{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.004369182939346164, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.004369182939346164, "step": 100, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 7.323769428184204, "val_fraction": 0.05995528919330289, "val_loss": 7.250895408258132, "within_bucket_gap": -0.07287401992607201}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.002496331974197335, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.002496331974197335, "step": 100, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 7.4335931019824155, "val_fraction": 0.019882039573820395, "val_loss": 7.308035964600777, "within_bucket_gap": -0.1255571373816382}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025602210363515965, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025602210363515965, "step": 100, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 7.329454152960526, "val_fraction": 0.011582001522070015, "val_loss": 7.307348983733316, "within_bucket_gap": -0.022105169227209842}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004555678959061397, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004555678959061397, "step": 100, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 7.290641918516996, "val_fraction": 0.007301179604261796, "val_loss": 7.2282454400575125, "within_bucket_gap": -0.06239647845948326}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019335047394851632, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019335047394851632, "step": 100, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 7.327886291823462, "val_fraction": 0.004827815829528158, "val_loss": 7.367935556496305, "within_bucket_gap": 0.04004926467284342}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0016022824700100008, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0016022824700100008, "step": 100, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 7.817550118087877, "val_fraction": 0.003305745814307458, "val_loss": 7.332853907303845, "within_bucket_gap": -0.48469621078403247}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001725653458591402, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001725653458591402, "step": 100, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 7.167820572164017, "val_fraction": 0.002877663622526636, "val_loss": 7.22778774293001, "within_bucket_gap": 0.05996717076599278}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018053819858968303, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018053819858968303, "step": 100, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 7.180953162057059, "val_fraction": 0.002449581430745814, "val_loss": 7.107251509879399, "within_bucket_gap": -0.07370165217766012}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006092558021267648, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006092558021267648, "step": 100, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 7.011685483090512, "val_fraction": 0.0025684931506849314, "val_loss": 7.2488890753851996, "within_bucket_gap": 0.23720359229468713}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009809034911153832, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009809034911153832, "step": 100, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 6.684971400669643, "val_fraction": 0.0014745053272450532, "val_loss": 7.350213819934476, "within_bucket_gap": 0.6652424192648327}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003310456063873125, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003310456063873125, "step": 100, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 7.102757284086045, "val_fraction": 0.0020214992389649923, "val_loss": 7.266519703584559, "within_bucket_gap": 0.16376241949851433}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002466359760239914, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002466359760239914, "step": 100, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 6.373685709635416, "val_fraction": 0.0016647640791476407, "val_loss": 6.521836417061942, "within_bucket_gap": 0.14815070742652559}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007764109033363786, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007764109033363786, "step": 100, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 6.974783504710478, "val_fraction": 0.0018074581430745813, "val_loss": 7.404343052914268, "within_bucket_gap": 0.42955954820379016}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007890172670282231, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007890172670282231, "step": 100, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 7.6418322394875915, "val_fraction": 0.0015220700152207, "val_loss": 7.123447895050049, "within_bucket_gap": -0.5183843444375427}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006350313855916633, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006350313855916633, "step": 100, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 7.745667770195799, "val_fraction": 0.0014745053272450532, "val_loss": 7.314993581464214, "within_bucket_gap": -0.4306741887315848}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006934880665992488, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006934880665992488, "step": 100, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 8.031367013620775, "val_fraction": 0.0010226407914764079, "val_loss": 7.3532324502634445, "within_bucket_gap": -0.6781345633573306}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.4061318144382444e-05, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.4061318144382444e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 6.9843612339185634, "val_fraction": 0.0004280821917808219, "val_loss": 7.110648473103841, "within_bucket_gap": 0.1262872391852774}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.889586496570776e-06, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.889586496570776e-06, "step": 100, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 6.615882873535156, "val_fraction": 0.0007372526636225266, "val_loss": 6.626584206857989, "within_bucket_gap": 0.010701333322832518}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.1403739860353955e-05, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.1403739860353955e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 7.169817788260324, "val_fraction": 0.000594558599695586, "val_loss": 7.23945556640625, "within_bucket_gap": 0.06963777814592653}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.999976892742612e-05, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.999976892742612e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 7.556638864370493, "val_fraction": 0.0007372526636225266, "val_loss": 7.638021776753087, "within_bucket_gap": 0.08138291238259399}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000645898974286548, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000645898974286548, "step": 100, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 6.88330078125, "val_fraction": 0.0006183409436834094, "val_loss": 7.927868476280799, "within_bucket_gap": 1.044567695030799}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.430875023387503e-05, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.430875023387503e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 6.912494795663016, "val_fraction": 0.0006659056316590563, "val_loss": 6.979033878871372, "within_bucket_gap": 0.06653908320835633}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019496172394382338, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019496172394382338, "step": 100, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 6.583657622337341, "val_fraction": 0.00047564687975646877, "val_loss": 6.993545150756836, "within_bucket_gap": 0.4098875284194943}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004358625484565259, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004358625484565259, "step": 100, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 6.26153564453125, "val_fraction": 0.0003091704718417047, "val_loss": 7.671316293569712, "within_bucket_gap": 1.4097806490384617}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019904763109141233, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019904763109141233, "step": 100, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.9505800519670755, "val_fraction": 0.000594558599695586, "val_loss": 6.285362243652344, "within_bucket_gap": 0.33478219168526824}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014867394300719234, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014867394300719234, "step": 100, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 8.08493741353353, "val_fraction": 0.0004280821917808219, "val_loss": 7.737635082668728, "within_bucket_gap": -0.34730233086480133}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007661421901591192, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007661421901591192, "step": 100, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 8.3373825970818, "val_fraction": 0.000594558599695586, "val_loss": 7.048792724609375, "within_bucket_gap": -1.2885898724724258}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003269165568749602, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003269165568749602, "step": 100, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 7.4398415305397725, "val_fraction": 0.00045186453576864533, "val_loss": 6.716357984040913, "within_bucket_gap": -0.7234835464988594}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008943413209627192, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008943413209627192, "step": 100, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.06657717304845, "val_fraction": 0.00040429984779299846, "val_loss": 7.278651517980239, "within_bucket_gap": 2.2120743449317892}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00026624163012002177, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00026624163012002177, "step": 100, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 6.947851392957899, "val_fraction": 0.00033295281582952814, "val_loss": 7.74748911176409, "within_bucket_gap": 0.7996377188061912}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013218938317745242, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013218938317745242, "step": 100, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 7.997731566429138, "val_fraction": 0.0002616057838660578, "val_loss": 7.492431640625, "within_bucket_gap": -0.5052999258041382}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022632179981194877, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022632179981194877, "step": 100, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 8.016413370768229, "val_fraction": 0.0005232115677321156, "val_loss": 7.583850687200373, "within_bucket_gap": -0.43256268356785554}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000129368494513796, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000129368494513796, "step": 100, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 8.348511377970377, "val_fraction": 0.0002616057838660578, "val_loss": 8.843028328635476, "within_bucket_gap": 0.4945169506650995}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020548876683038235, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020548876683038235, "step": 100, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 7.948301951090495, "val_fraction": 0.0002616057838660578, "val_loss": 7.162811799482866, "within_bucket_gap": -0.7854901516076289}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020161482024954878, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020161482024954878, "step": 100, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.927467346191406, "val_fraction": 0.00014269406392694063, "val_loss": 8.340384006500244, "within_bucket_gap": 1.412916660308838}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002976512202812538, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002976512202812538, "step": 100, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 8.570960839589437, "val_fraction": 0.0002616057838660578, "val_loss": 7.433175520463423, "within_bucket_gap": -1.1377853191260145}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001868901229298319, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001868901229298319, "step": 100, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 7.748486152062049, "val_fraction": 0.00028538812785388126, "val_loss": 7.093623161315918, "within_bucket_gap": -0.654862990746131}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.106626745772694e-05, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.106626745772694e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.625579833984375, "val_fraction": 0.0003567351598173516, "val_loss": 7.712664794921875, "within_bucket_gap": 0.08708496093750018}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.3051632027647854e-05, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.3051632027647854e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.995636940002441, "val_fraction": 0.0002616057838660578, "val_loss": 7.16020376032049, "within_bucket_gap": 0.1645668203180488}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.799822288907478e-05, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.799822288907478e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 6.570829564874822, "val_fraction": 9.512937595129375e-05, "val_loss": 5.645792245864868, "within_bucket_gap": -0.9250373190099541}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006268639052839584, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006268639052839584, "step": 100, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 5.512598266601563, "val_fraction": 0.00021404109589041095, "val_loss": 8.441306432088217, "within_bucket_gap": 2.928708165486654}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017901508571644968, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017901508571644968, "step": 100, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 7.6076942873207525, "val_fraction": 0.00014269406392694063, "val_loss": 6.353156566619873, "within_bucket_gap": -1.2545377207008794}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004817369023621871, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004817369023621871, "step": 100, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.107056209019253, "val_fraction": 0.00023782343987823439, "val_loss": 7.132663536071777, "within_bucket_gap": 2.0256073270525246}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.251073409796115e-05, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.251073409796115e-05, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.266345024108887, "val_fraction": 0.0002616057838660578, "val_loss": 6.505295146595348, "within_bucket_gap": 0.23895012248646097}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.90382783932672e-05, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.90382783932672e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 7.245488908555773, "val_fraction": 0.00016647640791476407, "val_loss": 7.840397698538644, "within_bucket_gap": 0.5949087899828713}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002912300180989676, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002912300180989676, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.292125701904297, "val_fraction": 0.000380517503805175, "val_loss": 8.057478189468384, "within_bucket_gap": 0.7653524875640869}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.6628419407867114e-05, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.6628419407867114e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.876225335257394, "val_fraction": 7.134703196347032e-05, "val_loss": 7.52976926167806, "within_bucket_gap": 0.6535439264206655}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.197568117574288e-06, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.197568117574288e-06, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.285853624343872, "val_fraction": 4.756468797564688e-05, "val_loss": 7.17657995223999, "within_bucket_gap": -0.10927367210388184}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.5290355452652833e-05, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.5290355452652833e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 8.137099988532789, "val_fraction": 4.756468797564688e-05, "val_loss": 7.605395555496216, "within_bucket_gap": -0.5317044330365732}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010249350218418571, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010249350218418571, "step": 100, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 6.878916306929155, "val_fraction": 0.00016647640791476407, "val_loss": 7.494580132620675, "within_bucket_gap": 0.6156638256915201}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002562519350008333, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002562519350008333, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.803544998168945, "val_fraction": 7.134703196347032e-05, "val_loss": 5.211917877197266, "within_bucket_gap": -3.5916271209716797}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002922057833301422, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002922057833301422, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.242481708526611, "val_fraction": 4.756468797564688e-05, "val_loss": 3.099147319793701, "within_bucket_gap": -6.14333438873291}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017705263262547783, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017705263262547783, "step": 100, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 8.541897024427142, "val_fraction": 0.00014269406392694063, "val_loss": 7.301112174987793, "within_bucket_gap": -1.2407848494393487}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.584300849172804e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.584300849172804e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 5.44206690788269, "val_fraction": 2.378234398782344e-05, "val_loss": 7.369673728942871, "within_bucket_gap": 1.9276068210601807}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001311960285657073, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001311960285657073, "step": 100, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.547484588623047, "val_fraction": 0.00016647640791476407, "val_loss": 7.335560389927456, "within_bucket_gap": 0.7880758013044087}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00026679426839906875, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00026679426839906875, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.838520050048828, "val_fraction": 0.00014269406392694063, "val_loss": 7.708214282989502, "within_bucket_gap": 1.8696942329406738}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001722522749022625, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001722522749022625, "step": 100, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.840224075317383, "val_fraction": 0.00011891171993911719, "val_loss": 7.28879680633545, "within_bucket_gap": 1.448572731018067}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002136126676605535, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002136126676605535, "step": 100, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 5.961740881204605, "val_fraction": 9.512937595129375e-05, "val_loss": 8.207237243652344, "within_bucket_gap": 2.2454963624477386}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005588972480813602, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005588972480813602, "step": 100, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.614945570627849, "val_fraction": 0.00023782343987823439, "val_loss": 7.964996719360352, "within_bucket_gap": 2.3500511487325033}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00023576263423379703, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00023576263423379703, "step": 100, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 8.742494583129883, "val_fraction": 0.00021404109589041095, "val_loss": 7.641011555989583, "within_bucket_gap": -1.1014830271402998}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001207255564142399, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001207255564142399, "step": 100, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 7.902168655395508, "val_fraction": 0.00011891171993911719, "val_loss": 6.886915016174316, "within_bucket_gap": -1.015253639221192}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.320858246659578e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.320858246659578e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.524552822113037, "val_fraction": 7.134703196347032e-05, "val_loss": 8.410484313964844, "within_bucket_gap": 0.8859314918518066}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00034818322577795846, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00034818322577795846, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.199923276901245, "val_fraction": 0.00016647640791476407, "val_loss": 9.291410173688616, "within_bucket_gap": 2.091486896787371}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.380095843552708e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.380095843552708e-05, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 8.12596861521403, "val_fraction": 0.0001902587519025875, "val_loss": 8.46130645275116, "within_bucket_gap": 0.3353378375371303}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00035430524991527545, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00035430524991527545, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.905963897705078, "val_fraction": 0.0002616057838660578, "val_loss": 7.551615975119851, "within_bucket_gap": -1.3543479225852275}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013115535996275948, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013115535996275948, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.98302412033081, "val_fraction": 4.756468797564688e-05, "val_loss": 7.225613832473755, "within_bucket_gap": -2.7574102878570557}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.498690141209006e-05, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.498690141209006e-05, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 9.80366063117981, "val_fraction": 7.134703196347032e-05, "val_loss": 8.752644220987955, "within_bucket_gap": -1.0510164101918544}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.8117725395539398e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.8117725395539398e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.564505577087402, "val_fraction": 4.756468797564688e-05, "val_loss": 8.183598518371582, "within_bucket_gap": -0.3809070587158203}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003944916630807351, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003944916630807351, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.726320266723633, "val_fraction": 9.512937595129375e-05, "val_loss": 8.87321662902832, "within_bucket_gap": 4.1468963623046875}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001604964381120702, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001604964381120702, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.6534223556518555, "val_fraction": 9.512937595129375e-05, "val_loss": 4.966283798217773, "within_bucket_gap": -1.687138557434082}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015055099034418253, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015055099034418253, "step": 100, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 8.711170196533203, "val_fraction": 7.134703196347032e-05, "val_loss": 6.601047515869141, "within_bucket_gap": -2.1101226806640625}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00026264337644185107, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00026264337644185107, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.9689273834228516, "val_fraction": 0.0001902587519025875, "val_loss": 5.349380970001221, "within_bucket_gap": 1.3804535865783691}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010633519092288372, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010633519092288372, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.551319122314453, "val_fraction": 2.378234398782344e-05, "val_loss": 3.080137014389038, "within_bucket_gap": -4.471182107925415}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00027478495371759023, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00027478495371759023, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.8204941749572754, "val_fraction": 7.134703196347032e-05, "val_loss": 6.6718800862630205, "within_bucket_gap": 3.851385911305745}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.66385658509474e-05, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.66385658509474e-05, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.532067775726318, "val_fraction": 0.00011891171993911719, "val_loss": 4.887568092346191, "within_bucket_gap": -0.6444996833801273}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020882580287561752, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020882580287561752, "step": 100, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.238186200459798, "val_fraction": 7.134703196347032e-05, "val_loss": 9.165088653564453, "within_bucket_gap": 2.9269024531046552}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.147175264140786e-05, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.147175264140786e-05, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.483477592468262, "val_fraction": 9.512937595129375e-05, "val_loss": 7.732166528701782, "within_bucket_gap": -0.7513110637664795}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.132213592529297, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.756037927290801e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.756037927290801e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.003090620040894, "val_fraction": 4.756468797564688e-05, "val_loss": 6.582701206207275, "within_bucket_gap": -1.4203894138336182}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001921389064832365, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001921389064832365, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.250332832336426, "val_fraction": 4.756468797564688e-05, "val_loss": 4.210804462432861, "within_bucket_gap": -4.0395283699035645}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.51022744178772, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00034860942280274366, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00034860942280274366, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.194926261901855, "val_fraction": 0.0001902587519025875, "val_loss": 6.362635135650635, "within_bucket_gap": -1.8322911262512207}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.950429318702384e-05, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.950429318702384e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.955963134765625, "val_fraction": 7.134703196347032e-05, "val_loss": 8.682590961456299, "within_bucket_gap": -0.27337217330932617}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.396805880458025e-05, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.396805880458025e-05, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.70816707611084, "val_fraction": 4.756468797564688e-05, "val_loss": 8.422311544418335, "within_bucket_gap": 0.7141444683074951}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020016679178331783, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020016679178331783, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.9049733479817705, "val_fraction": 9.512937595129375e-05, "val_loss": 8.009126663208008, "within_bucket_gap": 2.1041533152262373}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.174683917784436e-06, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.174683917784436e-06, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.46876335144043, "val_fraction": 9.512937595129375e-05, "val_loss": 8.51264762878418, "within_bucket_gap": 0.04388427734375}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.866762492391797e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.866762492391797e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.816666603088379, "val_fraction": 2.378234398782344e-05, "val_loss": 3.6678903102874756, "within_bucket_gap": -4.148776292800903}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.021097749335581e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.021097749335581e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.388405799865723, "val_fraction": 2.378234398782344e-05, "val_loss": 8.761116981506348, "within_bucket_gap": 3.372711181640625}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.945780733525603e-05, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.945780733525603e-05, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.828502019246419, "val_fraction": 4.756468797564688e-05, "val_loss": 9.499022960662842, "within_bucket_gap": 1.6705209414164228}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.609618172797982e-05, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.609618172797982e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.996699810028076, "val_fraction": 2.378234398782344e-05, "val_loss": 10.196392059326172, "within_bucket_gap": 3.1996922492980957}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.6995942393031475e-06, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.6995942393031475e-06, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.688807487487793, "val_fraction": 4.756468797564688e-05, "val_loss": 7.653075218200684, "within_bucket_gap": -0.035732269287109375}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.4707674131001506e-05, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.4707674131001506e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.215267181396484, "val_fraction": 2.378234398782344e-05, "val_loss": 8.515615463256836, "within_bucket_gap": 2.3003482818603516}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.31694499542724e-05, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.31694499542724e-05, "step": 100, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.633510971069336, "val_fraction": 2.378234398782344e-05, "val_loss": 10.130620002746582, "within_bucket_gap": 3.4971090316772457}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.4069498142332425e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.4069498142332425e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.464268088340759, "val_fraction": 4.756468797564688e-05, "val_loss": 6.390785217285156, "within_bucket_gap": 0.926517128944397}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025436658656034653, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025436658656034653, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.626093864440918, "val_fraction": 7.134703196347032e-05, "val_loss": 6.060891787211101, "within_bucket_gap": -3.5652020772298174}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.240874369939168, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.245244026184082, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.847294876262176e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.847294876262176e-05, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.61645491917928, "val_fraction": 2.378234398782344e-05, "val_loss": 9.49560546875, "within_bucket_gap": 2.8791505495707197}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.525188446044922, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.931909561157227, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.6340631509661854e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.6340631509661854e-05, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.40281867980957, "val_fraction": 2.378234398782344e-05, "val_loss": 9.089909553527832, "within_bucket_gap": 0.6870908737182617}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.1969170570373535, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.92463493347168, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.827924728393555, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003309837589253029, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003309837589253029, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.2032912969589233, "val_fraction": 4.756468797564688e-05, "val_loss": 10.161893844604492, "within_bucket_gap": 6.958602547645569}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.513677396730745e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.513677396730745e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.303939819335938, "val_fraction": 2.378234398782344e-05, "val_loss": 7.406028747558594, "within_bucket_gap": -1.8979110717773438}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003751343742716077, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003751343742716077, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.921950340270996, "val_fraction": 7.134703196347032e-05, "val_loss": 8.17983373006185, "within_bucket_gap": 5.2578833897908535}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.56120777130127, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003854029753799671, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003854029753799671, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.421292304992676, "val_fraction": 0.00014269406392694063, "val_loss": 6.720388253529866, "within_bucket_gap": -2.7009040514628095}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.653175212839785e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.653175212839785e-05, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.174209356307983, "val_fraction": 4.756468797564688e-05, "val_loss": 9.20369291305542, "within_bucket_gap": 2.0294835567474365}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010078091917154088, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010078091917154088, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.997379779815674, "val_fraction": 2.378234398782344e-05, "val_loss": 8.235015869140625, "within_bucket_gap": 4.237636089324951}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018983977865229276, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018983977865229276, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.050765037536621, "val_fraction": 2.378234398782344e-05, "val_loss": 3.0683820247650146, "within_bucket_gap": -7.9823830127716064}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.0078650994206491e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.0078650994206491e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.960386276245117, "val_fraction": 2.378234398782344e-05, "val_loss": 9.384173393249512, "within_bucket_gap": 0.42378711700439453}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.990965366363525, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.600379943847656, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.7076686593495546e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.7076686593495546e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.287749290466309, "val_fraction": 4.756468797564688e-05, "val_loss": 5.487729549407959, "within_bucket_gap": 1.1999802589416504}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.399764060974121, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.439523696899414, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.932448863983154, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.922815322875977, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.11894476939917e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.11894476939917e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.336816310882568, "val_fraction": 2.378234398782344e-05, "val_loss": 8.171150207519531, "within_bucket_gap": 3.834333896636963}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.464603424072266, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.6847755908966064, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6976561546325684, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.85552978515625, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010157252918458238, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010157252918458238, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.69686508178711, "val_fraction": 4.756468797564688e-05, "val_loss": 7.561404228210449, "within_bucket_gap": -2.13546085357666}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.727553844451904, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.82248830795288, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.458845138549805, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.181859649298216e-06, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.181859649298216e-06, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.863080024719238, "val_fraction": 2.378234398782344e-05, "val_loss": 8.477001190185547, "within_bucket_gap": -0.3860788345336914}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.213556752720198e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.213556752720198e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.207727432250977, "val_fraction": 2.378234398782344e-05, "val_loss": 9.558963775634766, "within_bucket_gap": 1.351236343383789}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.832998909362375e-06, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.832998909362375e-06, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.095848083496094, "val_fraction": 2.378234398782344e-05, "val_loss": 7.808534145355225, "within_bucket_gap": -0.28731393814086914}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.490460395812988, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.183038711547852, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.240102767944336, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00035407413449280156, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00035407413449280156, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.8050355911254883, "val_fraction": 4.756468797564688e-05, "val_loss": 11.249090194702148, "within_bucket_gap": 7.44405460357666}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.013128280639648, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.320490837097168, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.693978309631348, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.408249855041504, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.032221794128418, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.430161840839473e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.430161840839473e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.1329662799835205, "val_fraction": 2.378234398782344e-05, "val_loss": 3.588251829147339, "within_bucket_gap": -3.5447144508361816}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.843982696533203, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.7703574895858765, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.667648093522593e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.667648093522593e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.782924652099609, "val_fraction": 2.378234398782344e-05, "val_loss": 3.2407519817352295, "within_bucket_gap": -1.5421726703643799}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001164300245327855, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001164300245327855, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.5933010578155518, "val_fraction": 2.378234398782344e-05, "val_loss": 8.488950729370117, "within_bucket_gap": 4.895649671554565}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.076906442642212, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.224729537963867, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.296095848083496, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.489548683166504, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.453949928283691, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.512311935424805, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.932850360870361, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.990124034736436e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.990124034736436e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.780751705169678, "val_fraction": 2.378234398782344e-05, "val_loss": 3.580104351043701, "within_bucket_gap": -4.200647354125977}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.04991340637207, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013504092846649784, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013504092846649784, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.644710540771484, "val_fraction": 2.378234398782344e-05, "val_loss": 3.9665095806121826, "within_bucket_gap": -5.678200960159302}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001415704948263444, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001415704948263444, "step": 100, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 8.728963851928711, "val_fraction": 2.378234398782344e-05, "val_loss": 2.776207685470581, "within_bucket_gap": -5.95275616645813}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.262628705897105e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.262628705897105e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.240929126739502, "val_fraction": 2.378234398782344e-05, "val_loss": 9.135679244995117, "within_bucket_gap": 3.8947501182556152}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.768194198608398, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.0700483687754995e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.0700483687754995e-05, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.998570561408997, "val_fraction": 2.378234398782344e-05, "val_loss": 9.130424499511719, "within_bucket_gap": 2.131853938102722}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.951251983642578, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.725942850112915, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 10.303474426269531, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.25459098815918, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.357381343841553, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.034565448760986, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.268410682678223, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.653826713562012, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.9936041831970215, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.900764465332031, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.052916378009627e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.052916378009627e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.761232376098633, "val_fraction": 2.378234398782344e-05, "val_loss": 4.3751420974731445, "within_bucket_gap": -3.3860902786254883}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.118714809417725, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.60752010345459, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.0525335200482486e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.0525335200482486e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.6505630016326904, "val_fraction": 2.378234398782344e-05, "val_loss": 5.354572296142578, "within_bucket_gap": 1.7040092945098877}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014816179213821432, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014816179213821432, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.894991874694824, "val_fraction": 2.378234398782344e-05, "val_loss": 9.124898910522461, "within_bucket_gap": 6.229907035827637}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.904913425445557, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.576090812683105, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1730008125305176, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.20500512046901e-05, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.20500512046901e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.697754621505737, "val_fraction": 2.378234398782344e-05, "val_loss": 10.306835174560547, "within_bucket_gap": 2.6090805530548096}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.262457942309444e-05, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.262457942309444e-05, "step": 100, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.002891778945923, "val_fraction": 4.756468797564688e-05, "val_loss": 4.686272621154785, "within_bucket_gap": -1.3166191577911377}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.094945907592773, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.170877456665039, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.412536144256592, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.999727725982666, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.676854133605957, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.13369083404541, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.502588272094727, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.1804375648498535, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.756932018372929e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.756932018372929e-05, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.475661277770996, "val_fraction": 2.378234398782344e-05, "val_loss": 4.895946502685547, "within_bucket_gap": -1.5797147750854492}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.093755722045898, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.455105781555176, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.46603636037632e-07, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.46603636037632e-07, "step": 100, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.097510814666748, "val_fraction": 2.378234398782344e-05, "val_loss": 3.1373136043548584, "within_bucket_gap": 0.03980278968811035}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.1317012948714073e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.1317012948714073e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.722275733947754, "val_fraction": 4.756468797564688e-05, "val_loss": 3.0638668537139893, "within_bucket_gap": -0.6584088802337646}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.481013853981796e-05, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.481013853981796e-05, "step": 100, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.658403396606445, "val_fraction": 9.512937595129375e-05, "val_loss": 8.397599220275879, "within_bucket_gap": -0.2608041763305664}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.008175707794489345, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.008175707794489345, "step": 200, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 6.716986955362034, "val_fraction": 0.05995528919330289, "val_loss": 6.580623543484728, "within_bucket_gap": -0.13636341187730583}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.002417983511232747, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.002417983511232747, "step": 200, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 6.797132262981088, "val_fraction": 0.019882039573820395, "val_loss": 6.6755157908754486, "within_bucket_gap": -0.12161647210563942}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0013725575351814475, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0013725575351814475, "step": 200, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 6.707370476973685, "val_fraction": 0.011582001522070015, "val_loss": 6.588862675660934, "within_bucket_gap": -0.11850780131275052}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0012766536022111798, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0012766536022111798, "step": 200, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 6.679942991022478, "val_fraction": 0.007301179604261796, "val_loss": 6.505087190808225, "within_bucket_gap": -0.17485580021425307}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002032250617889105, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002032250617889105, "step": 200, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 6.64211109421016, "val_fraction": 0.004827815829528158, "val_loss": 6.600016476485529, "within_bucket_gap": -0.04209461772463108}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0018571479598715873, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0018571479598715873, "step": 200, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 7.017836550448803, "val_fraction": 0.003305745814307458, "val_loss": 6.456042612199303, "within_bucket_gap": -0.5617939382495001}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020059389451393267, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020059389451393267, "step": 200, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 6.538068297281431, "val_fraction": 0.002877663622526636, "val_loss": 6.468361090037448, "within_bucket_gap": -0.06970720724398216}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00026347389504245456, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00026347389504245456, "step": 200, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 6.48924446105957, "val_fraction": 0.002449581430745814, "val_loss": 6.381685719906705, "within_bucket_gap": -0.10755874115286534}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.221696073014614e-06, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.221696073014614e-06, "step": 200, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 6.622198156408362, "val_fraction": 0.0025684931506849314, "val_loss": 6.619775842737268, "within_bucket_gap": -0.00242231367109369}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007626569915772439, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007626569915772439, "step": 200, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 6.151054226622289, "val_fraction": 0.0014745053272450532, "val_loss": 6.6682832779422885, "within_bucket_gap": 0.5172290513199993}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013932499529927038, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013932499529927038, "step": 200, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 6.489105224609375, "val_fraction": 0.0020214992389649923, "val_loss": 6.558026841107536, "within_bucket_gap": 0.06892161649816142}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016458317166190777, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016458317166190777, "step": 200, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 5.959116448296441, "val_fraction": 0.0016647640791476407, "val_loss": 5.860253688267299, "within_bucket_gap": -0.0988627600291414}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008447220528474781, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008447220528474781, "step": 200, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 6.078191084020278, "val_fraction": 0.0018074581430745813, "val_loss": 6.545544674521999, "within_bucket_gap": 0.46735359050172054}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005180700173874505, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005180700173874505, "step": 200, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 6.839982425465303, "val_fraction": 0.0015220700152207, "val_loss": 6.499610424041748, "within_bucket_gap": -0.34037200142355495}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0011438060463488584, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0011438060463488584, "step": 200, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 7.55889867252334, "val_fraction": 0.0014745053272450532, "val_loss": 6.783176791283392, "within_bucket_gap": -0.7757218812399485}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007413055980223676, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007413055980223676, "step": 200, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 7.39324046290198, "val_fraction": 0.0010226407914764079, "val_loss": 6.6683470260265265, "within_bucket_gap": -0.7248934368754538}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.370604030444305e-05, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.370604030444305e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 6.531720368758492, "val_fraction": 0.0004280821917808219, "val_loss": 6.359543058607313, "within_bucket_gap": -0.17217731015117899}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.28436277144319e-05, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.28436277144319e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 6.1186635087176064, "val_fraction": 0.0007372526636225266, "val_loss": 6.006295481035786, "within_bucket_gap": -0.11236802768182041}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021918961718746537, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021918961718746537, "step": 200, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 6.216986656188965, "val_fraction": 0.000594558599695586, "val_loss": 6.585646057128907, "within_bucket_gap": 0.36865940093994176}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010421460747593196, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010421460747593196, "step": 200, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 7.363976405217097, "val_fraction": 0.0007372526636225266, "val_loss": 7.222621056341356, "within_bucket_gap": -0.14135534887574153}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005722100273372439, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005722100273372439, "step": 200, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 6.4491286739226314, "val_fraction": 0.0006183409436834094, "val_loss": 7.374524336594802, "within_bucket_gap": 0.9253956626721704}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003480502096666778, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003480502096666778, "step": 200, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 6.245575223650251, "val_fraction": 0.0006659056316590563, "val_loss": 6.768247195652553, "within_bucket_gap": 0.5226719720023025}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00027630068758669856, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00027630068758669856, "step": 200, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 5.98003625869751, "val_fraction": 0.00047564687975646877, "val_loss": 6.560930824279785, "within_bucket_gap": 0.580894565582275}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003599710805593924, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003599710805593924, "step": 200, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 5.498974914550781, "val_fraction": 0.0003091704718417047, "val_loss": 6.663287529578576, "within_bucket_gap": 1.1643126150277947}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003492712559817173, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003492712559817173, "step": 200, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.222113473074777, "val_fraction": 0.000594558599695586, "val_loss": 5.809559783935547, "within_bucket_gap": 0.5874463108607699}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013033848374945943, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013033848374945943, "step": 200, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 7.3235321044921875, "val_fraction": 0.0004280821917808219, "val_loss": 7.01906140645345, "within_bucket_gap": -0.3044706980387373}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007276318865724347, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007276318865724347, "step": 200, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 7.246576496199066, "val_fraction": 0.000594558599695586, "val_loss": 6.022757873535157, "within_bucket_gap": -1.2238186226639094}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005860434613386859, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005860434613386859, "step": 200, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 7.219030033458363, "val_fraction": 0.00045186453576864533, "val_loss": 5.922085009123149, "within_bucket_gap": -1.2969450243352139}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008234207634956979, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008234207634956979, "step": 200, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 4.340355780816847, "val_fraction": 0.00040429984779299846, "val_loss": 6.377014384550207, "within_bucket_gap": 2.0366586037333594}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003442809744250566, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003442809744250566, "step": 200, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 6.06575690375434, "val_fraction": 0.00033295281582952814, "val_loss": 7.099780218941825, "within_bucket_gap": 1.0340233151874845}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003407896018645172, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003407896018645172, "step": 200, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 7.612468719482422, "val_fraction": 0.0002616057838660578, "val_loss": 6.3097849759188565, "within_bucket_gap": -1.3026837435635654}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002456895595029911, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002456895595029911, "step": 200, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 7.524566014607747, "val_fraction": 0.0005232115677321156, "val_loss": 7.05498626015403, "within_bucket_gap": -0.4695797544537168}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016424837812804009, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016424837812804009, "step": 200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.335747718811035, "val_fraction": 0.0002616057838660578, "val_loss": 7.963594610040838, "within_bucket_gap": 0.6278468912298028}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013831934710353743, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013831934710353743, "step": 200, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 7.453119489881727, "val_fraction": 0.0002616057838660578, "val_loss": 6.924387498335405, "within_bucket_gap": -0.528731991546322}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002416070873878862, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002416070873878862, "step": 200, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.215144793192546, "val_fraction": 0.00014269406392694063, "val_loss": 7.9083272616068525, "within_bucket_gap": 1.6931824684143066}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00019904311494982102, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00019904311494982102, "step": 200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.747122605641683, "val_fraction": 0.0002616057838660578, "val_loss": 6.986271251331676, "within_bucket_gap": -0.7608513543100068}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.428192469404032e-05, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.428192469404032e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.720238172090971, "val_fraction": 0.00028538812785388126, "val_loss": 6.670194307963054, "within_bucket_gap": -0.05004386412791728}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016404071445697327, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016404071445697327, "step": 200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.2711489995320635, "val_fraction": 0.0003567351598173516, "val_loss": 6.811310068766276, "within_bucket_gap": -0.4598389307657875}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.16634205051754e-05, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.16634205051754e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.546690464019775, "val_fraction": 0.0002616057838660578, "val_loss": 6.782401691783559, "within_bucket_gap": 0.23571122776378317}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.86334317436815e-05, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.86334317436815e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 6.008442445234819, "val_fraction": 9.512937595129375e-05, "val_loss": 5.181847810745239, "within_bucket_gap": -0.82659463448958}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006426692027116654, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006426692027116654, "step": 200, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 5.1910786437988286, "val_fraction": 0.00021404109589041095, "val_loss": 8.19362915886773, "within_bucket_gap": 3.0025505150689007}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018207222260395583, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018207222260395583, "step": 200, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 6.8724892356178975, "val_fraction": 0.00014269406392694063, "val_loss": 5.596527099609375, "within_bucket_gap": -1.2759621360085225}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005341720635488152, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005341720635488152, "step": 200, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.9912776947021484, "val_fraction": 0.00023782343987823439, "val_loss": 6.237364387512207, "within_bucket_gap": 2.2460866928100582}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018556543046118103, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018556543046118103, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.008313179016113, "val_fraction": 0.0002616057838660578, "val_loss": 5.7176454717462715, "within_bucket_gap": 0.7093322927301582}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.83579133187135e-05, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.83579133187135e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 6.546552022298177, "val_fraction": 0.00016647640791476407, "val_loss": 7.077305385044643, "within_bucket_gap": 0.5307533627464665}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021033853156381544, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021033853156381544, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.039307594299316, "val_fraction": 0.000380517503805175, "val_loss": 7.592077255249023, "within_bucket_gap": 0.552769660949707}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.517024630177246e-05, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.517024630177246e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.39050156729562, "val_fraction": 7.134703196347032e-05, "val_loss": 7.303927739461263, "within_bucket_gap": 0.9134261721656429}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.8979299422268452e-05, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.8979299422268452e-05, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.250362038612366, "val_fraction": 4.756468797564688e-05, "val_loss": 5.851341247558594, "within_bucket_gap": -0.399020791053772}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.4639923171272173e-05, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.4639923171272173e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 6.83072382031065, "val_fraction": 4.756468797564688e-05, "val_loss": 7.138513565063477, "within_bucket_gap": 0.30778974475282617}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.2338087241052803e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.2338087241052803e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.987665349786932, "val_fraction": 0.00016647640791476407, "val_loss": 6.061778477260044, "within_bucket_gap": 0.0741131274731126}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003339015004115925, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003339015004115925, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.182243347167969, "val_fraction": 7.134703196347032e-05, "val_loss": 3.5022799173990884, "within_bucket_gap": -4.679963429768881}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003286959146981551, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003286959146981551, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.092840671539307, "val_fraction": 4.756468797564688e-05, "val_loss": 2.182337760925293, "within_bucket_gap": -6.910502910614014}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020830557284911942, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020830557284911942, "step": 200, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 7.907782520566668, "val_fraction": 0.00014269406392694063, "val_loss": 6.447977066040039, "within_bucket_gap": -1.459805454526629}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.626250814629472e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.626250814629472e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 4.366697311401367, "val_fraction": 2.378234398782344e-05, "val_loss": 5.050503253936768, "within_bucket_gap": 0.6838059425354004}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.408463898314735e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.408463898314735e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.703778266906738, "val_fraction": 0.00016647640791476407, "val_loss": 6.138625281197684, "within_bucket_gap": -0.5651529857090543}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001792243513161008, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001792243513161008, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.324026107788086, "val_fraction": 0.00014269406392694063, "val_loss": 6.58003036181132, "within_bucket_gap": 1.2560042540232343}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.829643251689036e-06, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.829643251689036e-06, "step": 200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.618420886993408, "val_fraction": 0.00011891171993911719, "val_loss": 5.577805519104004, "within_bucket_gap": -0.04061536788940412}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00024777365057435753, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00024777365057435753, "step": 200, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.9295055866241455, "val_fraction": 9.512937595129375e-05, "val_loss": 7.534102201461792, "within_bucket_gap": 2.6045966148376465}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005504840532273341, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005504840532273341, "step": 200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.2142148812611895, "val_fraction": 0.00023782343987823439, "val_loss": 7.528890228271484, "within_bucket_gap": 2.3146753470102945}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.914094998956269e-05, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.914094998956269e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.5487712860107425, "val_fraction": 0.00021404109589041095, "val_loss": 7.0119578043619795, "within_bucket_gap": 0.4631865183512369}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.274746772179679e-05, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.274746772179679e-05, "step": 200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.047442626953125, "val_fraction": 0.00011891171993911719, "val_loss": 6.3228357315063475, "within_bucket_gap": 0.2753931045532223}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.312478358705654e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.312478358705654e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.474094390869141, "val_fraction": 7.134703196347032e-05, "val_loss": 7.499011357625325, "within_bucket_gap": 1.0249169667561846}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00024185748699956106, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00024185748699956106, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.9116737842559814, "val_fraction": 0.00016647640791476407, "val_loss": 8.364477157592773, "within_bucket_gap": 1.452803373336792}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.748365015381184e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.748365015381184e-05, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.28579568862915, "val_fraction": 0.0001902587519025875, "val_loss": 7.5353697538375854, "within_bucket_gap": 0.24957406520843506}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025989750862303023, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025989750862303023, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.900874853134155, "val_fraction": 0.0002616057838660578, "val_loss": 6.907404812899503, "within_bucket_gap": -0.9934700402346524}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.227893289365725e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.227893289365725e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.94545602798462, "val_fraction": 4.756468797564688e-05, "val_loss": 7.425863742828369, "within_bucket_gap": -1.51959228515625}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014096490551166154, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014096490551166154, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 9.500733852386475, "val_fraction": 7.134703196347032e-05, "val_loss": 7.524969736735026, "within_bucket_gap": -1.9757641156514483}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.3707543333915815e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.3707543333915815e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.839860916137695, "val_fraction": 4.756468797564688e-05, "val_loss": 7.341433525085449, "within_bucket_gap": -0.4984273910522461}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00038224939523403683, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00038224939523403683, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.316777229309082, "val_fraction": 9.512937595129375e-05, "val_loss": 8.334982872009277, "within_bucket_gap": 4.018205642700195}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.372413481388644e-05, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.372413481388644e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.306774139404297, "val_fraction": 9.512937595129375e-05, "val_loss": 4.847146034240723, "within_bucket_gap": -0.4596281051635742}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011442697392932542, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011442697392932542, "step": 200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.695261383056641, "val_fraction": 7.134703196347032e-05, "val_loss": 5.091452916463216, "within_bucket_gap": -1.6038084665934251}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002615653670178882, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002615653670178882, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.5781333446502686, "val_fraction": 0.0001902587519025875, "val_loss": 4.952920913696289, "within_bucket_gap": 1.3747875690460205}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.442083903825083e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.442083903825083e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.0462236404418945, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4964962005615234, "within_bucket_gap": -3.549727439880371}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002649672885100773, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002649672885100773, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.490659236907959, "val_fraction": 7.134703196347032e-05, "val_loss": 6.204440752665202, "within_bucket_gap": 3.713781515757243}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.036222850895364e-05, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.036222850895364e-05, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.289236545562744, "val_fraction": 0.00011891171993911719, "val_loss": 4.697518348693848, "within_bucket_gap": -0.5917181968688965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016383150850438275, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016383150850438275, "step": 200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.782779375712077, "val_fraction": 7.134703196347032e-05, "val_loss": 8.079041798909506, "within_bucket_gap": 2.2962624231974287}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.8805158613660744e-05, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.8805158613660744e-05, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.599315404891968, "val_fraction": 9.512937595129375e-05, "val_loss": 7.086275577545166, "within_bucket_gap": -0.5130398273468018}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.788692951202393, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.266116922485955e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.266116922485955e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.485126972198486, "val_fraction": 4.756468797564688e-05, "val_loss": 5.957498550415039, "within_bucket_gap": -1.5276284217834473}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018732646647472178, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018732646647472178, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.701672077178955, "val_fraction": 4.756468797564688e-05, "val_loss": 2.7633204460144043, "within_bucket_gap": -3.938351631164551}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.321829319000244, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00028606741214270276, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00028606741214270276, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.443150997161865, "val_fraction": 0.0001902587519025875, "val_loss": 5.939580678939819, "within_bucket_gap": -1.503570318222046}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.469535978598865e-05, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.469535978598865e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.232540130615234, "val_fraction": 7.134703196347032e-05, "val_loss": 7.905289967854817, "within_bucket_gap": -1.327250162760417}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.2889215254892496e-05, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.2889215254892496e-05, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.670939087867737, "val_fraction": 4.756468797564688e-05, "val_loss": 7.941921949386597, "within_bucket_gap": 0.27098286151885986}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.87478815414529e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.87478815414529e-05, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.204541524251302, "val_fraction": 9.512937595129375e-05, "val_loss": 5.376743793487549, "within_bucket_gap": -0.827797730763753}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.859541368992543e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.859541368992543e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.337347030639648, "val_fraction": 9.512937595129375e-05, "val_loss": 7.7430620193481445, "within_bucket_gap": 0.4057149887084961}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.444655854408055e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.444655854408055e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.487730979919434, "val_fraction": 2.378234398782344e-05, "val_loss": 3.3574020862579346, "within_bucket_gap": -3.130328893661499}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.3466164628483207e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.3466164628483207e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.4155471324920654, "val_fraction": 2.378234398782344e-05, "val_loss": 3.9817724227905273, "within_bucket_gap": 0.5662252902984619}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014653072358387367, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014653072358387367, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.46157964070638, "val_fraction": 4.756468797564688e-05, "val_loss": 8.54224157333374, "within_bucket_gap": 3.0806619326273603}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013054245048099093, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013054245048099093, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.377738952636719, "val_fraction": 2.378234398782344e-05, "val_loss": 9.866787910461426, "within_bucket_gap": 5.489048957824707}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.945903648161634e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.945903648161634e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.94595718383789, "val_fraction": 4.756468797564688e-05, "val_loss": 6.854930400848389, "within_bucket_gap": -2.091026782989502}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.587761465998726e-05, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.587761465998726e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.060422897338867, "val_fraction": 2.378234398782344e-05, "val_loss": 8.671404838562012, "within_bucket_gap": 3.6109819412231445}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011654556976968658, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011654556976968658, "step": 200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.534530448913574, "val_fraction": 2.378234398782344e-05, "val_loss": 10.435038566589355, "within_bucket_gap": 4.900508117675781}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.0123757058626984e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.0123757058626984e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.944725155830383, "val_fraction": 4.756468797564688e-05, "val_loss": 5.157567024230957, "within_bucket_gap": 0.21284186840057373}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00032690183303490434, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00032690183303490434, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.096905708312988, "val_fraction": 7.134703196347032e-05, "val_loss": 5.5150496164957685, "within_bucket_gap": -4.58185609181722}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.6490322748819985, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.162979125976562, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.712742628632676e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.712742628632676e-05, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.022909482320149, "val_fraction": 2.378234398782344e-05, "val_loss": 7.584043502807617, "within_bucket_gap": 1.5611340204874677}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.534084320068359, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.683520317077637, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.1484656069013807e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.1484656069013807e-05, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.1453447341918945, "val_fraction": 2.378234398782344e-05, "val_loss": 6.662437915802002, "within_bucket_gap": -0.4829068183898926}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.875741958618164, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.339428901672363, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.819671630859375, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003074315154026269, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003074315154026269, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.5877625942230225, "val_fraction": 4.756468797564688e-05, "val_loss": 9.051202774047852, "within_bucket_gap": 6.463440179824829}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.12025737907607e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.12025737907607e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.895852088928223, "val_fraction": 2.378234398782344e-05, "val_loss": 5.481446266174316, "within_bucket_gap": -3.4144058227539062}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00043865020530046024, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00043865020530046024, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.9671671390533447, "val_fraction": 7.134703196347032e-05, "val_loss": 8.115288416544596, "within_bucket_gap": 6.148121277491251}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.787341117858887, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00037408210734072705, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00037408210734072705, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.635202884674072, "val_fraction": 0.00014269406392694063, "val_loss": 6.013635476430257, "within_bucket_gap": -2.6215674082438154}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011271816304587642, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011271816304587642, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.599968989690145, "val_fraction": 4.756468797564688e-05, "val_loss": 8.96975564956665, "within_bucket_gap": 2.369786659876506}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.899469633987867e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.899469633987867e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.13751220703125, "val_fraction": 2.378234398782344e-05, "val_loss": 7.300041198730469, "within_bucket_gap": 4.162528991699219}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014969941279659532, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014969941279659532, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.948878288269043, "val_fraction": 2.378234398782344e-05, "val_loss": 2.6543173789978027, "within_bucket_gap": -6.29456090927124}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.032706160886465e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.032706160886465e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.804885864257812, "val_fraction": 2.378234398782344e-05, "val_loss": 7.6887335777282715, "within_bucket_gap": -2.116152286529541}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.011641025543213, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.012570381164551, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.9561663791650686e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.9561663791650686e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.0272064208984375, "val_fraction": 4.756468797564688e-05, "val_loss": 4.438470840454102, "within_bucket_gap": 0.41126441955566406}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.5507917404174805, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.306380271911621, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.853536605834961, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.483293851216634, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.415743807316551e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.415743807316551e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.4633572101593018, "val_fraction": 2.378234398782344e-05, "val_loss": 6.422489166259766, "within_bucket_gap": 3.959131956100464}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.545156478881836, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.5112462043762207, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.4177744388580322, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.474492073059082, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.309914254343854e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.309914254343854e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.670716285705566, "val_fraction": 4.756468797564688e-05, "val_loss": 6.5543599128723145, "within_bucket_gap": -1.116356372833252}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.140349388122559, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.880685806274414, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.248991966247559, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.0246312044889777e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.0246312044889777e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.577939510345459, "val_fraction": 2.378234398782344e-05, "val_loss": 6.726622581481934, "within_bucket_gap": -0.8513169288635254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.7561619503131557e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.7561619503131557e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.463963508605957, "val_fraction": 2.378234398782344e-05, "val_loss": 8.202394485473633, "within_bucket_gap": 0.7384309768676758}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.4355154149971413e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.4355154149971413e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.420580863952637, "val_fraction": 2.378234398782344e-05, "val_loss": 5.396495342254639, "within_bucket_gap": -1.024085521697998}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.060998439788818, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.462810754776001, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.104780197143555, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003001890747695934, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003001890747695934, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.742358922958374, "val_fraction": 4.756468797564688e-05, "val_loss": 10.053534030914307, "within_bucket_gap": 6.311175107955933}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.716684341430664, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.266064643859863, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.660707473754883, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.139440536499023, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.127243995666504, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.393797956463772e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.393797956463772e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.854060649871826, "val_fraction": 2.378234398782344e-05, "val_loss": 3.1655964851379395, "within_bucket_gap": -2.6884641647338867}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.150468826293945, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.132315516471863, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.0392482910344953e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.0392482910344953e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.1890869140625, "val_fraction": 2.378234398782344e-05, "val_loss": 3.3316237926483154, "within_bucket_gap": -0.8574631214141846}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.736487179586332e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.736487179586332e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.4278812408447266, "val_fraction": 2.378234398782344e-05, "val_loss": 5.4194793701171875, "within_bucket_gap": 1.991598129272461}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.307706832885742, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.864500999450684, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.341307997703552, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.25223970413208, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.759805679321289, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.201567649841309, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.9205868244171143, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.5456831711068e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.5456831711068e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.6104736328125, "val_fraction": 2.378234398782344e-05, "val_loss": 3.2786247730255127, "within_bucket_gap": -2.3318488597869873}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.729511260986328, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017784059025018363, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017784059025018363, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.230121612548828, "val_fraction": 2.378234398782344e-05, "val_loss": 2.7522804737091064, "within_bucket_gap": -7.477841138839722}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012946828353048155, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012946828353048155, "step": 200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 8.309237162272135, "val_fraction": 2.378234398782344e-05, "val_loss": 2.8653547763824463, "within_bucket_gap": -5.4438823858896885}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.554359194350569e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.554359194350569e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.558205604553223, "val_fraction": 2.378234398782344e-05, "val_loss": 8.57562255859375, "within_bucket_gap": 4.017416954040527}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.776626110076904, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.70866836859211e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.70866836859211e-05, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.588633298873901, "val_fraction": 2.378234398782344e-05, "val_loss": 9.829974174499512, "within_bucket_gap": 3.2413408756256104}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.331708908081055, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.3448925018310547, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.77652359008789, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.445995807647705, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.844644546508789, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.643924236297607, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.008567810058594, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.406895399093628, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.308266639709473, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.031325101852417, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.4997054815655247e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.4997054815655247e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.777157783508301, "val_fraction": 2.378234398782344e-05, "val_loss": 5.146561622619629, "within_bucket_gap": -0.6305961608886719}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.0797648429870605, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.36139440536499, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.469207473677587e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.469207473677587e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.324951946735382, "val_fraction": 2.378234398782344e-05, "val_loss": 4.465604305267334, "within_bucket_gap": 3.140652358531952}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012948119635814218, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012948119635814218, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7806904315948486, "val_fraction": 2.378234398782344e-05, "val_loss": 8.225115776062012, "within_bucket_gap": 5.444425344467163}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.134915351867676, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.80126953125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.6463125944137573, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010944354593118757, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010944354593118757, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.88834547996521, "val_fraction": 2.378234398782344e-05, "val_loss": 10.490227699279785, "within_bucket_gap": 4.601882219314575}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010686464709778355, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010686464709778355, "step": 200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.558853030204773, "val_fraction": 4.756468797564688e-05, "val_loss": 3.3121306896209717, "within_bucket_gap": -2.2467223405838013}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.557822227478027, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.472695827484131, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.357528686523438, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.0991952419281006, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.0867193937301636, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.5236058235168457, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.433584213256836, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.5895520448684692, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.5237548670630839e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.5237548670630839e-05, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.829528331756592, "val_fraction": 2.378234398782344e-05, "val_loss": 5.188819885253906, "within_bucket_gap": -0.6407084465026855}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.190791130065918, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.10923957824707, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.034140832166512e-07, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.034140832166512e-07, "step": 200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.0993118286132812, "val_fraction": 2.378234398782344e-05, "val_loss": 1.137298583984375, "within_bucket_gap": 0.03798675537109375}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.6222046550187526e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.6222046550187526e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.874887466430664, "val_fraction": 4.756468797564688e-05, "val_loss": 1.1133551597595215, "within_bucket_gap": -0.7615323066711426}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.832367668050246e-05, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.832367668050246e-05, "step": 200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.46369743347168, "val_fraction": 9.512937595129375e-05, "val_loss": 7.271078944206238, "within_bucket_gap": -0.1926184892654419}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.011669820441479533, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.011669820441479533, "step": 300, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 6.3576493519132615, "val_fraction": 0.05995528919330289, "val_loss": 6.163007301170171, "within_bucket_gap": -0.1946420507430906}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.003438376102575289, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.003438376102575289, "step": 300, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 6.420486298316191, "val_fraction": 0.019882039573820395, "val_loss": 6.247547496448863, "within_bucket_gap": -0.17293880186732746}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0025741456745431626, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0025741456745431626, "step": 300, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 6.317067999588816, "val_fraction": 0.011582001522070015, "val_loss": 6.0948140420463295, "within_bucket_gap": -0.22225395754248645}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004100845501118478, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004100845501118478, "step": 300, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 6.266915518777412, "val_fraction": 0.007301179604261796, "val_loss": 6.210748628995318, "within_bucket_gap": -0.05616688978209439}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0009930047011027167, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0009930047011027167, "step": 300, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 6.344898323738138, "val_fraction": 0.004827815829528158, "val_loss": 6.13921427609298, "within_bucket_gap": -0.2056840476451578}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001770485394209759, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001770485394209759, "step": 300, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 6.665208748891844, "val_fraction": 0.003305745814307458, "val_loss": 6.129630548490895, "within_bucket_gap": -0.5355782004009493}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005805171191235591, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005805171191235591, "step": 300, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 6.170946418894508, "val_fraction": 0.002877663622526636, "val_loss": 5.969214321168001, "within_bucket_gap": -0.20173209772650758}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025446483248548694, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025446483248548694, "step": 300, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 6.030280249459403, "val_fraction": 0.002449581430745814, "val_loss": 5.926399305028823, "within_bucket_gap": -0.10388094443058016}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016789938697468163, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016789938697468163, "step": 300, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 6.015781436954533, "val_fraction": 0.0025684931506849314, "val_loss": 5.9504126089590565, "within_bucket_gap": -0.06536882799547605}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006091361409764006, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006091361409764006, "step": 300, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 5.667613438197544, "val_fraction": 0.0014745053272450532, "val_loss": 6.080725639097152, "within_bucket_gap": 0.413112200899608}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004514549177265275, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004514549177265275, "step": 300, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 6.189736980281464, "val_fraction": 0.0020214992389649923, "val_loss": 5.96641019933364, "within_bucket_gap": -0.22332678094782388}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001616859403778284, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001616859403778284, "step": 300, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 5.504665289984809, "val_fraction": 0.0016647640791476407, "val_loss": 5.4075428553989955, "within_bucket_gap": -0.09712243458581327}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000832818371829718, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000832818371829718, "step": 300, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.668329463285558, "val_fraction": 0.0018074581430745813, "val_loss": 6.129097185636821, "within_bucket_gap": 0.46076772235126295}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000652653334849266, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000652653334849266, "step": 300, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 6.305024315329159, "val_fraction": 0.0015220700152207, "val_loss": 5.876231074333191, "within_bucket_gap": -0.42879324099596783}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001990635067982467, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001990635067982467, "step": 300, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 7.468304519336748, "val_fraction": 0.0014745053272450532, "val_loss": 6.118268659037929, "within_bucket_gap": -1.350035860298819}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0013800292436027819, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0013800292436027819, "step": 300, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 7.492794569148574, "val_fraction": 0.0010226407914764079, "val_loss": 6.14331853112509, "within_bucket_gap": -1.3494760380234831}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.600009002973332e-05, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.600009002973332e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 6.153147821841032, "val_fraction": 0.0004280821917808219, "val_loss": 6.115771611531575, "within_bucket_gap": -0.037376210309457036}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004773101717422847, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004773101717422847, "step": 300, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.819738062416635, "val_fraction": 0.0007372526636225266, "val_loss": 5.172320704306325, "within_bucket_gap": -0.6474173581103093}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.128457593202755e-06, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.128457593202755e-06, "step": 300, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.6415106909615655, "val_fraction": 0.000594558599695586, "val_loss": 5.634566955566406, "within_bucket_gap": -0.006943735395159578}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013044026735138466, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013044026735138466, "step": 300, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.84264901968149, "val_fraction": 0.0007372526636225266, "val_loss": 6.665721524146296, "within_bucket_gap": -0.17692749553519427}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015704088758437853, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015704088758437853, "step": 300, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 6.351164048717868, "val_fraction": 0.0006183409436834094, "val_loss": 6.605135404146635, "within_bucket_gap": 0.25397135542876725}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017376659123320582, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017376659123320582, "step": 300, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.847994123186384, "val_fraction": 0.0006659056316590563, "val_loss": 6.108941895621164, "within_bucket_gap": 0.26094777243477996}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00044781182817491054, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00044781182817491054, "step": 300, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 5.3319411277771, "val_fraction": 0.00047564687975646877, "val_loss": 6.273420715332032, "within_bucket_gap": 0.941479587554932}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017983770080171572, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017983770080171572, "step": 300, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 5.139170837402344, "val_fraction": 0.0003091704718417047, "val_loss": 5.720848963810847, "within_bucket_gap": 0.5816781264085034}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005851377852327489, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005851377852327489, "step": 300, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.45651033946446, "val_fraction": 0.000594558599695586, "val_loss": 5.440665283203125, "within_bucket_gap": 0.9841549437386652}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.2497923689889998e-05, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.2497923689889998e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.680484771728516, "val_fraction": 0.0004280821917808219, "val_loss": 6.709679921468099, "within_bucket_gap": 0.029195149739583037}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0009206325055008327, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0009206325055008327, "step": 300, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 7.099835863300398, "val_fraction": 0.000594558599695586, "val_loss": 5.551405639648437, "within_bucket_gap": -1.5484302236519607}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004646373402120595, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004646373402120595, "step": 300, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 6.561986403031782, "val_fraction": 0.00045186453576864533, "val_loss": 5.533719514545641, "within_bucket_gap": -1.028266888486141}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000399113901293482, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000399113901293482, "step": 300, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 4.618543686405305, "val_fraction": 0.00040429984779299846, "val_loss": 5.605716705322266, "within_bucket_gap": 0.9871730189169607}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00035067759712557287, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00035067759712557287, "step": 300, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.584292623731825, "val_fraction": 0.00033295281582952814, "val_loss": 6.637527738298688, "within_bucket_gap": 1.0532351145668635}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005183701730890359, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005183701730890359, "step": 300, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 7.62872040271759, "val_fraction": 0.0002616057838660578, "val_loss": 5.647226853804155, "within_bucket_gap": -1.981493548913435}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003054324047992462, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003054324047992462, "step": 300, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.752291361490886, "val_fraction": 0.0005232115677321156, "val_loss": 6.168526736172763, "within_bucket_gap": -0.583764625318123}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001415681606134989, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001415681606134989, "step": 300, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.066908200581868, "val_fraction": 0.0002616057838660578, "val_loss": 7.608058929443359, "within_bucket_gap": 0.5411507288614912}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.5767242991942492e-05, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.5767242991942492e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 6.433610916137695, "val_fraction": 0.0002616057838660578, "val_loss": 6.493881919167259, "within_bucket_gap": 0.06027100302956345}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00026760707344275806, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00026760707344275806, "step": 300, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.4483184814453125, "val_fraction": 0.00014269406392694063, "val_loss": 7.323708852132161, "within_bucket_gap": 1.8753903706868487}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.479496960468173e-05, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.479496960468173e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.917936642964681, "val_fraction": 0.0002616057838660578, "val_loss": 6.746705835515803, "within_bucket_gap": -0.17123080744887798}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00027005501617495954, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00027005501617495954, "step": 300, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.578541682316707, "val_fraction": 0.00028538812785388126, "val_loss": 5.632268905639648, "within_bucket_gap": -0.9462727766770582}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006409277246422962, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006409277246422962, "step": 300, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.700350125630696, "val_fraction": 0.0003567351598173516, "val_loss": 5.9037015279134115, "within_bucket_gap": -1.7966485977172848}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.240491111891822e-05, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.240491111891822e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.0578837394714355, "val_fraction": 0.0002616057838660578, "val_loss": 6.181752985174006, "within_bucket_gap": 0.1238692457025703}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.788751182078462e-05, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.788751182078462e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.693676515059038, "val_fraction": 9.512937595129375e-05, "val_loss": 4.87492299079895, "within_bucket_gap": -0.8187535242600879}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004507826125785096, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004507826125785096, "step": 300, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 5.031452026367187, "val_fraction": 0.00021404109589041095, "val_loss": 7.137508392333984, "within_bucket_gap": 2.106056365966797}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002520996736649287, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002520996736649287, "step": 300, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 6.56039184306091, "val_fraction": 0.00014269406392694063, "val_loss": 4.79367733001709, "within_bucket_gap": -1.7667145130438202}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00046247538729163974, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00046247538729163974, "step": 300, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.660294532775879, "val_fraction": 0.00023782343987823439, "val_loss": 5.604911041259766, "within_bucket_gap": 1.944616508483887}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.345472440690568e-05, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.345472440690568e-05, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.342243671417236, "val_fraction": 0.0002616057838660578, "val_loss": 4.661253148859197, "within_bucket_gap": 0.31900947744196095}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.252186433446001e-05, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.252186433446001e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.972749498155382, "val_fraction": 0.00016647640791476407, "val_loss": 6.047966548374721, "within_bucket_gap": 0.07521705021933922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004115736103493329, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004115736103493329, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.5618085861206055, "val_fraction": 0.000380517503805175, "val_loss": 7.643424034118652, "within_bucket_gap": 1.0816154479980469}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.037302509290442e-05, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.037302509290442e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.948923110961914, "val_fraction": 7.134703196347032e-05, "val_loss": 7.2155914306640625, "within_bucket_gap": 1.2666683197021484}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.5669438528687985e-06, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.5669438528687985e-06, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.55830317735672, "val_fraction": 4.756468797564688e-05, "val_loss": 4.591246604919434, "within_bucket_gap": 0.03294342756271362}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.049201316188214e-05, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.049201316188214e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 6.68939547105269, "val_fraction": 4.756468797564688e-05, "val_loss": 6.25857138633728, "within_bucket_gap": -0.4308240847154101}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.713550648650344e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.713550648650344e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.785021868619052, "val_fraction": 0.00016647640791476407, "val_loss": 5.3817484719412665, "within_bucket_gap": -0.4032733966777853}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002714044200412396, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002714044200412396, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.605963706970215, "val_fraction": 7.134703196347032e-05, "val_loss": 3.8019593556722007, "within_bucket_gap": -3.804004351298014}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000256206976360382, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000256206976360382, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.061208724975586, "val_fraction": 4.756468797564688e-05, "val_loss": 1.6747132539749146, "within_bucket_gap": -5.386495471000671}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00024171043159029076, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00024171043159029076, "step": 300, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 7.319612979888916, "val_fraction": 0.00014269406392694063, "val_loss": 5.625706275304158, "within_bucket_gap": -1.6939067045847578}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.4953372572836444e-06, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.4953372572836444e-06, "step": 300, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 3.8152449131011963, "val_fraction": 2.378234398782344e-05, "val_loss": 3.6682729721069336, "within_bucket_gap": -0.1469719409942627}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014078486909242154, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014078486909242154, "step": 300, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.504327583312988, "val_fraction": 0.00016647640791476407, "val_loss": 5.658652986798968, "within_bucket_gap": -0.8456745965140202}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010074722712442754, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010074722712442754, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.2466780344645185, "val_fraction": 0.00014269406392694063, "val_loss": 5.952714602152507, "within_bucket_gap": 0.7060365676879883}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010835367454785743, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010835367454785743, "step": 300, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.226299905776978, "val_fraction": 0.00011891171993911719, "val_loss": 4.315088844299316, "within_bucket_gap": -0.9112110614776618}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001869105551068641, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001869105551068641, "step": 300, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 5.11240953207016, "val_fraction": 9.512937595129375e-05, "val_loss": 7.077213287353516, "within_bucket_gap": 1.9648037552833557}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006605717795322174, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006605717795322174, "step": 300, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.5994766553243, "val_fraction": 0.00023782343987823439, "val_loss": 7.3770488739013675, "within_bucket_gap": 2.7775722185770677}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.8069252394468653e-05, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.8069252394468653e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.249039840698242, "val_fraction": 0.00021404109589041095, "val_loss": 6.164620293511285, "within_bucket_gap": -0.08441954718695754}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.414676653921514e-05, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.414676653921514e-05, "step": 300, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.861659812927246, "val_fraction": 0.00011891171993911719, "val_loss": 6.5693004608154295, "within_bucket_gap": 0.7076406478881836}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.320304303408758e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.320304303408758e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.935915470123291, "val_fraction": 7.134703196347032e-05, "val_loss": 6.9619293212890625, "within_bucket_gap": 1.0260138511657715}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004162403331046052, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004162403331046052, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.578874826431274, "val_fraction": 0.00016647640791476407, "val_loss": 8.079171044485909, "within_bucket_gap": 2.5002962180546344}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00027764980590990634, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00027764980590990634, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.814615090688069, "val_fraction": 0.0001902587519025875, "val_loss": 8.273942470550537, "within_bucket_gap": 1.4593273798624677}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.387241466948977e-05, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.387241466948977e-05, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.4294114112854, "val_fraction": 0.0002616057838660578, "val_loss": 6.558890256014737, "within_bucket_gap": 0.1294788447293369}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.117279353569995e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.117279353569995e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.861262321472168, "val_fraction": 4.756468797564688e-05, "val_loss": 7.575165510177612, "within_bucket_gap": -1.2860968112945557}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.62788365774503e-05, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.62788365774503e-05, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.586252331733704, "val_fraction": 7.134703196347032e-05, "val_loss": 7.79744815826416, "within_bucket_gap": -0.7888041734695435}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.424984991459723e-06, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.424984991459723e-06, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.31827449798584, "val_fraction": 4.756468797564688e-05, "val_loss": 7.141147613525391, "within_bucket_gap": -0.17712688446044922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00028791801388768117, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00028791801388768117, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.763734817504883, "val_fraction": 9.512937595129375e-05, "val_loss": 7.7903289794921875, "within_bucket_gap": 3.0265941619873047}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.585193898216956e-05, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.585193898216956e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.200517654418945, "val_fraction": 9.512937595129375e-05, "val_loss": 4.718522071838379, "within_bucket_gap": -0.4819955825805664}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018285515645867614, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018285515645867614, "step": 300, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.137932968139649, "val_fraction": 7.134703196347032e-05, "val_loss": 3.5750350952148438, "within_bucket_gap": -2.562897872924805}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003274166602159381, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003274166602159381, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.555917739868164, "val_fraction": 0.0001902587519025875, "val_loss": 4.276819705963135, "within_bucket_gap": 1.7209019660949707}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.077625738612774e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.077625738612774e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.8056864738464355, "val_fraction": 2.378234398782344e-05, "val_loss": 2.829686403274536, "within_bucket_gap": -2.9760000705718994}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001779649820592668, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001779649820592668, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.903311014175415, "val_fraction": 7.134703196347032e-05, "val_loss": 5.397668202718099, "within_bucket_gap": 2.4943571885426836}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.668474924437353e-05, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.668474924437353e-05, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.7075135707855225, "val_fraction": 0.00011891171993911719, "val_loss": 4.062625503540039, "within_bucket_gap": -0.6448880672454838}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002004358480872992, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002004358480872992, "step": 300, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.422586917877197, "val_fraction": 7.134703196347032e-05, "val_loss": 8.231895764668783, "within_bucket_gap": 2.809308846791586}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.916427681253744e-05, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.916427681253744e-05, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.866078615188599, "val_fraction": 9.512937595129375e-05, "val_loss": 7.067533493041992, "within_bucket_gap": 0.20145487785339355}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.882610321044922, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.88838842397774e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.88838842397774e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.484771251678467, "val_fraction": 4.756468797564688e-05, "val_loss": 5.616076469421387, "within_bucket_gap": -1.86869478225708}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022606152468438923, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022606152468438923, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.095997333526611, "val_fraction": 4.756468797564688e-05, "val_loss": 2.3432798385620117, "within_bucket_gap": -4.7527174949646}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.627617597579956, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004202880452938457, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004202880452938457, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.527581691741943, "val_fraction": 0.0001902587519025875, "val_loss": 4.31854772567749, "within_bucket_gap": -2.209033966064453}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010791994756033791, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010791994756033791, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.957828521728516, "val_fraction": 7.134703196347032e-05, "val_loss": 7.445222536722819, "within_bucket_gap": -1.5126059850056963}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.673952322572332e-05, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.673952322572332e-05, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.108492851257324, "val_fraction": 4.756468797564688e-05, "val_loss": 6.284881114959717, "within_bucket_gap": -1.8236117362976074}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.581637042116719e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.581637042116719e-05, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.286571502685547, "val_fraction": 9.512937595129375e-05, "val_loss": 4.699829816818237, "within_bucket_gap": -0.5867416858673096}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.820577397919862e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.820577397919862e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.782842636108398, "val_fraction": 9.512937595129375e-05, "val_loss": 5.7505035400390625, "within_bucket_gap": -1.032339096069336}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.31554361257016e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.31554361257016e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.875121116638184, "val_fraction": 2.378234398782344e-05, "val_loss": 3.2195613384246826, "within_bucket_gap": -2.655559778213501}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.062322027244162e-06, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.062322027244162e-06, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.005253791809082, "val_fraction": 2.378234398782344e-05, "val_loss": 4.3442583084106445, "within_bucket_gap": 0.3390045166015625}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001243363927911829, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001243363927911829, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.361828486124675, "val_fraction": 4.756468797564688e-05, "val_loss": 7.975876808166504, "within_bucket_gap": 2.614048322041829}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001528121111418377, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001528121111418377, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7741518020629883, "val_fraction": 2.378234398782344e-05, "val_loss": 9.19959545135498, "within_bucket_gap": 6.425443649291992}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.2743844800888126e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.2743844800888126e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.344381809234619, "val_fraction": 4.756468797564688e-05, "val_loss": 6.445735216140747, "within_bucket_gap": -0.8986465930938721}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011362890107628054, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011362890107628054, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.425691843032837, "val_fraction": 2.378234398782344e-05, "val_loss": 8.203559875488281, "within_bucket_gap": 4.777868032455444}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015666282521716715, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015666282521716715, "step": 300, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.606755256652832, "val_fraction": 2.378234398782344e-05, "val_loss": 11.194113731384277, "within_bucket_gap": 6.587358474731445}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.330710734588007e-06, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.330710734588007e-06, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.512000143527985, "val_fraction": 4.756468797564688e-05, "val_loss": 4.666121006011963, "within_bucket_gap": 0.15412086248397827}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003277231206269751, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003277231206269751, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.272541046142578, "val_fraction": 7.134703196347032e-05, "val_loss": 4.6791737874348955, "within_bucket_gap": -4.593367258707683}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.275297999382019, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.753539085388184, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.1797124288337345e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.1797124288337345e-05, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.228095690409343, "val_fraction": 2.378234398782344e-05, "val_loss": 7.406061172485352, "within_bucket_gap": 2.177965482076009}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.9916725158691406, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.504803657531738, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.486604655170005e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.486604655170005e-05, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.807104825973511, "val_fraction": 2.378234398782344e-05, "val_loss": 7.2731523513793945, "within_bucket_gap": 1.4660475254058838}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.478935718536377, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.170896053314209, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.162546157836914, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00030936006473624, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00030936006473624, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.08161199092865, "val_fraction": 4.756468797564688e-05, "val_loss": 8.58559799194336, "within_bucket_gap": 6.5039860010147095}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.0809319433011964e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.0809319433011964e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.212743759155273, "val_fraction": 2.378234398782344e-05, "val_loss": 6.076313495635986, "within_bucket_gap": -2.136430263519287}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003910616913433307, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003910616913433307, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.525775671005249, "val_fraction": 7.134703196347032e-05, "val_loss": 7.006896336873372, "within_bucket_gap": 5.481120665868123}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.451906204223633, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00040494190185944603, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00040494190185944603, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.744246959686279, "val_fraction": 0.00014269406392694063, "val_loss": 4.906414111455281, "within_bucket_gap": -2.837832848230998}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.924688204964183e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.924688204964183e-05, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.136268615722656, "val_fraction": 4.756468797564688e-05, "val_loss": 8.222835063934326, "within_bucket_gap": 2.08656644821167}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.034986818943575e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.034986818943575e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6688308715820312, "val_fraction": 2.378234398782344e-05, "val_loss": 7.467862129211426, "within_bucket_gap": 3.7990312576293945}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016530964926134687, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016530964926134687, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.353569030761719, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4026288986206055, "within_bucket_gap": -6.950940132141113}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.682459037960755e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.682459037960755e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.272237777709961, "val_fraction": 2.378234398782344e-05, "val_loss": 7.564797401428223, "within_bucket_gap": -0.7074403762817383}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.834115982055664, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.787428855895996, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.9321017149195095e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.9321017149195095e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.600111961364746, "val_fraction": 4.756468797564688e-05, "val_loss": 4.216557025909424, "within_bucket_gap": 0.6164450645446777}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.986868381500244, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.5135462284088135, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.611411094665527, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.067507743835449, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.04359181195815e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.04359181195815e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.0412209033966064, "val_fraction": 2.378234398782344e-05, "val_loss": 5.4233903884887695, "within_bucket_gap": 3.382169485092163}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.192363739013672, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.6172609329223633, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.05474090203642845, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.631690979003906, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.6896511822530666e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.6896511822530666e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.729397296905518, "val_fraction": 4.756468797564688e-05, "val_loss": 5.533205032348633, "within_bucket_gap": -1.1961922645568848}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.425580024719238, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.562618732452393, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.075861930847168, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.38508794296822e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.38508794296822e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.480013847351074, "val_fraction": 2.378234398782344e-05, "val_loss": 5.056652069091797, "within_bucket_gap": -1.4233617782592773}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.1716435819274575e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.1716435819274575e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.6334919929504395, "val_fraction": 2.378234398782344e-05, "val_loss": 8.387584686279297, "within_bucket_gap": 1.7540926933288574}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.998253240614358e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.998253240614358e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.3313140869140625, "val_fraction": 2.378234398782344e-05, "val_loss": 4.388688564300537, "within_bucket_gap": -2.9426255226135254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.838747978210449, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.096197605133057, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.479950428009033, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003738860901632265, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003738860901632265, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.8312232494354248, "val_fraction": 4.756468797564688e-05, "val_loss": 9.6918044090271, "within_bucket_gap": 7.860581159591675}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.859506607055664, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.612155914306641, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.4529337882995605, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.651346206665039, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.649660110473633, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.7476212637700264e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.7476212637700264e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.191475212574005, "val_fraction": 2.378234398782344e-05, "val_loss": 2.7747154235839844, "within_bucket_gap": -2.4167597889900208}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.614273071289062, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.6618072986602783, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.5107135159421366e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.5107135159421366e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.9795870780944824, "val_fraction": 2.378234398782344e-05, "val_loss": 3.344362258911133, "within_bucket_gap": -0.6352248191833496}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.197959894458997e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.197959894458997e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.5901503562927246, "val_fraction": 2.378234398782344e-05, "val_loss": 4.355308532714844, "within_bucket_gap": 1.7651581764221191}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.6814122200012207, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.452616691589355, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.069194197654724, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.7186689376831055, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.642879486083984, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.665305137634277, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.9772199392318726, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.8057503010765784e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.8057503010765784e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7907276153564453, "val_fraction": 2.378234398782344e-05, "val_loss": 2.6109657287597656, "within_bucket_gap": -1.1797618865966797}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.480751037597656, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001624225606203442, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001624225606203442, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.068685531616211, "val_fraction": 2.378234398782344e-05, "val_loss": 2.2391417026519775, "within_bucket_gap": -6.829543828964233}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011812219276215326, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011812219276215326, "step": 300, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.763380368550618, "val_fraction": 2.378234398782344e-05, "val_loss": 2.7965784072875977, "within_bucket_gap": -4.9668019612630205}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.679280534363953e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.679280534363953e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.891845703125, "val_fraction": 2.378234398782344e-05, "val_loss": 8.120829582214355, "within_bucket_gap": 3.2289838790893555}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.0414276123046875, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.4942605566216384e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.4942605566216384e-05, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.049117803573608, "val_fraction": 2.378234398782344e-05, "val_loss": 8.359344482421875, "within_bucket_gap": 2.3102266788482666}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.957880020141602, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.1580322980880737, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.291138648986816, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.360133647918701, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.768321990966797, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.6276726722717285, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.121036529541016, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.8672077655792236, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.743687152862549, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.581315517425537, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011806746255861568, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011806746255861568, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.8074357509613037, "val_fraction": 2.378234398782344e-05, "val_loss": 8.771936416625977, "within_bucket_gap": 4.964500665664673}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.584356307983398, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.419151782989502, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.2718799123299537e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.2718799123299537e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.3997292518615723, "val_fraction": 2.378234398782344e-05, "val_loss": 3.3550093173980713, "within_bucket_gap": 0.955280065536499}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011992468704190247, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011992468704190247, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.296400785446167, "val_fraction": 2.378234398782344e-05, "val_loss": 7.338994026184082, "within_bucket_gap": 5.042593240737915}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.185527801513672, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.075313568115234, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.05006144195795059, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010321228630285103, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010321228630285103, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.304344415664673, "val_fraction": 2.378234398782344e-05, "val_loss": 9.644214630126953, "within_bucket_gap": 4.33987021446228}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012801350569253278, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012801350569253278, "step": 300, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.580368280410767, "val_fraction": 4.756468797564688e-05, "val_loss": 2.889012336730957, "within_bucket_gap": -2.6913559436798096}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.589831829071045, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.90925931930542, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.46795129776001, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.5694445371627808, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.39153018593788147, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.12614676356315613, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.456954002380371, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.019739028066396713, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.050238747937857e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.050238747937857e-05, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.828929901123047, "val_fraction": 2.378234398782344e-05, "val_loss": 5.691014289855957, "within_bucket_gap": 0.8620843887329102}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.452451705932617, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.840707778930664, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.3785462878428997e-07, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.3785462878428997e-07, "step": 300, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.5195840001106262, "val_fraction": 2.378234398782344e-05, "val_loss": 0.5053778886795044, "within_bucket_gap": -0.014206111431121826}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.727291326000265e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.727291326000265e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5178194046020508, "val_fraction": 4.756468797564688e-05, "val_loss": 0.5239536762237549, "within_bucket_gap": -0.9938657283782959}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.238911943711465e-05, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.238911943711465e-05, "step": 300, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.912688255310059, "val_fraction": 9.512937595129375e-05, "val_loss": 7.148042678833008, "within_bucket_gap": 0.23535442352294922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.014691605910561225, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.014691605910561225, "step": 400, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 6.1335491670030855, "val_fraction": 0.05995528919330289, "val_loss": 5.8885064675476, "within_bucket_gap": -0.2450426994554853}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0031432657074733053, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0031432657074733053, "step": 400, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 6.195345872905607, "val_fraction": 0.019882039573820395, "val_loss": 6.0372501355038874, "within_bucket_gap": -0.15809573740171956}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.003017059970358518, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.003017059970358518, "step": 400, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 6.101125873766447, "val_fraction": 0.011582001522070015, "val_loss": 5.840630313943788, "within_bucket_gap": -0.26049555982265904}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0010671910445341091, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0010671910445341091, "step": 400, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.974918191474781, "val_fraction": 0.007301179604261796, "val_loss": 5.82875125648921, "within_bucket_gap": -0.14616693498557076}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0013342079241574604, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0013342079241574604, "step": 400, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 6.112335684411812, "val_fraction": 0.004827815829528158, "val_loss": 5.835977187884852, "within_bucket_gap": -0.2763584965269601}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0015291614732806428, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0015291614732806428, "step": 400, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 6.3818082335992905, "val_fraction": 0.003305745814307458, "val_loss": 5.919231387351057, "within_bucket_gap": -0.4625768462482336}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0013405047135848073, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0013405047135848073, "step": 400, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 6.180149320922146, "val_fraction": 0.002877663622526636, "val_loss": 5.714318393675749, "within_bucket_gap": -0.4658309272463965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00033861151154234754, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00033861151154234754, "step": 400, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.815540858677456, "val_fraction": 0.002449581430745814, "val_loss": 5.677308462198498, "within_bucket_gap": -0.13823239647895758}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00019988557019234396, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00019988557019234396, "step": 400, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 5.892882201048705, "val_fraction": 0.0025684931506849314, "val_loss": 5.815060085720486, "within_bucket_gap": -0.07782211532821925}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000832824417015669, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000832824417015669, "step": 400, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 5.2595766028579405, "val_fraction": 0.0014745053272450532, "val_loss": 5.824392749417212, "within_bucket_gap": 0.5648161465592718}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008845743774380979, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008845743774380979, "step": 400, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 6.012873610405073, "val_fraction": 0.0020214992389649923, "val_loss": 5.5752902760225185, "within_bucket_gap": -0.4375833343825546}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003589300381115154, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003589300381115154, "step": 400, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 5.337212117513021, "val_fraction": 0.0016647640791476407, "val_loss": 5.121607971191406, "within_bucket_gap": -0.2156041463216143}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008210446480426493, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008210446480426493, "step": 400, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.424765530754538, "val_fraction": 0.0018074581430745813, "val_loss": 5.879019285503187, "within_bucket_gap": 0.45425375474864893}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000523353361573255, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000523353361573255, "step": 400, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.750453051398782, "val_fraction": 0.0015220700152207, "val_loss": 5.406609892845154, "within_bucket_gap": -0.34384315855362857}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001817336225476216, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001817336225476216, "step": 400, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 6.981346035399378, "val_fraction": 0.0014745053272450532, "val_loss": 5.74884033203125, "within_bucket_gap": -1.232505703368128}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0015524029731750488, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0015524029731750488, "step": 400, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 7.316045095754224, "val_fraction": 0.0010226407914764079, "val_loss": 5.798011602357376, "within_bucket_gap": -1.5180334933968478}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.898832426689016e-05, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.898832426689016e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.843631578528362, "val_fraction": 0.0004280821917808219, "val_loss": 5.729194853040907, "within_bucket_gap": -0.11443672548745543}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005746691158895272, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005746691158895272, "step": 400, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.746514297113186, "val_fraction": 0.0007372526636225266, "val_loss": 4.9670405234059976, "within_bucket_gap": -0.7794737737071884}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.239683754471594e-05, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.239683754471594e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.65426390511649, "val_fraction": 0.000594558599695586, "val_loss": 5.515679016113281, "within_bucket_gap": -0.13858488900320864}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010083472892029663, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010083472892029663, "step": 400, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.2514495849609375, "val_fraction": 0.0007372526636225266, "val_loss": 6.3882205101751515, "within_bucket_gap": 0.13677092521421397}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00028115619514829825, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00028115619514829825, "step": 400, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.902157568162488, "val_fraction": 0.0006183409436834094, "val_loss": 6.356852017916166, "within_bucket_gap": 0.4546944497536787}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001335875265855948, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001335875265855948, "step": 400, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.662931714739118, "val_fraction": 0.0006659056316590563, "val_loss": 5.863542011805943, "within_bucket_gap": 0.20061029706682465}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003862433033446741, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003862433033446741, "step": 400, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 5.084352612495422, "val_fraction": 0.00047564687975646877, "val_loss": 5.896390533447265, "within_bucket_gap": 0.8120379209518429}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003198172331219211, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003198172331219211, "step": 400, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.762265625, "val_fraction": 0.0003091704718417047, "val_loss": 5.796702164870042, "within_bucket_gap": 1.0344365398700415}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006043048225996103, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006043048225996103, "step": 400, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.384072644369943, "val_fraction": 0.000594558599695586, "val_loss": 5.40046501159668, "within_bucket_gap": 1.0163923672267368}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013605676466652954, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013605676466652954, "step": 400, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.34061845143636, "val_fraction": 0.0004280821917808219, "val_loss": 6.022789849175347, "within_bucket_gap": -0.31782860226101306}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007013100968957831, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007013100968957831, "step": 400, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 6.977696587057674, "val_fraction": 0.000594558599695586, "val_loss": 5.798149108886719, "within_bucket_gap": -1.1795474781709556}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001624699399005135, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001624699399005135, "step": 400, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.503832730379972, "val_fraction": 0.00045186453576864533, "val_loss": 5.8633872584292765, "within_bucket_gap": 0.3595545280493049}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007640227376200195, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007640227376200195, "step": 400, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 3.6113239411384828, "val_fraction": 0.00040429984779299846, "val_loss": 5.5010667688706345, "within_bucket_gap": 1.8897428277321517}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00044734523761258324, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00044734523761258324, "step": 400, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.1484875149197045, "val_fraction": 0.00033295281582952814, "val_loss": 6.49205698285784, "within_bucket_gap": 1.3435694679381358}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005381801158552663, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005381801158552663, "step": 400, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 7.3690255880355835, "val_fraction": 0.0002616057838660578, "val_loss": 5.311807632446289, "within_bucket_gap": -2.0572179555892944}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004692404993532391, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004692404993532391, "step": 400, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.614248593648274, "val_fraction": 0.0005232115677321156, "val_loss": 5.717402024702593, "within_bucket_gap": -0.8968465689456817}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0651684182359283e-05, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0651684182359283e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.76062552134196, "val_fraction": 0.0002616057838660578, "val_loss": 6.719908974387429, "within_bucket_gap": -0.040716546954531196}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.8730442092117494e-05, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.8730442092117494e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 6.0477800369262695, "val_fraction": 0.0002616057838660578, "val_loss": 6.11937800320712, "within_bucket_gap": 0.07159796628085058}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003036206851811169, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003036206851811169, "step": 400, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.025333722432454, "val_fraction": 0.00014269406392694063, "val_loss": 7.153107484181722, "within_bucket_gap": 2.1277737617492676}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010982371265666859, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010982371265666859, "step": 400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.682501792907715, "val_fraction": 0.0002616057838660578, "val_loss": 6.26269565929066, "within_bucket_gap": -0.41980613361705466}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003304442530590592, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003304442530590592, "step": 400, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.404676290658804, "val_fraction": 0.00028538812785388126, "val_loss": 5.24679962793986, "within_bucket_gap": -1.1578766627189436}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006177239630320301, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006177239630320301, "step": 400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.331626574198405, "val_fraction": 0.0003567351598173516, "val_loss": 5.600022761027018, "within_bucket_gap": -1.731603813171387}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002789758429316806, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002789758429316806, "step": 400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.120680809020996, "val_fraction": 0.0002616057838660578, "val_loss": 6.1870786493474785, "within_bucket_gap": 1.0663978403264824}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.282847749951041e-06, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.282847749951041e-06, "step": 400, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.824397087097168, "val_fraction": 9.512937595129375e-05, "val_loss": 4.869418382644653, "within_bucket_gap": 0.04502129554748535}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00043309732477777435, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00043309732477777435, "step": 400, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.924581756591797, "val_fraction": 0.00021404109589041095, "val_loss": 6.948012457953559, "within_bucket_gap": 2.023430701361762}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00021256740848607476, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00021256740848607476, "step": 400, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 6.45542674869686, "val_fraction": 0.00014269406392694063, "val_loss": 4.965754350026448, "within_bucket_gap": -1.489672398670412}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004055397902967929, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004055397902967929, "step": 400, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.7847322736467635, "val_fraction": 0.00023782343987823439, "val_loss": 5.489945983886718, "within_bucket_gap": 1.705213710239955}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011987507887626896, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011987507887626896, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.671738624572754, "val_fraction": 0.0002616057838660578, "val_loss": 4.129966562444514, "within_bucket_gap": 0.45822793787175975}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.850081848287754e-05, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.850081848287754e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.4810536702473955, "val_fraction": 0.00016647640791476407, "val_loss": 6.072734015328543, "within_bucket_gap": 0.5916803450811479}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006453688830545504, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006453688830545504, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.60436487197876, "val_fraction": 0.000380517503805175, "val_loss": 7.300394296646118, "within_bucket_gap": 1.6960294246673584}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.933778046369288e-05, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.933778046369288e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.922020503452846, "val_fraction": 7.134703196347032e-05, "val_loss": 7.034018834431966, "within_bucket_gap": 1.1119983309791195}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.9153036348170157e-05, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.9153036348170157e-05, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.7315937876701355, "val_fraction": 4.756468797564688e-05, "val_loss": 3.908440351486206, "within_bucket_gap": -0.8231534361839294}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010623647253411397, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010623647253411397, "step": 400, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 6.193853118202903, "val_fraction": 4.756468797564688e-05, "val_loss": 3.960337519645691, "within_bucket_gap": -2.2335155985572124}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001517624937186553, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001517624937186553, "step": 400, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.69530539079146, "val_fraction": 0.00016647640791476407, "val_loss": 4.783689771379743, "within_bucket_gap": -0.9116156194117169}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000316981039090788, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000316981039090788, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.154791831970215, "val_fraction": 7.134703196347032e-05, "val_loss": 2.7119855880737305, "within_bucket_gap": -4.442806243896484}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001603564930742371, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001603564930742371, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.9508525133132935, "val_fraction": 4.756468797564688e-05, "val_loss": 1.5795176029205322, "within_bucket_gap": -3.3713349103927612}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017321897112926627, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017321897112926627, "step": 400, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.817238433020456, "val_fraction": 0.00014269406392694063, "val_loss": 5.603319883346558, "within_bucket_gap": -1.213918549673898}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.3997427053647498e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.3997427053647498e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 3.4652737379074097, "val_fraction": 2.378234398782344e-05, "val_loss": 4.47431755065918, "within_bucket_gap": 1.00904381275177}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.180001160144079e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.180001160144079e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.832535934448242, "val_fraction": 0.00016647640791476407, "val_loss": 5.581449236188616, "within_bucket_gap": -0.25108669825962604}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.336673375134054e-05, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.336673375134054e-05, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.490310033162435, "val_fraction": 0.00014269406392694063, "val_loss": 5.794224103291829, "within_bucket_gap": 0.30391407012939453}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00026600507387106095, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00026600507387106095, "step": 400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.328039169311523, "val_fraction": 0.00011891171993911719, "val_loss": 3.091042900085449, "within_bucket_gap": -2.2369962692260743}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019790798062784668, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019790798062784668, "step": 400, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.771282076835632, "val_fraction": 9.512937595129375e-05, "val_loss": 6.851690769195557, "within_bucket_gap": 2.0804086923599243}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005740735061818244, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005740735061818244, "step": 400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.729597330093384, "val_fraction": 0.00023782343987823439, "val_loss": 7.143461608886719, "within_bucket_gap": 2.4138642787933353}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00028996156412353986, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00028996156412353986, "step": 400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.6533763885498045, "val_fraction": 0.00021404109589041095, "val_loss": 5.298675960964626, "within_bucket_gap": -1.3547004275851782}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.942234339053595e-05, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.942234339053595e-05, "step": 400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.2829992294311525, "val_fraction": 0.00011891171993911719, "val_loss": 5.867377090454101, "within_bucket_gap": -0.41562213897705114}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.8665842349489325e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.8665842349489325e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.267204284667969, "val_fraction": 7.134703196347032e-05, "val_loss": 5.865423838297526, "within_bucket_gap": -0.4017804463704424}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00025883379790654106, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00025883379790654106, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.848153293132782, "val_fraction": 0.00016647640791476407, "val_loss": 7.40293094090053, "within_bucket_gap": 1.5547776477677484}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014121028325812945, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014121028325812945, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.470403989156087, "val_fraction": 0.0001902587519025875, "val_loss": 8.212605237960815, "within_bucket_gap": 0.7422012488047285}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.182172994091086e-05, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.182172994091086e-05, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.241112232208252, "val_fraction": 0.0002616057838660578, "val_loss": 6.157697677612305, "within_bucket_gap": -0.08341455459594727}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014540986292619864, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014540986292619864, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.971834182739258, "val_fraction": 4.756468797564688e-05, "val_loss": 5.914737224578857, "within_bucket_gap": -3.0570969581604004}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015830439111413477, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015830439111413477, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.747858047485352, "val_fraction": 7.134703196347032e-05, "val_loss": 6.529063701629639, "within_bucket_gap": -2.218794345855713}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.195516181318726e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.195516181318726e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.787494659423828, "val_fraction": 4.756468797564688e-05, "val_loss": 6.064469337463379, "within_bucket_gap": -1.7230253219604492}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002588190625973125, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002588190625973125, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.381875991821289, "val_fraction": 9.512937595129375e-05, "val_loss": 7.102581977844238, "within_bucket_gap": 2.720705986022949}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012364171592793688, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012364171592793688, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.1061930656433105, "val_fraction": 9.512937595129375e-05, "val_loss": 4.405914783477783, "within_bucket_gap": 1.2997217178344727}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020951021298244482, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020951021298244482, "step": 400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.9147186279296875, "val_fraction": 7.134703196347032e-05, "val_loss": 2.9782234827677407, "within_bucket_gap": -2.9364951451619468}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021526829837119742, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021526829837119742, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.120907783508301, "val_fraction": 0.0001902587519025875, "val_loss": 4.2523579597473145, "within_bucket_gap": 1.1314501762390137}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.546550765792347e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.546550765792347e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.861319065093994, "val_fraction": 2.378234398782344e-05, "val_loss": 2.688145399093628, "within_bucket_gap": -3.173173666000366}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016723640115656624, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016723640115656624, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.7090635299682617, "val_fraction": 7.134703196347032e-05, "val_loss": 5.053048928578694, "within_bucket_gap": 2.3439853986104326}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010067897833192007, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010067897833192007, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.459424614906311, "val_fraction": 0.00011891171993911719, "val_loss": 4.306094551086426, "within_bucket_gap": 0.8466699361801151}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016282308319387915, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016282308319387915, "step": 400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.656531651814778, "val_fraction": 7.134703196347032e-05, "val_loss": 7.9386599858601885, "within_bucket_gap": 2.28212833404541}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.110373168956924e-05, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.110373168956924e-05, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.991543292999268, "val_fraction": 9.512937595129375e-05, "val_loss": 6.8441057205200195, "within_bucket_gap": 0.852562427520752}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.8393354415893555, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.244616776296537e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.244616776296537e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.951578617095947, "val_fraction": 4.756468797564688e-05, "val_loss": 5.848950386047363, "within_bucket_gap": -1.102628231048584}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002194591875852878, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002194591875852878, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.202066421508789, "val_fraction": 4.756468797564688e-05, "val_loss": 2.5881564617156982, "within_bucket_gap": -4.613909959793091}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.145775556564331, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00031409005505490703, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00031409005505490703, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.424902439117432, "val_fraction": 0.0001902587519025875, "val_loss": 3.7740451097488403, "within_bucket_gap": -1.6508573293685913}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013562574234182975, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013562574234182975, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.150545120239258, "val_fraction": 7.134703196347032e-05, "val_loss": 7.249614715576172, "within_bucket_gap": -1.900930404663086}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010196848728522498, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010196848728522498, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.5024213790893555, "val_fraction": 4.756468797564688e-05, "val_loss": 5.358635902404785, "within_bucket_gap": -2.1437854766845703}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.667961274954556e-06, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.667961274954556e-06, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.567483266194661, "val_fraction": 9.512937595129375e-05, "val_loss": 4.528925657272339, "within_bucket_gap": -0.038557608922322295}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.846079476529241e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.846079476529241e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.904348373413086, "val_fraction": 9.512937595129375e-05, "val_loss": 6.93936824798584, "within_bucket_gap": 1.035019874572754}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.513265682319165e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.513265682319165e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.099898815155029, "val_fraction": 2.378234398782344e-05, "val_loss": 3.043120861053467, "within_bucket_gap": -1.0567779541015625}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.1703620354334513e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.1703620354334513e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.206451416015625, "val_fraction": 2.378234398782344e-05, "val_loss": 3.7143375873565674, "within_bucket_gap": -0.4921138286590576}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011542052000322538, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011542052000322538, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.835067431131999, "val_fraction": 4.756468797564688e-05, "val_loss": 7.26166844367981, "within_bucket_gap": 2.4266010125478106}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001525038759730178, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001525038759730178, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7109243869781494, "val_fraction": 2.378234398782344e-05, "val_loss": 10.123407363891602, "within_bucket_gap": 6.412482976913452}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.3579380566671014e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.3579380566671014e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.155820846557617, "val_fraction": 4.756468797564688e-05, "val_loss": 5.870327949523926, "within_bucket_gap": -0.2854928970336914}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014072633950859082, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014072633950859082, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.8117198944091797, "val_fraction": 2.378234398782344e-05, "val_loss": 8.728981018066406, "within_bucket_gap": 5.917261123657227}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014323354719980664, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014323354719980664, "step": 400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.501887226104737, "val_fraction": 2.378234398782344e-05, "val_loss": 10.524571418762207, "within_bucket_gap": 6.02268419265747}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.0384964594557948e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.0384964594557948e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.303201675415039, "val_fraction": 4.756468797564688e-05, "val_loss": 4.942015171051025, "within_bucket_gap": 0.6388134956359863}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003787627843416989, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003787627843416989, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.257732391357422, "val_fraction": 7.134703196347032e-05, "val_loss": 4.94899320602417, "within_bucket_gap": -5.308739185333252}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.747528394063314, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.767694473266602, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.516663457463078e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.516663457463078e-05, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.243396441141765, "val_fraction": 2.378234398782344e-05, "val_loss": 6.72208309173584, "within_bucket_gap": 1.4786866505940752}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.046268939971924, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.750438690185547, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.6359179192300614e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.6359179192300614e-05, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.518648386001587, "val_fraction": 2.378234398782344e-05, "val_loss": 7.047479152679443, "within_bucket_gap": 1.5288307666778564}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.923671007156372, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.034490346908569, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.9154052734375, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000328227034890443, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000328227034890443, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.6073771268129349, "val_fraction": 4.756468797564688e-05, "val_loss": 8.50802230834961, "within_bucket_gap": 6.9006451815366745}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.095159965381593e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.095159965381593e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.221155166625977, "val_fraction": 2.378234398782344e-05, "val_loss": 6.340182304382324, "within_bucket_gap": -0.8809728622436523}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012207719606533078, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012207719606533078, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.8207926750183105, "val_fraction": 7.134703196347032e-05, "val_loss": 3.531826655069987, "within_bucket_gap": 1.7110339800516763}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.584195137023926, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00035861636904639557, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00035861636904639557, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.402681350708008, "val_fraction": 0.00014269406392694063, "val_loss": 3.8894978364308677, "within_bucket_gap": -2.51318351427714}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010734525112740466, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010734525112740466, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.7751234372456866, "val_fraction": 4.756468797564688e-05, "val_loss": 8.031949996948242, "within_bucket_gap": 2.2568265597025556}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011867287609856604, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011867287609856604, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.9161789417266846, "val_fraction": 2.378234398782344e-05, "val_loss": 7.9061360359191895, "within_bucket_gap": 4.989957094192505}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001709044569340652, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001709044569340652, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.492205619812012, "val_fraction": 2.378234398782344e-05, "val_loss": 2.3060150146484375, "within_bucket_gap": -7.186190605163574}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.2392193273140774e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.2392193273140774e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.96092414855957, "val_fraction": 2.378234398782344e-05, "val_loss": 8.743431091308594, "within_bucket_gap": 1.7825069427490234}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.769642114639282, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.182008743286133, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.7868392079206722e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.7868392079206722e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.796323299407959, "val_fraction": 4.756468797564688e-05, "val_loss": 4.382228374481201, "within_bucket_gap": 0.5859050750732422}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.144511699676514, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.2027206420898438, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.3070855140686035, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.956567764282227, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.329998877585568e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.329998877585568e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.4571360349655151, "val_fraction": 2.378234398782344e-05, "val_loss": 4.959733963012695, "within_bucket_gap": 3.50259792804718}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1606762409210205, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.9458611011505127, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.08521035313606262, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.6285905838012695, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.890473185790365e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.890473185790365e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.892983436584473, "val_fraction": 4.756468797564688e-05, "val_loss": 5.075050354003906, "within_bucket_gap": -0.8179330825805664}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.253958702087402, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.917013645172119, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.586730003356934, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.4460907860615119e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.4460907860615119e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.6723785400390625, "val_fraction": 2.378234398782344e-05, "val_loss": 4.064326286315918, "within_bucket_gap": -0.6080522537231445}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.875791965191404e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.875791965191404e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.0237860679626465, "val_fraction": 2.378234398782344e-05, "val_loss": 7.812519073486328, "within_bucket_gap": 0.7887330055236816}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.255803451146165e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.255803451146165e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.7133588790893555, "val_fraction": 2.378234398782344e-05, "val_loss": 4.082918643951416, "within_bucket_gap": -2.6304402351379395}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.489222526550293, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.701934337615967, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.7295966148376465, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004005746085259105, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004005746085259105, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.5784462690353394, "val_fraction": 4.756468797564688e-05, "val_loss": 10.000126838684082, "within_bucket_gap": 8.421680569648743}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.2292121648788452, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.181527137756348, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.134507179260254, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1910319328308105, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.474652290344238, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.4201401792795505e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.4201401792795505e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.054681718349457, "val_fraction": 2.378234398782344e-05, "val_loss": 2.775621175765991, "within_bucket_gap": -2.2790605425834656}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.436336517333984, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.224312424659729, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.672646869444593e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.672646869444593e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.2102246284484863, "val_fraction": 2.378234398782344e-05, "val_loss": 2.086430072784424, "within_bucket_gap": -1.1237945556640625}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.4690345224543067e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.4690345224543067e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7867894172668457, "val_fraction": 2.378234398782344e-05, "val_loss": 2.748609781265259, "within_bucket_gap": -1.038179636001587}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.050306558609009, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.622091293334961, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.786350846290588, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.873661994934082, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.693995475769043, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.953819751739502, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.980093240737915, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.5689528492850617e-06, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.5689528492850617e-06, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.5748608112335205, "val_fraction": 2.378234398782344e-05, "val_loss": 2.724928140640259, "within_bucket_gap": 0.15006732940673828}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.297596454620361, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001545472572383271, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001545472572383271, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.040979385375977, "val_fraction": 2.378234398782344e-05, "val_loss": 2.542576313018799, "within_bucket_gap": -6.498403072357178}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013004475094109604, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013004475094109604, "step": 400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 8.022002855936686, "val_fraction": 2.378234398782344e-05, "val_loss": 2.5538811683654785, "within_bucket_gap": -5.468121687571207}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.376204606786348e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.376204606786348e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.043494701385498, "val_fraction": 2.378234398782344e-05, "val_loss": 7.7245612144470215, "within_bucket_gap": 2.6810665130615234}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.270880699157715, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.716498251737888e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.716498251737888e-05, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.777308225631714, "val_fraction": 2.378234398782344e-05, "val_loss": 8.601461410522461, "within_bucket_gap": 2.824153184890747}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.97082233428955, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.5358507037162781, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.360548973083496, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.798493385314941, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.353928565979004, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.7468252182006836, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.499528884887695, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.2700513005256653, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.480453968048096, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.529045343399048, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.412315035338089e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.412315035338089e-05, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.126178503036499, "val_fraction": 2.378234398782344e-05, "val_loss": 6.663388729095459, "within_bucket_gap": 3.53721022605896}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.21610164642334, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.484703540802002, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.3887426092018636e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.3887426092018636e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.8396142721176147, "val_fraction": 2.378234398782344e-05, "val_loss": 2.8440327644348145, "within_bucket_gap": 1.0044184923171997}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012052597838086805, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012052597838086805, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.4532084465026855, "val_fraction": 2.378234398782344e-05, "val_loss": 7.521084785461426, "within_bucket_gap": 5.06787633895874}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.781436920166016, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.219301223754883, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.028097914531826973, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.954847685641168e-05, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.954847685641168e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.422050476074219, "val_fraction": 2.378234398782344e-05, "val_loss": 8.766904830932617, "within_bucket_gap": 3.3448543548583984}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011124057040366951, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011124057040366951, "step": 400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.495800018310547, "val_fraction": 4.756468797564688e-05, "val_loss": 3.157078266143799, "within_bucket_gap": -2.338721752166748}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.969113349914551, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.747060775756836, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.300612449645996, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.3966299891471863, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.10144676268100739, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.09045321494340897, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.30977725982666, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.06604339182376862, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.959181413984371e-07, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.959181413984371e-07, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.915311336517334, "val_fraction": 2.378234398782344e-05, "val_loss": 4.8776397705078125, "within_bucket_gap": -0.037671566009521484}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.269659042358398, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.583558082580566, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.490547395823027e-06, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.490547395823027e-06, "step": 400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.4786292016506195, "val_fraction": 2.378234398782344e-05, "val_loss": 0.33185866475105286, "within_bucket_gap": -0.14677053689956665}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.987697096289202e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.987697096289202e-05, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.2332414388656616, "val_fraction": 4.756468797564688e-05, "val_loss": 0.39486800134181976, "within_bucket_gap": -0.8383734375238419}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.118807551704767e-06, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.118807551704767e-06, "step": 400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.889082908630371, "val_fraction": 9.512937595129375e-05, "val_loss": 5.793226003646851, "within_bucket_gap": -0.09585690498352051}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.01726843198061813, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.01726843198061813, "step": 500, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 5.945768242308025, "val_fraction": 0.05995528919330289, "val_loss": 5.657746414493257, "within_bucket_gap": -0.28802182781476837}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0048850270917732445, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0048850270917732445, "step": 500, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 6.006286742597047, "val_fraction": 0.019882039573820395, "val_loss": 5.76058624121561, "within_bucket_gap": -0.24570050138143706}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0024781652815896046, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0024781652815896046, "step": 500, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 5.894276058799342, "val_fraction": 0.011582001522070015, "val_loss": 5.680309131160164, "within_bucket_gap": -0.21396692763917802}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008588132598778363, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008588132598778363, "step": 500, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.703188390899123, "val_fraction": 0.007301179604261796, "val_loss": 5.585561746106474, "within_bucket_gap": -0.11762664479264906}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0016576262881583363, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0016576262881583363, "step": 500, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 5.92427150866124, "val_fraction": 0.004827815829528158, "val_loss": 5.580922394550493, "within_bucket_gap": -0.3433491141107474}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0017976083367072413, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0017976083367072413, "step": 500, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 6.150136528285683, "val_fraction": 0.003305745814307458, "val_loss": 5.606353540214704, "within_bucket_gap": -0.543782988070979}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0014123434548941795, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0014123434548941795, "step": 500, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 5.888312367345557, "val_fraction": 0.002877663622526636, "val_loss": 5.39751718063985, "within_bucket_gap": -0.4907951867057063}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.6818152326270762e-05, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.6818152326270762e-05, "step": 500, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.5068849836077005, "val_fraction": 0.002449581430745814, "val_loss": 5.500019258665807, "within_bucket_gap": -0.006865724941893525}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003078399414027351, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003078399414027351, "step": 500, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 5.494707812060107, "val_fraction": 0.0025684931506849314, "val_loss": 5.614560162579572, "within_bucket_gap": 0.11985235051946486}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009160097131257226, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009160097131257226, "step": 500, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 5.031183982382015, "val_fraction": 0.0014745053272450532, "val_loss": 5.652415860083796, "within_bucket_gap": 0.6212318777017805}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006955236159856139, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006955236159856139, "step": 500, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 5.668454522955908, "val_fraction": 0.0020214992389649923, "val_loss": 5.324391264073989, "within_bucket_gap": -0.3440632588819188}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005305181516521208, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005305181516521208, "step": 500, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 5.097793240017361, "val_fraction": 0.0016647640791476407, "val_loss": 4.77911856515067, "within_bucket_gap": -0.3186746748666911}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003745139088111107, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003745139088111107, "step": 500, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.44196959102855, "val_fraction": 0.0018074581430745813, "val_loss": 5.649174338892887, "within_bucket_gap": 0.20720474786433662}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008444403756731075, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008444403756731075, "step": 500, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.668221866383272, "val_fraction": 0.0015220700152207, "val_loss": 5.11342453956604, "within_bucket_gap": -0.5547973268172317}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.002188944434576348, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.002188944434576348, "step": 500, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 7.01761864626556, "val_fraction": 0.0014745053272450532, "val_loss": 5.533090652958039, "within_bucket_gap": -1.4845279933075206}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001195088641284264, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001195088641284264, "step": 500, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 6.8797065380007725, "val_fraction": 0.0010226407914764079, "val_loss": 5.71107660337936, "within_bucket_gap": -1.1686299346214124}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014849658063194687, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014849658063194687, "step": 500, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.642749786376953, "val_fraction": 0.0004280821917808219, "val_loss": 5.295861774020725, "within_bucket_gap": -0.3468880123562279}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004921039618189676, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004921039618189676, "step": 500, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.451291526236186, "val_fraction": 0.0007372526636225266, "val_loss": 4.7838080621534775, "within_bucket_gap": -0.6674834640827081}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.933478651318379e-05, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.933478651318379e-05, "step": 500, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.769974844796317, "val_fraction": 0.000594558599695586, "val_loss": 5.686997680664063, "within_bucket_gap": -0.08297716413225409}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015658928769880562, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015658928769880562, "step": 500, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.0934929480919475, "val_fraction": 0.0007372526636225266, "val_loss": 6.305888637419669, "within_bucket_gap": 0.2123956893277219}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005410520424165985, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005410520424165985, "step": 500, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.485710882371472, "val_fraction": 0.0006183409436834094, "val_loss": 6.3607168931227465, "within_bucket_gap": 0.8750060107512745}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00029550603898511447, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00029550603898511447, "step": 500, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.6995375497000555, "val_fraction": 0.0006659056316590563, "val_loss": 6.143303189958845, "within_bucket_gap": 0.44376564025878906}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006347628058000967, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006347628058000967, "step": 500, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 4.57067334651947, "val_fraction": 0.00047564687975646877, "val_loss": 5.905198669433593, "within_bucket_gap": 1.3345253229141232}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00023741238556314632, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00023741238556314632, "step": 500, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.218281936645508, "val_fraction": 0.0003091704718417047, "val_loss": 4.986183166503906, "within_bucket_gap": 0.7679012298583983}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00043140553629743083, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00043140553629743083, "step": 500, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.279766082763672, "val_fraction": 0.000594558599695586, "val_loss": 5.005355682373047, "within_bucket_gap": 0.7255895996093749}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.411692421911716e-05, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.411692421911716e-05, "step": 500, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.086927096048991, "val_fraction": 0.0004280821917808219, "val_loss": 5.9137899610731335, "within_bucket_gap": -0.1731371349758577}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000596246442710268, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000596246442710268, "step": 500, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 6.398838267606847, "val_fraction": 0.000594558599695586, "val_loss": 5.395999450683593, "within_bucket_gap": -1.002838816923254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016889815692075532, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016889815692075532, "step": 500, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.277785041115501, "val_fraction": 0.00045186453576864533, "val_loss": 5.6515655517578125, "within_bucket_gap": 0.37378051064231155}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0010842672736806169, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0010842672736806169, "step": 500, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.631242198328818, "val_fraction": 0.00040429984779299846, "val_loss": 5.313081629136029, "within_bucket_gap": 2.681839430807211}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003451844288293588, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003451844288293588, "step": 500, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.989404254489475, "val_fraction": 0.00033295281582952814, "val_loss": 6.026141030447824, "within_bucket_gap": 1.0367367759583486}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00045586375822986456, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00045586375822986456, "step": 500, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 7.040784299373627, "val_fraction": 0.0002616057838660578, "val_loss": 5.29822436246005, "within_bucket_gap": -1.742559936913577}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005545215459114276, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005545215459114276, "step": 500, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.387606938680013, "val_fraction": 0.0005232115677321156, "val_loss": 5.32776503129439, "within_bucket_gap": -1.0598419073856231}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.209563259180872e-05, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.209563259180872e-05, "step": 500, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.377493540445964, "val_fraction": 0.0002616057838660578, "val_loss": 6.7295332821932705, "within_bucket_gap": 0.3520397417473067}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.620599339461132e-05, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.620599339461132e-05, "step": 500, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.707947519090441, "val_fraction": 0.0002616057838660578, "val_loss": 5.8845720291137695, "within_bucket_gap": 0.1766245100233288}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00024185143694667143, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00024185143694667143, "step": 500, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.062195936838786, "val_fraction": 0.00014269406392694063, "val_loss": 6.75709080696106, "within_bucket_gap": 1.6948948701222735}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016990096087627515, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016990096087627515, "step": 500, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.939479509989421, "val_fraction": 0.0002616057838660578, "val_loss": 6.29002536426891, "within_bucket_gap": -0.6494541457205107}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00031206640199536875, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00031206640199536875, "step": 500, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.976426051213191, "val_fraction": 0.00028538812785388126, "val_loss": 4.882945378621419, "within_bucket_gap": -1.093480672591772}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006286002457414042, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006286002457414042, "step": 500, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.16108512878418, "val_fraction": 0.0003567351598173516, "val_loss": 5.398992919921875, "within_bucket_gap": -1.7620922088623043}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00027287041533120333, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00027287041533120333, "step": 500, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.212625026702881, "val_fraction": 0.0002616057838660578, "val_loss": 6.2556845925071025, "within_bucket_gap": 1.0430595658042217}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010136332323199175, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010136332323199175, "step": 500, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 3.731034278869629, "val_fraction": 9.512937595129375e-05, "val_loss": 4.796565532684326, "within_bucket_gap": 1.0655312538146973}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004030209698089182, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004030209698089182, "step": 500, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.920087432861328, "val_fraction": 0.00021404109589041095, "val_loss": 6.803001403808594, "within_bucket_gap": 1.882913970947266}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00019514246587733124, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00019514246587733124, "step": 500, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 5.596671661773286, "val_fraction": 0.00014269406392694063, "val_loss": 4.229113260904948, "within_bucket_gap": -1.3675584008683375}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003853524656600604, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003853524656600604, "step": 500, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.56417179107666, "val_fraction": 0.00023782343987823439, "val_loss": 5.184501838684082, "within_bucket_gap": 1.620330047607422}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004140761832577634, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004140761832577634, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.622978925704956, "val_fraction": 0.0002616057838660578, "val_loss": 4.20580395785245, "within_bucket_gap": 1.5828250321474941}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011242512577623958, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011242512577623958, "step": 500, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.025620778401692, "val_fraction": 0.00016647640791476407, "val_loss": 5.70094244820731, "within_bucket_gap": 0.6753216698056175}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00037344550796113965, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00037344550796113965, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.548839569091797, "val_fraction": 0.000380517503805175, "val_loss": 6.530254364013672, "within_bucket_gap": 0.981414794921875}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001016847496059672, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001016847496059672, "step": 500, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.727320398603167, "val_fraction": 7.134703196347032e-05, "val_loss": 7.152533849080403, "within_bucket_gap": 1.4252134504772362}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.43270197723917e-05, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.43270197723917e-05, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.608593583106995, "val_fraction": 4.756468797564688e-05, "val_loss": 3.4664223194122314, "within_bucket_gap": -1.1421712636947632}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.4160600350322195e-05, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.4160600350322195e-05, "step": 500, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 5.4282254016760625, "val_fraction": 4.756468797564688e-05, "val_loss": 4.710032939910889, "within_bucket_gap": -0.7181924617651738}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.300399654647592e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.300399654647592e-05, "step": 500, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.543032906272194, "val_fraction": 0.00016647640791476407, "val_loss": 4.984371185302734, "within_bucket_gap": -0.55866172096946}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003078068681684984, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003078068681684984, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.59584903717041, "val_fraction": 7.134703196347032e-05, "val_loss": 3.281627972920736, "within_bucket_gap": -4.314221064249674}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010473407305992538, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010473407305992538, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.208468019962311, "val_fraction": 4.756468797564688e-05, "val_loss": 2.0065388679504395, "within_bucket_gap": -2.2019291520118713}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013744373422883122, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013744373422883122, "step": 500, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.603591237749372, "val_fraction": 0.00014269406392694063, "val_loss": 5.640385548273723, "within_bucket_gap": -0.9632056894756493}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.577795069330177e-06, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.577795069330177e-06, "step": 500, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 3.7473392486572266, "val_fraction": 2.378234398782344e-05, "val_loss": 4.150066375732422, "within_bucket_gap": 0.4027271270751953}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.924745442841874e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.924745442841874e-05, "step": 500, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.750205039978027, "val_fraction": 0.00016647640791476407, "val_loss": 5.214108330862863, "within_bucket_gap": -0.5360967091151645}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.976758350883266e-05, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.976758350883266e-05, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.654918034871419, "val_fraction": 0.00014269406392694063, "val_loss": 5.354089260101318, "within_bucket_gap": 0.6991712252298994}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003723431452314966, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003723431452314966, "step": 500, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.3530182361602785, "val_fraction": 0.00011891171993911719, "val_loss": 2.2217613220214845, "within_bucket_gap": -3.131256914138794}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00024097547774264074, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00024097547774264074, "step": 500, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.175568580627441, "val_fraction": 9.512937595129375e-05, "val_loss": 6.708702802658081, "within_bucket_gap": 2.5331342220306396}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004966076839581048, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004966076839581048, "step": 500, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.712889591852824, "val_fraction": 0.00023782343987823439, "val_loss": 6.8010255813598635, "within_bucket_gap": 2.0881359895070393}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00026805845161551213, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00026805845161551213, "step": 500, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.8372087478637695, "val_fraction": 0.00021404109589041095, "val_loss": 4.584839661916097, "within_bucket_gap": -1.2523690859476728}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.185194771765189e-06, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.185194771765189e-06, "step": 500, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.404840278625488, "val_fraction": 0.00011891171993911719, "val_loss": 6.344415664672852, "within_bucket_gap": -0.06042461395263654}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.3071432200741004e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.3071432200741004e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.795600891113281, "val_fraction": 7.134703196347032e-05, "val_loss": 6.399290084838867, "within_bucket_gap": 0.6036891937255859}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.9105424382915296e-05, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.9105424382915296e-05, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.684368014335632, "val_fraction": 0.00016647640791476407, "val_loss": 5.979337283543178, "within_bucket_gap": 0.29496926920754607}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.486167547728191e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.486167547728191e-05, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.343777179718018, "val_fraction": 0.0001902587519025875, "val_loss": 7.632130146026611, "within_bucket_gap": 0.28835296630859375}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000116827088313197, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000116827088313197, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.429575443267822, "val_fraction": 0.0002616057838660578, "val_loss": 5.982998587868431, "within_bucket_gap": -0.4465768553993916}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010598879562665338, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010598879562665338, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.608805179595947, "val_fraction": 4.756468797564688e-05, "val_loss": 5.3804967403411865, "within_bucket_gap": -2.2283084392547607}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013379342116903134, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013379342116903134, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.074690341949463, "val_fraction": 7.134703196347032e-05, "val_loss": 6.199441750844319, "within_bucket_gap": -1.8752485911051435}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.745709238531382e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.745709238531382e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.843961715698242, "val_fraction": 4.756468797564688e-05, "val_loss": 7.005263805389404, "within_bucket_gap": -1.838697910308838}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003750889676890961, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003750889676890961, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.846832036972046, "val_fraction": 9.512937595129375e-05, "val_loss": 7.789767265319824, "within_bucket_gap": 3.9429352283477783}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010983262613665204, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010983262613665204, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.916691303253174, "val_fraction": 9.512937595129375e-05, "val_loss": 4.07125186920166, "within_bucket_gap": 1.1545605659484863}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014016050363421619, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014016050363421619, "step": 500, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.599891662597656, "val_fraction": 7.134703196347032e-05, "val_loss": 3.635402043660482, "within_bucket_gap": -1.9644896189371743}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000248171150956524, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000248171150956524, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.5991384983062744, "val_fraction": 0.0001902587519025875, "val_loss": 3.9035260677337646, "within_bucket_gap": 1.3043875694274902}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.925806236412245e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.925806236412245e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.338642597198486, "val_fraction": 2.378234398782344e-05, "val_loss": 2.8469595909118652, "within_bucket_gap": -2.491683006286621}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00025023856516388814, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00025023856516388814, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5812209844589233, "val_fraction": 7.134703196347032e-05, "val_loss": 5.0885647137959795, "within_bucket_gap": 3.507343729337056}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010462470999047096, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010462470999047096, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.4633147716522217, "val_fraction": 0.00011891171993911719, "val_loss": 4.343166732788086, "within_bucket_gap": 0.8798519611358646}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001822830244467864, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001822830244467864, "step": 500, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.970109144846599, "val_fraction": 7.134703196347032e-05, "val_loss": 7.524988015492757, "within_bucket_gap": 2.5548788706461583}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001346130530765249, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001346130530765249, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.445675373077393, "val_fraction": 9.512937595129375e-05, "val_loss": 6.860727787017822, "within_bucket_gap": 1.4150524139404297}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.101971626281738, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.0171109116785058e-06, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.0171109116785058e-06, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.92356538772583, "val_fraction": 4.756468797564688e-05, "val_loss": 5.860133647918701, "within_bucket_gap": -0.0634317398071289}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020129207516006862, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020129207516006862, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.83842134475708, "val_fraction": 4.756468797564688e-05, "val_loss": 2.606456756591797, "within_bucket_gap": -4.231964588165283}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.942353248596191, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003274948629614425, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003274948629614425, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.1220479011535645, "val_fraction": 0.0001902587519025875, "val_loss": 3.4007349014282227, "within_bucket_gap": -1.7213129997253418}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00026560335489531447, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00026560335489531447, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.566286087036133, "val_fraction": 7.134703196347032e-05, "val_loss": 6.843589464823405, "within_bucket_gap": -3.7226966222127276}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013850638401925654, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013850638401925654, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.225881576538086, "val_fraction": 4.756468797564688e-05, "val_loss": 4.313923358917236, "within_bucket_gap": -2.9119582176208496}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.843860873469596e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.843860873469596e-05, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.082461039225261, "val_fraction": 9.512937595129375e-05, "val_loss": 4.907007694244385, "within_bucket_gap": 0.824546655019124}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011487311968520352, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011487311968520352, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.393494606018066, "val_fraction": 9.512937595129375e-05, "val_loss": 5.185948371887207, "within_bucket_gap": -1.2075462341308594}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.5540005953888916e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.5540005953888916e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.7821927070617676, "val_fraction": 2.378234398782344e-05, "val_loss": 3.1287665367126465, "within_bucket_gap": -0.6534261703491211}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.218852613977464e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.218852613977464e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1308209896087646, "val_fraction": 2.378234398782344e-05, "val_loss": 3.643324136734009, "within_bucket_gap": 0.5125031471252441}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018489798545111622, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018489798545111622, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.669156074523926, "val_fraction": 4.756468797564688e-05, "val_loss": 8.556451320648193, "within_bucket_gap": 3.8872952461242676}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017200698976828807, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017200698976828807, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.4730794429779053, "val_fraction": 2.378234398782344e-05, "val_loss": 10.705629348754883, "within_bucket_gap": 7.2325499057769775}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.3427783671760847e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.3427783671760847e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.724679470062256, "val_fraction": 4.756468797564688e-05, "val_loss": 6.232133746147156, "within_bucket_gap": -0.4925457239151001}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014133611056086135, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014133611056086135, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.6949299573898315, "val_fraction": 2.378234398782344e-05, "val_loss": 7.63783073425293, "within_bucket_gap": 5.942900776863098}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014920596209109464, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014920596209109464, "step": 500, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.582422256469727, "val_fraction": 2.378234398782344e-05, "val_loss": 10.856234550476074, "within_bucket_gap": 6.273812294006348}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.12986436849134e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.12986436849134e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.710572212934494, "val_fraction": 4.756468797564688e-05, "val_loss": 4.789074897766113, "within_bucket_gap": 1.0785026848316193}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004807886762524667, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004807886762524667, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.409538269042969, "val_fraction": 7.134703196347032e-05, "val_loss": 4.6708041826883955, "within_bucket_gap": -6.738734086354573}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.729051351547241, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.8239827156066895, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.643875503467461e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.643875503467461e-05, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.8923444747924805, "val_fraction": 2.378234398782344e-05, "val_loss": 7.6859612464904785, "within_bucket_gap": 2.793616771697998}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.600205421447754, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.064889907836914, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.129943103187886e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.129943103187886e-05, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.9048120975494385, "val_fraction": 2.378234398782344e-05, "val_loss": 7.902810573577881, "within_bucket_gap": 2.9979984760284424}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.8750667572021484, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.429352164268494, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.383187294006348, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003182290240986162, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003182290240986162, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5044802129268646, "val_fraction": 4.756468797564688e-05, "val_loss": 8.194927215576172, "within_bucket_gap": 6.690447002649307}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.809482579543347e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.809482579543347e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.64361572265625, "val_fraction": 2.378234398782344e-05, "val_loss": 4.939404487609863, "within_bucket_gap": -3.7042112350463867}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000216818605996158, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000216818605996158, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.2811921834945679, "val_fraction": 7.134703196347032e-05, "val_loss": 4.320121765136719, "within_bucket_gap": 3.038929581642151}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.700902938842773, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016579548495182346, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016579548495182346, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.927768230438232, "val_fraction": 0.00014269406392694063, "val_loss": 3.7658734718958535, "within_bucket_gap": -1.161894758542379}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.483591091687276e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.483591091687276e-05, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.153339862823486, "val_fraction": 4.756468797564688e-05, "val_loss": 7.726690053939819, "within_bucket_gap": 1.573350191116333}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.93041956461002e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.93041956461002e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1077425479888916, "val_fraction": 2.378234398782344e-05, "val_loss": 6.862805366516113, "within_bucket_gap": 3.7550628185272217}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001724757866500175, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001724757866500175, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.674875259399414, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4226133823394775, "within_bucket_gap": -7.2522618770599365}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.460256022768296e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.460256022768296e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.539910316467285, "val_fraction": 2.378234398782344e-05, "val_loss": 8.994878768920898, "within_bucket_gap": 1.4549684524536133}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.425747871398926, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.894270420074463, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0363723589405076e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0363723589405076e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.047560691833496, "val_fraction": 4.756468797564688e-05, "val_loss": 3.8296737670898438, "within_bucket_gap": -0.21788692474365234}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.707311153411865, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.5578844547271729, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.1872682571411133, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.7717695236206055, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.895325825094632e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.895325825094632e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.7785234451293945, "val_fraction": 2.378234398782344e-05, "val_loss": 5.0983500480651855, "within_bucket_gap": 3.319826602935791}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1850337982177734, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.467334747314453, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.10872182995080948, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.289946556091309, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.677502374489376e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.677502374489376e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.0946502685546875, "val_fraction": 4.756468797564688e-05, "val_loss": 5.480532169342041, "within_bucket_gap": -1.6141180992126465}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.147661209106445, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.2731218338012695, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.624236583709717, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.289769933648305e-06, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.289769933648305e-06, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.8205935955047607, "val_fraction": 2.378234398782344e-05, "val_loss": 4.085065841674805, "within_bucket_gap": 0.26447224617004395}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.321248291834304e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.321248291834304e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.1290693283081055, "val_fraction": 2.378234398782344e-05, "val_loss": 7.946067810058594, "within_bucket_gap": 1.8169984817504883}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.9925064488995925e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.9925064488995925e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.806020736694336, "val_fraction": 2.378234398782344e-05, "val_loss": 4.127251625061035, "within_bucket_gap": -1.6787691116333008}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.089811325073242, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.575811862945557, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.575113296508789, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00036790456673870345, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00036790456673870345, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.2422080039978027, "val_fraction": 4.756468797564688e-05, "val_loss": 9.977033615112305, "within_bucket_gap": 7.734825611114502}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.15486973524093628, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.761131763458252, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.557084560394287, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.382248878479004, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.962882995605469, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.4281228423753646e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.4281228423753646e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.100946009159088, "val_fraction": 2.378234398782344e-05, "val_loss": 2.6594889163970947, "within_bucket_gap": -1.4414570927619934}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.754544258117676, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.6175448894500732, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.3843655348549874e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.3843655348549874e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.9002089500427246, "val_fraction": 2.378234398782344e-05, "val_loss": 1.8976309299468994, "within_bucket_gap": -1.0025780200958252}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.913626871152555e-06, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.913626871152555e-06, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7506113052368164, "val_fraction": 2.378234398782344e-05, "val_loss": 3.167459487915039, "within_bucket_gap": 0.41684818267822266}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.045769453048706, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.793178558349609, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.817558526992798, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.791425704956055, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.458054542541504, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.952934741973877, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.8543970584869385, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.6528426965803497e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.6528426965803497e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.049131155014038, "val_fraction": 2.378234398782344e-05, "val_loss": 2.3541438579559326, "within_bucket_gap": -0.6949872970581055}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.644381523132324, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.44649474624448e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.44649474624448e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.671867847442627, "val_fraction": 2.378234398782344e-05, "val_loss": 3.699805736541748, "within_bucket_gap": -3.972062110900879}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013593293042245822, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013593293042245822, "step": 500, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.941116650899251, "val_fraction": 2.378234398782344e-05, "val_loss": 2.2254087924957275, "within_bucket_gap": -5.7157078584035235}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.4235701963781764e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.4235701963781764e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.231045246124268, "val_fraction": 2.378234398782344e-05, "val_loss": 7.511548042297363, "within_bucket_gap": 2.2805027961730957}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.593810558319092, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.806816918120536e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.806816918120536e-05, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.255780458450317, "val_fraction": 2.378234398782344e-05, "val_loss": 8.27695083618164, "within_bucket_gap": 2.0211703777313232}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.816438674926758, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.2686847746372223, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.2514195442199707, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.562807083129883, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.820601463317871, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4078450202941895, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.447539329528809, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.2821490168571472, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.329977989196777, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.342071056365967, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.14070158951903e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.14070158951903e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.6667792797088623, "val_fraction": 2.378234398782344e-05, "val_loss": 4.828341484069824, "within_bucket_gap": 2.161562204360962}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.556097030639648, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.489554405212402, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.804158892306744e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.804158892306744e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.995170533657074, "val_fraction": 2.378234398782344e-05, "val_loss": 3.015223264694214, "within_bucket_gap": 2.02005273103714}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014378185436424417, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014378185436424417, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.642470121383667, "val_fraction": 2.378234398782344e-05, "val_loss": 7.688209533691406, "within_bucket_gap": 6.045739412307739}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.2031450271606445, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.678438663482666, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.016513574868440628, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015456218391520792, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015456218391520792, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.876048445701599, "val_fraction": 2.378234398782344e-05, "val_loss": 10.375079154968262, "within_bucket_gap": 6.499030709266663}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001314456264177958, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001314456264177958, "step": 500, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.088986873626709, "val_fraction": 4.756468797564688e-05, "val_loss": 2.3254740238189697, "within_bucket_gap": -2.7635128498077393}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.11134672164917, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.888692378997803, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.116929054260254, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.20898762345314026, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.05139026790857315, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.01632019877433777, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.321072578430176, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.030070779845118523, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.726369895528622e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.726369895528622e-05, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.756336212158203, "val_fraction": 2.378234398782344e-05, "val_loss": 6.323200225830078, "within_bucket_gap": 1.566864013671875}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.618939399719238, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.476749420166016, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.3222654252291814e-06, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.3222654252291814e-06, "step": 500, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.3690125346183777, "val_fraction": 2.378234398782344e-05, "val_loss": 0.4246111512184143, "within_bucket_gap": 0.05559861660003662}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.876474096123667e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.876474096123667e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.839194416999817, "val_fraction": 4.756468797564688e-05, "val_loss": 0.393484503030777, "within_bucket_gap": -1.44570991396904}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.648867383213711e-05, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.648867383213711e-05, "step": 500, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.511363983154297, "val_fraction": 9.512937595129375e-05, "val_loss": 5.232915043830872, "within_bucket_gap": -0.2784489393234253}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.016122572768458025, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.016122572768458025, "step": 600, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 6.069805844315003, "val_fraction": 0.05995528919330289, "val_loss": 5.80089591184054, "within_bucket_gap": -0.2689099324744637}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.003879483250541038, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.003879483250541038, "step": 600, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 6.149502840909091, "val_fraction": 0.019882039573820395, "val_loss": 5.954377826891448, "within_bucket_gap": -0.19512501401764304}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0031887508491040004, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0031887508491040004, "step": 600, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 6.0835595703125, "val_fraction": 0.011582001522070015, "val_loss": 5.808240071948793, "within_bucket_gap": -0.2753194983637064}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0018526790520801634, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0018526790520801634, "step": 600, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.904686215049342, "val_fraction": 0.007301179604261796, "val_loss": 5.650935567551405, "within_bucket_gap": -0.2537506474979372}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0017686793470291112, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0017686793470291112, "step": 600, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 6.135476317081152, "val_fraction": 0.004827815829528158, "val_loss": 5.769124449180265, "within_bucket_gap": -0.36635186790088703}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0021467960481655916, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0021467960481655916, "step": 600, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 6.421668086491578, "val_fraction": 0.003305745814307458, "val_loss": 5.772254559633543, "within_bucket_gap": -0.6494135268580346}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0015932312518170252, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0015932312518170252, "step": 600, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 6.121061468400018, "val_fraction": 0.002877663622526636, "val_loss": 5.567407024793388, "within_bucket_gap": -0.5536544436066304}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014830275213337317, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014830275213337317, "step": 600, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.605356216430664, "val_fraction": 0.002449581430745814, "val_loss": 5.66589829528216, "within_bucket_gap": 0.06054207885149587}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008104931661370535, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008104931661370535, "step": 600, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 5.841487850154842, "val_fraction": 0.0025684931506849314, "val_loss": 5.525935844138816, "within_bucket_gap": -0.3155520060160262}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005524425979385331, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005524425979385331, "step": 600, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 5.585963657924107, "val_fraction": 0.0014745053272450532, "val_loss": 5.960626663700227, "within_bucket_gap": 0.37466300577612}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0010147817225181924, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0010147817225181924, "step": 600, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 5.822867197533176, "val_fraction": 0.0020214992389649923, "val_loss": 5.320872587316177, "within_bucket_gap": -0.5019946102169994}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006732559639569282, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006732559639569282, "step": 600, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 5.2644287109375, "val_fraction": 0.0016647640791476407, "val_loss": 4.860013471330915, "within_bucket_gap": -0.40441523960658454}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00036325935220662763, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00036325935220662763, "step": 600, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.935745127060834, "val_fraction": 0.0018074581430745813, "val_loss": 5.734767110724198, "within_bucket_gap": -0.20097801633663526}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014518030712924536, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014518030712924536, "step": 600, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.450312670539407, "val_fraction": 0.0015220700152207, "val_loss": 5.354929208755493, "within_bucket_gap": -0.09538346178391421}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0027474986049811326, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0027474986049811326, "step": 600, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 7.427653838984699, "val_fraction": 0.0014745053272450532, "val_loss": 5.564318010883946, "within_bucket_gap": -1.8633358281007526}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008653629134597665, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008653629134597665, "step": 600, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 6.799838221350382, "val_fraction": 0.0010226407914764079, "val_loss": 5.953634040300236, "within_bucket_gap": -0.8462041810501457}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013601774415508504, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013601774415508504, "step": 600, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.7876454228940215, "val_fraction": 0.0004280821917808219, "val_loss": 5.469907972547743, "within_bucket_gap": -0.31773745034627865}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004347650082608692, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004347650082608692, "step": 600, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.438586723513719, "val_fraction": 0.0007372526636225266, "val_loss": 4.84887707617975, "within_bucket_gap": -0.5897096473339687}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003221340025474413, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003221340025474413, "step": 600, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.805633817400251, "val_fraction": 0.000594558599695586, "val_loss": 6.347437438964843, "within_bucket_gap": 0.5418036215645925}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00046995997833655727, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00046995997833655727, "step": 600, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.792895977313702, "val_fraction": 0.0007372526636225266, "val_loss": 6.4303436279296875, "within_bucket_gap": 0.6374476506159859}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004138917677613323, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004138917677613323, "step": 600, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.351386408652028, "val_fraction": 0.0006183409436834094, "val_loss": 6.020744910606971, "within_bucket_gap": 0.6693585019549424}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005580875427211256, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005580875427211256, "step": 600, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.504665510995047, "val_fraction": 0.0006659056316590563, "val_loss": 6.342753546578543, "within_bucket_gap": 0.8380880355834961}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008100185407215056, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008100185407215056, "step": 600, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 4.536706656217575, "val_fraction": 0.00047564687975646877, "val_loss": 6.239689636230469, "within_bucket_gap": 1.7029829800128935}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00035378764027692777, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00035378764027692777, "step": 600, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 3.9779943084716796, "val_fraction": 0.0003091704718417047, "val_loss": 5.122306823730469, "within_bucket_gap": 1.1443125152587892}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006101789760133396, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006101789760133396, "step": 600, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.573437554495675, "val_fraction": 0.000594558599695586, "val_loss": 5.5997097778320315, "within_bucket_gap": 1.0262722233363561}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013949754938505935, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013949754938505935, "step": 600, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.930412928263347, "val_fraction": 0.0004280821917808219, "val_loss": 5.604546652899848, "within_bucket_gap": -0.32586627536349866}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00044185478535655403, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00044185478535655403, "step": 600, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 6.186586641797833, "val_fraction": 0.000594558599695586, "val_loss": 5.443422241210937, "within_bucket_gap": -0.7431644005868954}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00045070849267282113, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00045070849267282113, "step": 600, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.025884454900568, "val_fraction": 0.00045186453576864533, "val_loss": 6.023326070685136, "within_bucket_gap": 0.9974416157845676}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0010929728792828958, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0010929728792828958, "step": 600, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.9278865937263734, "val_fraction": 0.00040429984779299846, "val_loss": 5.63125857184915, "within_bucket_gap": 2.7033719781227767}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.6730406590369494e-05, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.6730406590369494e-05, "step": 600, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.86860720316569, "val_fraction": 0.00033295281582952814, "val_loss": 5.698221479143415, "within_bucket_gap": -0.17038572402227548}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007320990703739169, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007320990703739169, "step": 600, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 7.454815030097961, "val_fraction": 0.0002616057838660578, "val_loss": 4.656333056363192, "within_bucket_gap": -2.798481973734769}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008374373360009198, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008374373360009198, "step": 600, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.428276697794597, "val_fraction": 0.0005232115677321156, "val_loss": 4.827705556696111, "within_bucket_gap": -1.6005711410984853}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00024579443842671596, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00024579443842671596, "step": 600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.382660388946533, "val_fraction": 0.0002616057838660578, "val_loss": 5.4430999755859375, "within_bucket_gap": -0.9395604133605957}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001548078309939273, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001548078309939273, "step": 600, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.621031125386556, "val_fraction": 0.0002616057838660578, "val_loss": 6.212791096080434, "within_bucket_gap": 0.5917599706938779}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00022694466230168924, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00022694466230168924, "step": 600, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 4.94147523244222, "val_fraction": 0.00014269406392694063, "val_loss": 6.531903425852458, "within_bucket_gap": 1.5904281934102382}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.801591652773856e-06, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.801591652773856e-06, "step": 600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.791068871815999, "val_fraction": 0.0002616057838660578, "val_loss": 6.77653711492365, "within_bucket_gap": -0.014531756892348646}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016655117018609967, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016655117018609967, "step": 600, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.660508816058819, "val_fraction": 0.00028538812785388126, "val_loss": 5.076913515726726, "within_bucket_gap": -0.5835953003320933}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005715330396431584, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005715330396431584, "step": 600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.347235043843587, "val_fraction": 0.0003567351598173516, "val_loss": 5.745113627115885, "within_bucket_gap": -1.6021214167277016}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00026443638833629986, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00026443638833629986, "step": 600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.446122010548909, "val_fraction": 0.0002616057838660578, "val_loss": 6.456942124800249, "within_bucket_gap": 1.0108201142513398}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.587359318417167e-05, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.587359318417167e-05, "step": 600, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.500001560557973, "val_fraction": 9.512937595129375e-05, "val_loss": 5.297584772109985, "within_bucket_gap": 0.7975832115520127}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002884271391631988, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002884271391631988, "step": 600, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 5.958547210693359, "val_fraction": 0.00021404109589041095, "val_loss": 7.306078804863824, "within_bucket_gap": 1.3475315941704649}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022515778894887614, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022515778894887614, "step": 600, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 5.510289163300485, "val_fraction": 0.00014269406392694063, "val_loss": 3.932383378346761, "within_bucket_gap": -1.5779057849537241}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00037919478679278735, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00037919478679278735, "step": 600, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.9862536702837263, "val_fraction": 0.00023782343987823439, "val_loss": 5.580691909790039, "within_bucket_gap": 1.5944382395063124}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004901381682950431, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004901381682950431, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.0944619178771973, "val_fraction": 0.0002616057838660578, "val_loss": 4.968037345192649, "within_bucket_gap": 1.8735754273154521}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010872818120094849, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010872818120094849, "step": 600, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.605730268690321, "val_fraction": 0.00016647640791476407, "val_loss": 6.258844920567104, "within_bucket_gap": 0.6531146518767832}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001671743356655359, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001671743356655359, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.409887313842773, "val_fraction": 0.000380517503805175, "val_loss": 5.970553159713745, "within_bucket_gap": -0.4393341541290283}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012981869599486875, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012981869599486875, "step": 600, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.9093843187604636, "val_fraction": 7.134703196347032e-05, "val_loss": 7.728923161824544, "within_bucket_gap": 1.8195388430640804}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012133211532502776, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012133211532502776, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.02002215385437, "val_fraction": 4.756468797564688e-05, "val_loss": 3.4691357612609863, "within_bucket_gap": -2.550886392593384}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.254820473744724e-05, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.254820473744724e-05, "step": 600, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 3.897808855230158, "val_fraction": 4.756468797564688e-05, "val_loss": 5.002582311630249, "within_bucket_gap": 1.1047734564000908}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011985755088798757, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011985755088798757, "step": 600, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.971379713578657, "val_fraction": 0.00016647640791476407, "val_loss": 5.251412527901786, "within_bucket_gap": -0.7199671856768717}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017764336533016624, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017764336533016624, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.888510704040527, "val_fraction": 7.134703196347032e-05, "val_loss": 4.398661295572917, "within_bucket_gap": -2.4898494084676104}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.091960704399207e-05, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.091960704399207e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.151474893093109, "val_fraction": 4.756468797564688e-05, "val_loss": 2.8707010746002197, "within_bucket_gap": -1.2807738184928894}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001530207525092276, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001530207525092276, "step": 600, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.763624208314078, "val_fraction": 0.00014269406392694063, "val_loss": 5.691254774729411, "within_bucket_gap": -1.072369433584667}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.507861490481883e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.507861490481883e-05, "step": 600, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 3.208786964416504, "val_fraction": 2.378234398782344e-05, "val_loss": 4.683772563934326, "within_bucket_gap": 1.4749855995178223}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.5408006853890505e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.5408006853890505e-05, "step": 600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.385769271850586, "val_fraction": 0.00016647640791476407, "val_loss": 5.113009861537388, "within_bucket_gap": -0.2727594103131983}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00023186277127519955, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00023186277127519955, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.339042663574219, "val_fraction": 0.00014269406392694063, "val_loss": 6.963936964670817, "within_bucket_gap": 1.6248943010965986}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00040219141286620627, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00040219141286620627, "step": 600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.192648601531983, "val_fraction": 0.00011891171993911719, "val_loss": 2.810379695892334, "within_bucket_gap": -3.3822689056396484}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.46953125849162e-05, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.46953125849162e-05, "step": 600, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.104550242424011, "val_fraction": 9.512937595129375e-05, "val_loss": 6.46926736831665, "within_bucket_gap": 0.36471712589263916}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005529924456024461, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005529924456024461, "step": 600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.756833791732788, "val_fraction": 0.00023782343987823439, "val_loss": 7.0820564270019535, "within_bucket_gap": 2.3252226352691654}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00029011709867909983, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00029011709867909983, "step": 600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.153458786010742, "val_fraction": 0.00021404109589041095, "val_loss": 4.798031700981988, "within_bucket_gap": -1.3554270850287544}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.2172791772781653e-06, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.2172791772781653e-06, "step": 600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.8917749404907225, "val_fraction": 0.00011891171993911719, "val_loss": 5.910421371459961, "within_bucket_gap": 0.01864643096923846}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.421995027968872e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.421995027968872e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.022310256958008, "val_fraction": 7.134703196347032e-05, "val_loss": 6.782257080078125, "within_bucket_gap": 0.7599468231201172}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020827507512300887, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020827507512300887, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.951164126396179, "val_fraction": 0.00016647640791476407, "val_loss": 5.700085503714425, "within_bucket_gap": -1.2510786226817538}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.282102375209548e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.282102375209548e-05, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.869027296702067, "val_fraction": 0.0001902587519025875, "val_loss": 7.5388399958610535, "within_bucket_gap": -0.3301873008410139}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.479224537787002e-05, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.479224537787002e-05, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.9426493644714355, "val_fraction": 0.0002616057838660578, "val_loss": 5.618527152321556, "within_bucket_gap": -0.3241222121498799}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.200655638355098e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.200655638355098e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.0103243589401245, "val_fraction": 4.756468797564688e-05, "val_loss": 7.9446702003479, "within_bucket_gap": 1.9343458414077759}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00019195126979136936, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00019195126979136936, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.325824737548828, "val_fraction": 7.134703196347032e-05, "val_loss": 5.635435740152995, "within_bucket_gap": -2.690388997395833}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001137922071430781, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001137922071430781, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.716609954833984, "val_fraction": 4.756468797564688e-05, "val_loss": 6.32424259185791, "within_bucket_gap": -2.392367362976074}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00026116906870808594, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00026116906870808594, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.370977401733398, "val_fraction": 9.512937595129375e-05, "val_loss": 8.116386651992798, "within_bucket_gap": 2.7454092502593994}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019738109870225508, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019738109870225508, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.5595827102661133, "val_fraction": 9.512937595129375e-05, "val_loss": 4.634452819824219, "within_bucket_gap": 2.0748701095581055}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00021145426791552538, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00021145426791552538, "step": 600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.132880020141601, "val_fraction": 7.134703196347032e-05, "val_loss": 3.1691370010375977, "within_bucket_gap": -2.9637430191040037}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005390307493224354, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005390307493224354, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4438605308532715, "val_fraction": 0.0001902587519025875, "val_loss": 4.277006149291992, "within_bucket_gap": 2.8331456184387207}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.9910055594901515e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.9910055594901515e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.446911573410034, "val_fraction": 2.378234398782344e-05, "val_loss": 4.125049591064453, "within_bucket_gap": 1.678138017654419}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015636372466428455, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015636372466428455, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.691072940826416, "val_fraction": 7.134703196347032e-05, "val_loss": 4.882666905721028, "within_bucket_gap": 2.1915939648946123}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002678297431728429, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002678297431728429, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.637593150138855, "val_fraction": 0.00011891171993911719, "val_loss": 4.889934158325195, "within_bucket_gap": 2.25234100818634}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001535389793517927, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001535389793517927, "step": 600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.568990071614583, "val_fraction": 7.134703196347032e-05, "val_loss": 7.72099240620931, "within_bucket_gap": 2.1520023345947266}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016272528440077735, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016272528440077735, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.579641103744507, "val_fraction": 9.512937595129375e-05, "val_loss": 7.2902092933654785, "within_bucket_gap": 1.7105681896209717}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.162806987762451, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.902028481891348e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.902028481891348e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.179097890853882, "val_fraction": 4.756468797564688e-05, "val_loss": 7.630180358886719, "within_bucket_gap": 1.451082468032837}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017943012115617863, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017943012115617863, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.180602073669434, "val_fraction": 4.756468797564688e-05, "val_loss": 3.4082632064819336, "within_bucket_gap": -3.7723388671875}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.522624492645264, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.075424493357108e-05, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.075424493357108e-05, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.256091117858887, "val_fraction": 0.0001902587519025875, "val_loss": 4.041886806488037, "within_bucket_gap": -0.2142043113708496}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00044056985249439515, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00044056985249439515, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.929985046386719, "val_fraction": 7.134703196347032e-05, "val_loss": 6.754957993825276, "within_bucket_gap": -6.175027052561442}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018475353831207006, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018475353831207006, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.286019325256348, "val_fraction": 4.756468797564688e-05, "val_loss": 4.401760935783386, "within_bucket_gap": -3.8842583894729614}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.063818508208188e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.063818508208188e-05, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.169501622517903, "val_fraction": 9.512937595129375e-05, "val_loss": 4.532073020935059, "within_bucket_gap": -0.6374286015828448}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00021840339381945186, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00021840339381945186, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.099206924438477, "val_fraction": 9.512937595129375e-05, "val_loss": 3.8033504486083984, "within_bucket_gap": -2.295856475830078}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.574825549053093e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.574825549053093e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4329049587249756, "val_fraction": 2.378234398782344e-05, "val_loss": 4.61796760559082, "within_bucket_gap": 3.1850626468658447}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.282594263462897e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.282594263462897e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1538729667663574, "val_fraction": 2.378234398782344e-05, "val_loss": 4.534138202667236, "within_bucket_gap": 1.380265235900879}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002000433449996782, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002000433449996782, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.364529927571614, "val_fraction": 4.756468797564688e-05, "val_loss": 8.570241212844849, "within_bucket_gap": 4.205711285273234}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001706324937318195, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001706324937318195, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7984657287597656, "val_fraction": 2.378234398782344e-05, "val_loss": 9.973220825195312, "within_bucket_gap": 7.174755096435547}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.203500169960149e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.203500169960149e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.683245658874512, "val_fraction": 4.756468797564688e-05, "val_loss": 7.00974178314209, "within_bucket_gap": -0.6735038757324219}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014449106004865928, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014449106004865928, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.2533831596374512, "val_fraction": 2.378234398782344e-05, "val_loss": 7.328943252563477, "within_bucket_gap": 6.075560092926025}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001225058890913175, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001225058890913175, "step": 600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.149591636657715, "val_fraction": 2.378234398782344e-05, "val_loss": 10.300719261169434, "within_bucket_gap": 5.151127624511719}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.06849203452672e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.06849203452672e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.9686166644096375, "val_fraction": 4.756468797564688e-05, "val_loss": 4.875176429748535, "within_bucket_gap": 1.9065597653388977}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004785997591788003, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004785997591788003, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.176162719726562, "val_fraction": 7.134703196347032e-05, "val_loss": 5.468108495076497, "within_bucket_gap": -6.708054224650065}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.833500067392985, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.20033073425293, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.279372099690604e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.279372099690604e-05, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.81204628944397, "val_fraction": 2.378234398782344e-05, "val_loss": 8.713836669921875, "within_bucket_gap": 3.9017903804779053}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.821950912475586, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.009940147399902, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.1271776470782364e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.1271776470782364e-05, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.819235801696777, "val_fraction": 2.378234398782344e-05, "val_loss": 6.134151458740234, "within_bucket_gap": 1.314915657043457}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.354515552520752, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.821383953094482, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.377986907958984, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003677416997543526, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003677416997543526, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.010505624115467, "val_fraction": 4.756468797564688e-05, "val_loss": 8.741907119750977, "within_bucket_gap": 7.7314014956355095}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.713343409461336e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.713343409461336e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.8718843460083, "val_fraction": 2.378234398782344e-05, "val_loss": 5.628577709197998, "within_bucket_gap": -3.2433066368103027}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.794153827510467e-05, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.794153827510467e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.0889968872070312, "val_fraction": 7.134703196347032e-05, "val_loss": 3.0412654876708984, "within_bucket_gap": 0.9522686004638672}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.266161918640137, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00021977093824875768, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00021977093824875768, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.996205568313599, "val_fraction": 0.00014269406392694063, "val_loss": 3.4560508330663047, "within_bucket_gap": -1.540154735247294}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.6965231262808928e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.6965231262808928e-05, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.159789244333903, "val_fraction": 4.756468797564688e-05, "val_loss": 6.726706266403198, "within_bucket_gap": 0.566917022069295}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.222434035719258e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.222434035719258e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.2690041065216064, "val_fraction": 2.378234398782344e-05, "val_loss": 5.88541316986084, "within_bucket_gap": 2.6164090633392334}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00019906815623402412, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00019906815623402412, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.519744873046875, "val_fraction": 2.378234398782344e-05, "val_loss": 2.149327039718628, "within_bucket_gap": -8.370417833328247}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.383220256009965e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.383220256009965e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.2785139083862305, "val_fraction": 2.378234398782344e-05, "val_loss": 9.5420503616333, "within_bucket_gap": 2.2635364532470703}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.712961196899414, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.4087592363357544, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.630133260149208e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.630133260149208e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.681794166564941, "val_fraction": 4.756468797564688e-05, "val_loss": 5.655233383178711, "within_bucket_gap": 0.9734392166137695}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.96086311340332, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.9707757234573364, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.8531546592712402, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.917680581410726, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010373861290786636, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010373861290786636, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.11144660413265228, "val_fraction": 2.378234398782344e-05, "val_loss": 4.473447799682617, "within_bucket_gap": 4.362001195549965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.877264022827148, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.9712564945220947, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.3748067319393158, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.529828071594238, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.5689392409186745e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.5689392409186745e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.898310661315918, "val_fraction": 4.756468797564688e-05, "val_loss": 7.147976875305176, "within_bucket_gap": -0.7503337860107422}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.800919532775879, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.98886513710022, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.6703548431396484, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.305586104160756e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.305586104160756e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1621360778808594, "val_fraction": 2.378234398782344e-05, "val_loss": 3.711108922958374, "within_bucket_gap": 0.5489728450775146}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.14737825560606e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.14737825560606e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.1612868309021, "val_fraction": 2.378234398782344e-05, "val_loss": 8.166616439819336, "within_bucket_gap": 3.0053296089172363}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.7203164712116235e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.7203164712116235e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.58355712890625, "val_fraction": 2.378234398782344e-05, "val_loss": 3.5987584590911865, "within_bucket_gap": -1.9847986698150635}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.608244895935059, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.770736694335938, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.403120040893555, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00043033077020079033, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00043033077020079033, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.2095346450805664, "val_fraction": 4.756468797564688e-05, "val_loss": 11.256808757781982, "within_bucket_gap": 9.047274112701416}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.0511893630027771, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.752969741821289, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.1132612228393555, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.6456570625305176, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.564507484436035, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.324398184294388e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.324398184294388e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.702929496765137, "val_fraction": 2.378234398782344e-05, "val_loss": 2.2026865482330322, "within_bucket_gap": -3.5002429485321045}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.22864818572998, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.5246479213237762, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.900438632413495e-07, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.900438632413495e-07, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.3535492420196533, "val_fraction": 2.378234398782344e-05, "val_loss": 1.316124677658081, "within_bucket_gap": -0.037424564361572266}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.035564362186275e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.035564362186275e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.757901430130005, "val_fraction": 2.378234398782344e-05, "val_loss": 3.19333553314209, "within_bucket_gap": 0.43543410301208496}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7230846881866455, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.7421746253967285, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.085307598114014, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.90293550491333, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.161269187927246, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.18783950805664, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.320061683654785, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.965576595185917e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.965576595185917e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.2740046977996826, "val_fraction": 2.378234398782344e-05, "val_loss": 2.027039051055908, "within_bucket_gap": -1.2469656467437744}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.812924861907959, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012201036764606492, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012201036764606492, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.074392318725586, "val_fraction": 2.378234398782344e-05, "val_loss": 2.9441003799438477, "within_bucket_gap": -5.130291938781738}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020297814466773646, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020297814466773646, "step": 600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 9.061750411987305, "val_fraction": 2.378234398782344e-05, "val_loss": 0.5269253849983215, "within_bucket_gap": -8.534825026988983}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.866202189316306e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.866202189316306e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.024375915527344, "val_fraction": 2.378234398782344e-05, "val_loss": 7.9114766120910645, "within_bucket_gap": 2.8871006965637207}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.892456531524658, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.69008904069526e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.69008904069526e-05, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.350583076477051, "val_fraction": 2.378234398782344e-05, "val_loss": 10.425071716308594, "within_bucket_gap": 4.074488639831543}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.146032333374023, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.7418472766876221, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.899688720703125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.42284631729126, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.945402145385742, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.1545785665512085, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.286455154418945, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.18594689667224884, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.605470180511475, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.2947869300842285, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.846481812054708e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.846481812054708e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.353151321411133, "val_fraction": 2.378234398782344e-05, "val_loss": 4.1295599937438965, "within_bucket_gap": 0.7764086723327637}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.111644744873047, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6853015422821045, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.4689824988036395e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.4689824988036395e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.8351146578788757, "val_fraction": 2.378234398782344e-05, "val_loss": 2.87327241897583, "within_bucket_gap": 1.0381577610969543}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001768226426270545, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001768226426270545, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.3516899049282074, "val_fraction": 2.378234398782344e-05, "val_loss": 7.786728382110596, "within_bucket_gap": 7.435038477182388}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.591018676757812, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.307783126831055, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.07286370545625687, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018872921688190152, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018872921688190152, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.423125267028809, "val_fraction": 2.378234398782344e-05, "val_loss": 12.358811378479004, "within_bucket_gap": 7.935686111450195}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.833702069620745e-05, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.833702069620745e-05, "step": 600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.584853172302246, "val_fraction": 4.756468797564688e-05, "val_loss": 3.7276556491851807, "within_bucket_gap": -1.8571975231170654}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.1466169357299805, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.817986011505127, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.66228723526001, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.04965577274560928, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.16006474196910858, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.04026229679584503, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.178289413452148, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.012329667806625366, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.1942878116029945e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.1942878116029945e-05, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.638824224472046, "val_fraction": 2.378234398782344e-05, "val_loss": 4.140998363494873, "within_bucket_gap": 0.5021741390228271}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.3962788581848145, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.293645858764648, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.076012783262827e-06, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.076012783262827e-06, "step": 600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.43330803513526917, "val_fraction": 2.378234398782344e-05, "val_loss": 0.8149362206459045, "within_bucket_gap": 0.3816281855106354}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.5517936302193042e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.5517936302193042e-05, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.8824465274810791, "val_fraction": 4.756468797564688e-05, "val_loss": 0.5561974346637726, "within_bucket_gap": -0.3262490928173065}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.212637237943652e-06, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.212637237943652e-06, "step": 600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.872578144073486, "val_fraction": 9.512937595129375e-05, "val_loss": 4.96942138671875, "within_bucket_gap": 0.09684324264526367}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.021157548517421923, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.021157548517421923, "step": 700, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 5.965099330646789, "val_fraction": 0.05995528919330289, "val_loss": 5.612210556326854, "within_bucket_gap": -0.35288877431993537}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0030185820800515406, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0030185820800515406, "step": 700, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 6.0079077585849365, "val_fraction": 0.019882039573820395, "val_loss": 5.856083190041866, "within_bucket_gap": -0.1518245685430708}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.002827780596063771, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.002827780596063771, "step": 700, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 5.94500154194079, "val_fraction": 0.011582001522070015, "val_loss": 5.700848526533624, "within_bucket_gap": -0.24415301540716516}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0010253343707191133, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0010253343707191133, "step": 700, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.775780821683114, "val_fraction": 0.007301179604261796, "val_loss": 5.63534675125967, "within_bucket_gap": -0.1404340704234439}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001433468140862456, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001433468140862456, "step": 700, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 5.967177785503926, "val_fraction": 0.004827815829528158, "val_loss": 5.670259222021244, "within_bucket_gap": -0.29691856348268253}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001989086104837873, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001989086104837873, "step": 700, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 6.161602047318262, "val_fraction": 0.003305745814307458, "val_loss": 5.559896345618817, "within_bucket_gap": -0.6017057016994452}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0017181190992274083, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0017181190992274083, "step": 700, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 5.985630277953396, "val_fraction": 0.002877663622526636, "val_loss": 5.388576791306172, "within_bucket_gap": -0.5970534866472237}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.7513544205403196e-05, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.7513544205403196e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.669536863054548, "val_fraction": 0.002449581430745814, "val_loss": 5.680768800013274, "within_bucket_gap": 0.011231936958726152}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004890922595975116, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004890922595975116, "step": 700, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 5.557226232580237, "val_fraction": 0.0025684931506849314, "val_loss": 5.366806312843606, "within_bucket_gap": -0.1904199197366312}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000554795856634021, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000554795856634021, "step": 700, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 5.3449302206234055, "val_fraction": 0.0014745053272450532, "val_loss": 5.721189191264491, "within_bucket_gap": 0.3762589706410857}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001254711210577317, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001254711210577317, "step": 700, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 5.7879049222763275, "val_fraction": 0.0020214992389649923, "val_loss": 5.167221428366268, "within_bucket_gap": -0.6206834939100592}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005341840641119141, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005341840641119141, "step": 700, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 5.112539333767361, "val_fraction": 0.0016647640791476407, "val_loss": 4.79166259765625, "within_bucket_gap": -0.3208767361111109}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008625275062590398, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008625275062590398, "step": 700, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 6.176999260397518, "val_fraction": 0.0018074581430745813, "val_loss": 5.699794568513569, "within_bucket_gap": -0.47720469188394876}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007649711555435192, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007649711555435192, "step": 700, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.7270424786735985, "val_fraction": 0.0015220700152207, "val_loss": 5.224456429481506, "within_bucket_gap": -0.5025860491920922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0011040915148634509, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0011040915148634509, "step": 700, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 6.212479207525609, "val_fraction": 0.0014745053272450532, "val_loss": 5.4636914653162805, "within_bucket_gap": -0.7487877422093288}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004703822564135217, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004703822564135217, "step": 700, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 5.553586383198583, "val_fraction": 0.0010226407914764079, "val_loss": 6.013554595237554, "within_bucket_gap": 0.4599682120389712}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.877999706166057e-05, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.877999706166057e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.711878569229789, "val_fraction": 0.0004280821917808219, "val_loss": 5.55120849609375, "within_bucket_gap": -0.1606700731360391}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.296217024695856e-05, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.296217024695856e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.508958025676448, "val_fraction": 0.0007372526636225266, "val_loss": 5.409993079400832, "within_bucket_gap": -0.09896494627561658}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00032851431104871946, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00032851431104871946, "step": 700, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.300132751464844, "val_fraction": 0.000594558599695586, "val_loss": 5.852667541503906, "within_bucket_gap": 0.5525347900390623}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00040486341560630847, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00040486341560630847, "step": 700, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.821956634521484, "val_fraction": 0.0007372526636225266, "val_loss": 6.371108147405809, "within_bucket_gap": 0.5491515128843245}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005068148031743719, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005068148031743719, "step": 700, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.596728909400202, "val_fraction": 0.0006183409436834094, "val_loss": 6.4163654033954325, "within_bucket_gap": 0.8196364939952305}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.4791184139396525e-05, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.4791184139396525e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.991138049534389, "val_fraction": 0.0006659056316590563, "val_loss": 6.073418753487723, "within_bucket_gap": 0.08228070395333376}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006761984783401953, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006761984783401953, "step": 700, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 4.678799390792847, "val_fraction": 0.00047564687975646877, "val_loss": 6.100439071655273, "within_bucket_gap": 1.4216396808624268}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015205771229945744, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015205771229945744, "step": 700, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.352199249267578, "val_fraction": 0.0003091704718417047, "val_loss": 4.844024071326623, "within_bucket_gap": 0.4918248220590451}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004524281100621302, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004524281100621302, "step": 700, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.6212248120989114, "val_fraction": 0.000594558599695586, "val_loss": 5.3821726989746095, "within_bucket_gap": 0.760947886875698}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00021914532195487497, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00021914532195487497, "step": 700, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.346848805745442, "val_fraction": 0.0004280821917808219, "val_loss": 5.8349253336588545, "within_bucket_gap": -0.511923472086588}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003567438780661964, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003567438780661964, "step": 700, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 4.980110916436887, "val_fraction": 0.000594558599695586, "val_loss": 5.580125579833984, "within_bucket_gap": 0.6000146633970971}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.3488261004577807e-05, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.3488261004577807e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.6364899548617275, "val_fraction": 0.00045186453576864533, "val_loss": 5.5845091970343335, "within_bucket_gap": -0.051980757827394086}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000984623474359992, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000984623474359992, "step": 700, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.6597057465584046, "val_fraction": 0.00040429984779299846, "val_loss": 5.09508503184599, "within_bucket_gap": 2.4353792852875853}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001409832100244862, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001409832100244862, "step": 700, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.430731667412652, "val_fraction": 0.00033295281582952814, "val_loss": 5.854164668491909, "within_bucket_gap": 0.42343300107925685}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00038397130719421477, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00038397130719421477, "step": 700, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.802324295043945, "val_fraction": 0.0002616057838660578, "val_loss": 5.334576520052823, "within_bucket_gap": -1.467747774991122}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0010445797612738453, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0010445797612738453, "step": 700, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.576478640238444, "val_fraction": 0.0005232115677321156, "val_loss": 4.5800018310546875, "within_bucket_gap": -1.9964768091837568}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003020142250893264, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003020142250893264, "step": 700, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.656984965006511, "val_fraction": 0.0002616057838660578, "val_loss": 5.502521861683238, "within_bucket_gap": -1.1544631033232724}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.345329239508228e-06, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.345329239508228e-06, "step": 700, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.784338421291775, "val_fraction": 0.0002616057838660578, "val_loss": 5.760083111849698, "within_bucket_gap": -0.024255309442076545}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002720255966055883, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002720255966055883, "step": 700, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 4.869939963022868, "val_fraction": 0.00014269406392694063, "val_loss": 6.776295344034831, "within_bucket_gap": 1.906355381011963}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001329354684798018, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001329354684798018, "step": 700, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.439459403355916, "val_fraction": 0.0002616057838660578, "val_loss": 6.931307532570579, "within_bucket_gap": -0.508151870785337}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.3408930414162002e-05, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.3408930414162002e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.387434005737305, "val_fraction": 0.00028538812785388126, "val_loss": 5.469458897908528, "within_bucket_gap": 0.08202489217122366}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005333863164736255, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005333863164736255, "step": 700, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.381663640340169, "val_fraction": 0.0003567351598173516, "val_loss": 5.886475118001302, "within_bucket_gap": -1.495188522338867}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00028141317695364146, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00028141317695364146, "step": 700, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.474878629048665, "val_fraction": 0.0002616057838660578, "val_loss": 6.550593289462003, "within_bucket_gap": 1.0757146604133379}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.767942464350005e-05, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.767942464350005e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.528070623224432, "val_fraction": 9.512937595129375e-05, "val_loss": 5.239516735076904, "within_bucket_gap": 0.7114461118524726}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002382023838919954, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002382023838919954, "step": 700, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 5.383909454345703, "val_fraction": 0.00021404109589041095, "val_loss": 6.496790991889106, "within_bucket_gap": 1.1128815375434025}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.70422497065984e-05, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.70422497065984e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 4.739454252895339, "val_fraction": 0.00014269406392694063, "val_loss": 4.269622166951497, "within_bucket_gap": -0.4698320859438416}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004752253346869312, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004752253346869312, "step": 700, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.59035130909511, "val_fraction": 0.00023782343987823439, "val_loss": 5.588578796386718, "within_bucket_gap": 1.9982274872916084}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006788176834537922, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006788176834537922, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 1.9055582284927368, "val_fraction": 0.0002616057838660578, "val_loss": 4.500369678844105, "within_bucket_gap": 2.5948114503513686}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014199945258545866, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014199945258545866, "step": 700, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.316260443793403, "val_fraction": 0.00016647640791476407, "val_loss": 6.169230869838169, "within_bucket_gap": 0.8529704260447666}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014110569540224117, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014110569540224117, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.389824390411377, "val_fraction": 0.000380517503805175, "val_loss": 5.018998622894287, "within_bucket_gap": -0.37082576751708984}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012422564444943186, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012422564444943186, "step": 700, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.7213320050920755, "val_fraction": 7.134703196347032e-05, "val_loss": 7.4624786376953125, "within_bucket_gap": 1.741146632603237}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.883135946918296e-05, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.883135946918296e-05, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.758967638015747, "val_fraction": 4.756468797564688e-05, "val_loss": 5.15487813949585, "within_bucket_gap": 0.39591050148010254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000133175915427151, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000133175915427151, "step": 700, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 3.138927055127693, "val_fraction": 4.756468797564688e-05, "val_loss": 5.938817501068115, "within_bucket_gap": 2.7998904459404224}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013187129302557306, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013187129302557306, "step": 700, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 6.064619931307706, "val_fraction": 0.00016647640791476407, "val_loss": 5.272487912859235, "within_bucket_gap": -0.7921320184484708}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003614691135727288, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003614691135727288, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.597120761871338, "val_fraction": 7.134703196347032e-05, "val_loss": 2.53076966603597, "within_bucket_gap": -5.066351095835367}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001315775595299185, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001315775595299185, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.721663236618042, "val_fraction": 4.756468797564688e-05, "val_loss": 1.9553766250610352, "within_bucket_gap": -2.766286611557007}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.977460153778377e-05, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.977460153778377e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.876346179417202, "val_fraction": 0.00014269406392694063, "val_loss": 6.177125771840413, "within_bucket_gap": -0.6992204075767887}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.178618375387729e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.178618375387729e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 3.123491048812866, "val_fraction": 2.378234398782344e-05, "val_loss": 2.207425594329834, "within_bucket_gap": -0.9160654544830322}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.1490825808393054e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.1490825808393054e-05, "step": 700, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.775513648986816, "val_fraction": 0.00016647640791476407, "val_loss": 5.5863527570452005, "within_bucket_gap": -0.1891608919416159}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016197675620765634, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016197675620765634, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.77432378133138, "val_fraction": 0.00014269406392694063, "val_loss": 6.909456888834636, "within_bucket_gap": 1.1351331075032558}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00035645813316877937, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00035645813316877937, "step": 700, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.687448453903198, "val_fraction": 0.00011891171993911719, "val_loss": 2.689778137207031, "within_bucket_gap": -2.9976703166961673}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013790415994017815, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013790415994017815, "step": 700, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.533266633749008, "val_fraction": 9.512937595129375e-05, "val_loss": 5.982915163040161, "within_bucket_gap": 1.449648529291153}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004659740642688408, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004659740642688408, "step": 700, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.005669355392456, "val_fraction": 0.00023782343987823439, "val_loss": 6.964997100830078, "within_bucket_gap": 1.959327745437622}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002290894996085667, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002290894996085667, "step": 700, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.834632873535156, "val_fraction": 0.00021404109589041095, "val_loss": 4.764326731363933, "within_bucket_gap": -1.0703061421712237}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001104699059345588, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001104699059345588, "step": 700, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.601747512817383, "val_fraction": 0.00011891171993911719, "val_loss": 5.672739791870117, "within_bucket_gap": -0.9290077209472658}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.067049739204769e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.067049739204769e-05, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.961872100830078, "val_fraction": 7.134703196347032e-05, "val_loss": 6.111514409383138, "within_bucket_gap": -0.8503576914469404}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00021469989958781995, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00021469989958781995, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.876161575317383, "val_fraction": 0.00016647640791476407, "val_loss": 5.586489949907575, "within_bucket_gap": -1.2896716254098077}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017319863850183737, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017319863850183737, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.481927553812663, "val_fraction": 0.0001902587519025875, "val_loss": 7.39225959777832, "within_bucket_gap": 0.9103320439656573}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.218859004466235e-06, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.218859004466235e-06, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.455284357070923, "val_fraction": 0.0002616057838660578, "val_loss": 5.431512485850941, "within_bucket_gap": -0.023771871219981477}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.567756983424612e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.567756983424612e-05, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.027076005935669, "val_fraction": 4.756468797564688e-05, "val_loss": 6.566921234130859, "within_bucket_gap": 0.5398452281951904}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015955417214281843, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015955417214281843, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.189130306243896, "val_fraction": 7.134703196347032e-05, "val_loss": 5.952819029490153, "within_bucket_gap": -2.236311276753743}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.1882599949655286e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.1882599949655286e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.145118713378906, "val_fraction": 4.756468797564688e-05, "val_loss": 7.0543389320373535, "within_bucket_gap": -1.0907797813415527}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00022129948099272798, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00022129948099272798, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.636163234710693, "val_fraction": 9.512937595129375e-05, "val_loss": 7.96246337890625, "within_bucket_gap": 2.3263001441955566}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.630936153766045e-05, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.630936153766045e-05, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.308908462524414, "val_fraction": 9.512937595129375e-05, "val_loss": 4.795712471008301, "within_bucket_gap": 0.4868040084838867}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001725481706848609, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001725481706848609, "step": 700, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.748739624023438, "val_fraction": 7.134703196347032e-05, "val_loss": 3.3303044637044272, "within_bucket_gap": -2.4184351603190106}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004935827746964662, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004935827746964662, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.779496431350708, "val_fraction": 0.0001902587519025875, "val_loss": 4.3737674951553345, "within_bucket_gap": 2.5942710638046265}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.981351174538901e-06, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.981351174538901e-06, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.570132255554199, "val_fraction": 2.378234398782344e-05, "val_loss": 3.989828109741211, "within_bucket_gap": 0.4196958541870117}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001966894308000213, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001966894308000213, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.1422085762023926, "val_fraction": 7.134703196347032e-05, "val_loss": 4.899007638295491, "within_bucket_gap": 2.7567990620930987}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001938047689843577, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001938047689843577, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.7030746936798096, "val_fraction": 0.00011891171993911719, "val_loss": 4.332895278930664, "within_bucket_gap": 1.6298205852508545}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001747276254077662, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001747276254077662, "step": 700, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.230703830718994, "val_fraction": 7.134703196347032e-05, "val_loss": 7.679686228434245, "within_bucket_gap": 2.448982397715251}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001042012800001844, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001042012800001844, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.1096813678741455, "val_fraction": 9.512937595129375e-05, "val_loss": 7.205045223236084, "within_bucket_gap": 1.0953638553619385}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.684222221374512, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010698503010893521, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010698503010893521, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.7114949226379395, "val_fraction": 4.756468797564688e-05, "val_loss": 7.960748195648193, "within_bucket_gap": 2.249253273010254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018005993903318315, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018005993903318315, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.956321716308594, "val_fraction": 4.756468797564688e-05, "val_loss": 2.170741558074951, "within_bucket_gap": -3.7855801582336426}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.327844142913818, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0261774244555237e-05, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0261774244555237e-05, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.675897598266602, "val_fraction": 0.0001902587519025875, "val_loss": 4.621961712837219, "within_bucket_gap": -0.053935885429382324}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003707048283320039, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003707048283320039, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.234479904174805, "val_fraction": 7.134703196347032e-05, "val_loss": 7.0386810302734375, "within_bucket_gap": -5.195798873901367}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017723787570290004, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017723787570290004, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.809385299682617, "val_fraction": 4.756468797564688e-05, "val_loss": 4.083136200904846, "within_bucket_gap": -3.726249098777771}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.524503230201844e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.524503230201844e-05, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.203076680501302, "val_fraction": 9.512937595129375e-05, "val_loss": 4.67869246006012, "within_bucket_gap": 0.47561577955881784}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00029904122040514894, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00029904122040514894, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.558446884155273, "val_fraction": 9.512937595129375e-05, "val_loss": 3.4149255752563477, "within_bucket_gap": -3.143521308898926}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.212917757179457e-06, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.212917757179457e-06, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.086559295654297, "val_fraction": 2.378234398782344e-05, "val_loss": 3.2637040615081787, "within_bucket_gap": 0.17714476585388184}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.0331919703127949e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.0331919703127949e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.370018720626831, "val_fraction": 2.378234398782344e-05, "val_loss": 3.804455280303955, "within_bucket_gap": 0.434436559677124}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.235147682680569e-05, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.235147682680569e-05, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.073856353759766, "val_fraction": 4.756468797564688e-05, "val_loss": 8.594973802566528, "within_bucket_gap": 1.5211174488067627}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001643493408663269, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001643493408663269, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.2839369773864746, "val_fraction": 2.378234398782344e-05, "val_loss": 10.194498062133789, "within_bucket_gap": 6.9105610847473145}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.494879047620242e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.494879047620242e-05, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.0449934005737305, "val_fraction": 4.756468797564688e-05, "val_loss": 7.35927677154541, "within_bucket_gap": 0.3142833709716797}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016193461563307764, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016193461563307764, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.7625322341918945, "val_fraction": 2.378234398782344e-05, "val_loss": 8.571558952331543, "within_bucket_gap": 6.809026718139648}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001295266403636613, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001295266403636613, "step": 700, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.1234699249267575, "val_fraction": 2.378234398782344e-05, "val_loss": 10.569806098937988, "within_bucket_gap": 5.446336174011231}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013964394847415897, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013964394847415897, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.629335969686508, "val_fraction": 4.756468797564688e-05, "val_loss": 6.565210342407227, "within_bucket_gap": 2.9358743727207184}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004042104181452247, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004042104181452247, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.862345695495605, "val_fraction": 7.134703196347032e-05, "val_loss": 5.196932474772136, "within_bucket_gap": -5.66541322072347}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.927588701248169, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.142786026000977, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.852849702645171e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.852849702645171e-05, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.385141571362813, "val_fraction": 2.378234398782344e-05, "val_loss": 8.107587814331055, "within_bucket_gap": 3.7224462429682417}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5485808849334717, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.510927200317383, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.0360449076243186e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.0360449076243186e-05, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.03965699672699, "val_fraction": 2.378234398782344e-05, "val_loss": 4.475293159484863, "within_bucket_gap": 0.43563616275787354}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.4685306549072266, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.965483665466309, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.480000495910645, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00028911896497918406, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00028911896497918406, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.456827849149704, "val_fraction": 4.756468797564688e-05, "val_loss": 8.53526496887207, "within_bucket_gap": 6.078437119722366}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.653404639373268e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.653404639373268e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.356993675231934, "val_fraction": 2.378234398782344e-05, "val_loss": 6.241290092468262, "within_bucket_gap": -1.1157035827636719}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002336970117946738, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002336970117946738, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.7986127138137817, "val_fraction": 7.134703196347032e-05, "val_loss": 5.07411003112793, "within_bucket_gap": 3.275497317314148}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.396102905273438, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002996128171546274, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002996128171546274, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.718453884124756, "val_fraction": 0.00014269406392694063, "val_loss": 3.618767261505127, "within_bucket_gap": -2.099686622619629}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.150607974561926e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.150607974561926e-05, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.853977680206299, "val_fraction": 4.756468797564688e-05, "val_loss": 7.357321500778198, "within_bucket_gap": 1.5033438205718994}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.1713600742943935e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.1713600742943935e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.6252360343933105, "val_fraction": 2.378234398782344e-05, "val_loss": 6.379209518432617, "within_bucket_gap": 1.7539734840393066}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001619016266848943, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001619016266848943, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.923702239990234, "val_fraction": 2.378234398782344e-05, "val_loss": 3.116062641143799, "within_bucket_gap": -6.8076395988464355}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.809468848520218e-06, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.809468848520218e-06, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.063373565673828, "val_fraction": 2.378234398782344e-05, "val_loss": 7.265602111816406, "within_bucket_gap": 0.20222854614257812}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.371589422225952, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.3884204626083374, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.479384436092057e-06, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.479384436092057e-06, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.9767472743988037, "val_fraction": 4.756468797564688e-05, "val_loss": 4.155017852783203, "within_bucket_gap": 0.17827057838439941}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.03828239440918, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.141709327697754, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.1921048164367676, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.01716963450114, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.875635400119618e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.875635400119618e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.4322785139083862, "val_fraction": 2.378234398782344e-05, "val_loss": 5.164305686950684, "within_bucket_gap": 3.7320271730422974}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.060598373413086, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.4591450691223145, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.09734714776277542, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.839417457580566, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.870183234163978e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.870183234163978e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.725160121917725, "val_fraction": 4.756468797564688e-05, "val_loss": 6.07053279876709, "within_bucket_gap": -1.6546273231506348}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.723304748535156, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.855050086975098, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.5149197578430176, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.67453556663188e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.67453556663188e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.41729736328125, "val_fraction": 2.378234398782344e-05, "val_loss": 6.121406078338623, "within_bucket_gap": 0.704108715057373}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.928864119078289e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.928864119078289e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.562017440795898, "val_fraction": 2.378234398782344e-05, "val_loss": 8.634506225585938, "within_bucket_gap": 2.072488784790039}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.251600013657429e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.251600013657429e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.947413444519043, "val_fraction": 2.378234398782344e-05, "val_loss": 5.473686218261719, "within_bucket_gap": 0.5262727737426758}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.2444000244140625, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.868642807006836, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.211013793945312, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00027106917793529397, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00027106917793529397, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.093461513519287, "val_fraction": 4.756468797564688e-05, "val_loss": 10.792419910430908, "within_bucket_gap": 5.698958396911621}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.6678862571716309, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.878746032714844, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.839327096939087, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.983644962310791, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.287912368774414, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.9123344593759355e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.9123344593759355e-05, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.579140901565552, "val_fraction": 2.378234398782344e-05, "val_loss": 2.934082508087158, "within_bucket_gap": -1.6450583934783936}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.118412971496582, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5902132391929626, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.59901735490497e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.59901735490497e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.060922861099243, "val_fraction": 2.378234398782344e-05, "val_loss": 1.3885680437088013, "within_bucket_gap": -0.6723548173904419}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.4580061675206706e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.4580061675206706e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.44779109954834, "val_fraction": 2.378234398782344e-05, "val_loss": 2.152808666229248, "within_bucket_gap": -2.294982433319092}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.779355049133301, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.424088478088379, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.785870313644409, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.759277820587158, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.693254470825195, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.510499954223633, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.4976511001586914, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.1413886893648353e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.1413886893648353e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.4697985649108887, "val_fraction": 2.378234398782344e-05, "val_loss": 1.9898674488067627, "within_bucket_gap": -0.479931116104126}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.787382125854492, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.738461113048652e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.738461113048652e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.222426414489746, "val_fraction": 2.378234398782344e-05, "val_loss": 4.127598285675049, "within_bucket_gap": -4.094828128814697}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012740780864494938, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012740780864494938, "step": 700, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.448678016662598, "val_fraction": 2.378234398782344e-05, "val_loss": 2.0914344787597656, "within_bucket_gap": -5.357243537902832}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010307390622715245, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010307390622715245, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6177802085876465, "val_fraction": 2.378234398782344e-05, "val_loss": 7.951831817626953, "within_bucket_gap": 4.334051609039307}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.9434003829956055, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.2650491100830935e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.2650491100830935e-05, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.987899541854858, "val_fraction": 2.378234398782344e-05, "val_loss": 8.360787391662598, "within_bucket_gap": 1.3728878498077393}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.10361099243164, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.0364412069320679, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.21123719215393066, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.799554824829102, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.609945297241211, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.829737901687622, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.969086647033691, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.024907264858484268, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.9188926219940186, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.277150869369507, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.7826307899512837e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.7826307899512837e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7540791034698486, "val_fraction": 2.378234398782344e-05, "val_loss": 3.004518508911133, "within_bucket_gap": -0.7495605945587158}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.671324729919434, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.520967483520508, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.0781672266248152e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.0781672266248152e-05, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.2425753474235535, "val_fraction": 2.378234398782344e-05, "val_loss": 3.116403102874756, "within_bucket_gap": 0.8738277554512024}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001923265112804949, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001923265112804949, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.3669990599155426, "val_fraction": 2.378234398782344e-05, "val_loss": 8.453944206237793, "within_bucket_gap": 8.08694514632225}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.151853561401367, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.394941329956055, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.17358805239200592, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011981668197401038, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011981668197401038, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.889604330062866, "val_fraction": 2.378234398782344e-05, "val_loss": 9.927656173706055, "within_bucket_gap": 5.0380518436431885}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.195334736433565e-05, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.195334736433565e-05, "step": 700, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.029840469360352, "val_fraction": 4.756468797564688e-05, "val_loss": 4.306853294372559, "within_bucket_gap": -1.722987174987793}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.681855201721191, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.6307902336120605, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.2114105224609375, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.5903171896934509, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.8962134718894958, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.06118129938840866, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.987403869628906, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.004792749881744385, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.753728320790934e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.753728320790934e-05, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.001132965087891, "val_fraction": 2.378234398782344e-05, "val_loss": 4.7385406494140625, "within_bucket_gap": 0.7374076843261719}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.074990272521973, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.662652969360352, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.5212064476136566e-06, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.5212064476136566e-06, "step": 700, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.7540428638458252, "val_fraction": 2.378234398782344e-05, "val_loss": 0.6480311751365662, "within_bucket_gap": -0.10601168870925903}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.4289526078889722e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.4289526078889722e-05, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.9362096786499023, "val_fraction": 4.756468797564688e-05, "val_loss": 0.6357866823673248, "within_bucket_gap": -0.3004229962825775}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.6863085352666846e-05, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.6863085352666846e-05, "step": 700, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.525512218475342, "val_fraction": 9.512937595129375e-05, "val_loss": 5.243127465248108, "within_bucket_gap": -0.2823847532272339}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.020874002277946797, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.020874002277946797, "step": 800, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 6.091046184290205, "val_fraction": 0.05995528919330289, "val_loss": 5.742886704804641, "within_bucket_gap": -0.348159479485564}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00335344408809988, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00335344408809988, "step": 800, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 6.114633662699071, "val_fraction": 0.019882039573820395, "val_loss": 5.945966656698564, "within_bucket_gap": -0.16866700600050688}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.002615409819903903, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.002615409819903903, "step": 800, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 6.048398694490132, "val_fraction": 0.011582001522070015, "val_loss": 5.822581955050051, "within_bucket_gap": -0.22581673944008074}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005028107452209254, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005028107452209254, "step": 800, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.849702533922698, "val_fraction": 0.007301179604261796, "val_loss": 5.780835477847618, "within_bucket_gap": -0.0688670560750797}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0012622055550676134, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0012622055550676134, "step": 800, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 6.002012242821499, "val_fraction": 0.004827815829528158, "val_loss": 5.740567813365917, "within_bucket_gap": -0.26144442945558133}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0009912076810661798, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0009912076810661798, "step": 800, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 6.237387799202128, "val_fraction": 0.003305745814307458, "val_loss": 5.937543910184353, "within_bucket_gap": -0.29984388901777503}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.002538279053289358, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.002538279053289358, "step": 800, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 6.163469083047326, "val_fraction": 0.002877663622526636, "val_loss": 5.281406623272856, "within_bucket_gap": -0.8820624597744704}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013719220787682072, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013719220787682072, "step": 800, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.621447699410575, "val_fraction": 0.002449581430745814, "val_loss": 5.677454087340716, "within_bucket_gap": 0.056006387930141344}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003219536661223943, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003219536661223943, "step": 800, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 5.6455581252639355, "val_fraction": 0.0025684931506849314, "val_loss": 5.520210831253617, "within_bucket_gap": -0.12534729401031885}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006740406205113286, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006740406205113286, "step": 800, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 5.2867152934171715, "val_fraction": 0.0014745053272450532, "val_loss": 5.74384529359879, "within_bucket_gap": 0.45713000018161853}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0014340167064384004, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0014340167064384004, "step": 800, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 5.953379591850386, "val_fraction": 0.0020214992389649923, "val_loss": 5.243996833352481, "within_bucket_gap": -0.7093827584979042}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005113725309018737, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005113725309018737, "step": 800, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 5.379352823893229, "val_fraction": 0.0016647640791476407, "val_loss": 5.072178649902344, "within_bucket_gap": -0.30717417399088554}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.37344557182968e-05, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.37344557182968e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.952581405639648, "val_fraction": 0.0018074581430745813, "val_loss": 5.9062542162443465, "within_bucket_gap": -0.04632718939530189}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.192075273922085e-05, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.192075273922085e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.29644708072438, "val_fraction": 0.0015220700152207, "val_loss": 5.330559015274048, "within_bucket_gap": 0.0341119345496681}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008827045693425798, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008827045693425798, "step": 800, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 6.306676967510049, "val_fraction": 0.0014745053272450532, "val_loss": 5.708032423450101, "within_bucket_gap": -0.5986445440599484}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.4473858489290105e-06, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.4473858489290105e-06, "step": 800, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 5.951873247013536, "val_fraction": 0.0010226407914764079, "val_loss": 5.953288588412972, "within_bucket_gap": 0.0014153413994364428}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002866339678799371, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002866339678799371, "step": 800, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.896037723707116, "val_fraction": 0.0004280821917808219, "val_loss": 5.226460774739583, "within_bucket_gap": -0.6695769489675332}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012887518606919806, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012887518606919806, "step": 800, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.634263759706078, "val_fraction": 0.0007372526636225266, "val_loss": 5.459459120227445, "within_bucket_gap": -0.17480463947863356}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001500969277591541, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001500969277591541, "step": 800, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.740053858075823, "val_fraction": 0.000594558599695586, "val_loss": 5.9925048828125, "within_bucket_gap": 0.25245102473667647}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.40418727224011e-05, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.40418727224011e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.5746134244478664, "val_fraction": 0.0007372526636225266, "val_loss": 6.51487559656943, "within_bucket_gap": -0.05973782787843618}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004688077186715209, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004688077186715209, "step": 800, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.650914346018145, "val_fraction": 0.0006183409436834094, "val_loss": 6.409084613506611, "within_bucket_gap": 0.7581702674884658}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003885586875032977, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003885586875032977, "step": 800, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.953751564025879, "val_fraction": 0.0006659056316590563, "val_loss": 6.537255695887974, "within_bucket_gap": 0.583504131862095}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006744589208649353, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006744589208649353, "step": 800, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 4.890230417251587, "val_fraction": 0.00047564687975646877, "val_loss": 6.308212852478027, "within_bucket_gap": 1.41798243522644}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00023785866014489282, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00023785866014489282, "step": 800, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.548469772338867, "val_fraction": 0.0003091704718417047, "val_loss": 5.317814460167518, "within_bucket_gap": 0.7693446878286503}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016811991624402864, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016811991624402864, "step": 800, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.111163820539202, "val_fraction": 0.000594558599695586, "val_loss": 5.393928070068359, "within_bucket_gap": 0.2827642495291567}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004947511209926282, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004947511209926282, "step": 800, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 7.119254430135091, "val_fraction": 0.0004280821917808219, "val_loss": 5.963515811496311, "within_bucket_gap": -1.1557386186387797}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00045783046582863245, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00045783046582863245, "step": 800, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 4.884479485306085, "val_fraction": 0.000594558599695586, "val_loss": 5.6545137023925784, "within_bucket_gap": 0.7700342170864936}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.619795191184048e-05, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.619795191184048e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.993124701760032, "val_fraction": 0.00045186453576864533, "val_loss": 6.02897162186472, "within_bucket_gap": 0.03584692010468782}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0012081898692173135, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0012081898692173135, "step": 800, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.286865234375, "val_fraction": 0.00040429984779299846, "val_loss": 5.275216270895565, "within_bucket_gap": 2.988351036520565}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004741471470581706, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004741471470581706, "step": 800, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.838266372680664, "val_fraction": 0.00033295281582952814, "val_loss": 6.262333461216518, "within_bucket_gap": 1.4240670885358542}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003080938629454493, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003080938629454493, "step": 800, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.802602052688599, "val_fraction": 0.0002616057838660578, "val_loss": 5.624899257313121, "within_bucket_gap": -1.1777027953754775}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0014253458748125052, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0014253458748125052, "step": 800, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 7.15358829498291, "val_fraction": 0.0005232115677321156, "val_loss": 4.429363597523082, "within_bucket_gap": -2.724224697459828}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004943324293599158, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004943324293599158, "step": 800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.996912320454915, "val_fraction": 0.0002616057838660578, "val_loss": 5.107304139570757, "within_bucket_gap": -1.8896081808841583}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.126899147243347e-05, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.126899147243347e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.953486654493544, "val_fraction": 0.0002616057838660578, "val_loss": 5.795734058726918, "within_bucket_gap": -0.1577525957666257}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002087727075660973, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002087727075660973, "step": 800, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.107668240865071, "val_fraction": 0.00014269406392694063, "val_loss": 6.570747375488281, "within_bucket_gap": 1.46307913462321}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015137167690007112, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015137167690007112, "step": 800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.398343563079834, "val_fraction": 0.0002616057838660578, "val_loss": 6.819718447598544, "within_bucket_gap": -0.57862511548129}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.5543709353978504e-05, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.5543709353978504e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.692801255446214, "val_fraction": 0.00028538812785388126, "val_loss": 5.498176097869873, "within_bucket_gap": -0.1946251575763407}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004613562125593559, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004613562125593559, "step": 800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.537790616353353, "val_fraction": 0.0003567351598173516, "val_loss": 6.244516881306966, "within_bucket_gap": -1.2932737350463865}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018435321797397999, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018435321797397999, "step": 800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.832665363947551, "val_fraction": 0.0002616057838660578, "val_loss": 6.537363919344815, "within_bucket_gap": 0.7046985553972647}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013866456065976933, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013866456065976933, "step": 800, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.99231598593972, "val_fraction": 9.512937595129375e-05, "val_loss": 6.449957847595215, "within_bucket_gap": 1.4576418616554951}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00029643873282944957, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00029643873282944957, "step": 800, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 5.2454362487792965, "val_fraction": 0.00021404109589041095, "val_loss": 6.630398008558485, "within_bucket_gap": 1.3849617597791886}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.603674211882683e-05, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.603674211882683e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 4.737053710144836, "val_fraction": 0.00014269406392694063, "val_loss": 5.339999198913574, "within_bucket_gap": 0.6029454887687384}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00045087935160387845, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00045087935160387845, "step": 800, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.6439360891069685, "val_fraction": 0.00023782343987823439, "val_loss": 5.539793586730957, "within_bucket_gap": 1.8958574976239881}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008065409850629678, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008065409850629678, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 1.928200602531433, "val_fraction": 0.0002616057838660578, "val_loss": 5.011240178888494, "within_bucket_gap": 3.083039576357061}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.4489505267195558e-05, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.4489505267195558e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.795868343777126, "val_fraction": 0.00016647640791476407, "val_loss": 5.648763384137835, "within_bucket_gap": -0.14710495963929127}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008571484045350932, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008571484045350932, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.494958877563477, "val_fraction": 0.000380517503805175, "val_loss": 4.2423728704452515, "within_bucket_gap": -2.252586007118225}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011808274362003782, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011808274362003782, "step": 800, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.000368118286133, "val_fraction": 7.134703196347032e-05, "val_loss": 7.655415852864583, "within_bucket_gap": 1.6550477345784502}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.832872403629657e-05, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.832872403629657e-05, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.2183321714401245, "val_fraction": 4.756468797564688e-05, "val_loss": 4.202269077301025, "within_bucket_gap": -1.0160630941390991}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014686074743114282, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014686074743114282, "step": 800, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 1.8533431255456172, "val_fraction": 4.756468797564688e-05, "val_loss": 4.940943479537964, "within_bucket_gap": 3.0876003539923467}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.403586289456096e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.403586289456096e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.7287328893488105, "val_fraction": 0.00016647640791476407, "val_loss": 5.344078608921596, "within_bucket_gap": -0.38465428042721417}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025216003531190355, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025216003531190355, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.732486724853516, "val_fraction": 7.134703196347032e-05, "val_loss": 4.198211669921875, "within_bucket_gap": -3.5342750549316406}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016832319931715407, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016832319931715407, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.915351152420044, "val_fraction": 4.756468797564688e-05, "val_loss": 1.3765242099761963, "within_bucket_gap": -3.5388269424438477}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012221294438017263, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012221294438017263, "step": 800, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.832573686327253, "val_fraction": 0.00014269406392694063, "val_loss": 5.976105372111003, "within_bucket_gap": -0.8564683142162499}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.8465195492340912e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.8465195492340912e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 3.707022786140442, "val_fraction": 2.378234398782344e-05, "val_loss": 2.510118246078491, "within_bucket_gap": -1.1969045400619507}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000147000549409306, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000147000549409306, "step": 800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.169967269897461, "val_fraction": 0.00016647640791476407, "val_loss": 5.2869559696742465, "within_bucket_gap": -0.8830113002232141}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001504384604944667, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001504384604944667, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.374746958414714, "val_fraction": 0.00014269406392694063, "val_loss": 6.4290196895599365, "within_bucket_gap": 1.0542727311452227}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014777704109430673, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014777704109430673, "step": 800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.222549605369568, "val_fraction": 0.00011891171993911719, "val_loss": 3.9798038005828857, "within_bucket_gap": -1.242745804786682}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021477279790550969, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021477279790550969, "step": 800, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.926904410123825, "val_fraction": 9.512937595129375e-05, "val_loss": 7.184596061706543, "within_bucket_gap": 2.257691651582718}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007253719814110382, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007253719814110382, "step": 800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.149217367172241, "val_fraction": 0.00023782343987823439, "val_loss": 7.199261474609375, "within_bucket_gap": 3.0500441074371336}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003257635498881522, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003257635498881522, "step": 800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.366434097290039, "val_fraction": 0.00021404109589041095, "val_loss": 4.844466792212592, "within_bucket_gap": -1.521967305077447}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.224346041860832e-05, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.224346041860832e-05, "step": 800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 7.202144622802734, "val_fraction": 0.00011891171993911719, "val_loss": 6.594606018066406, "within_bucket_gap": -0.6075386047363285}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.693828291726078e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.693828291726078e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.202292442321777, "val_fraction": 7.134703196347032e-05, "val_loss": 6.5798594156901045, "within_bucket_gap": 0.3775669733683271}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.2922001801760027e-05, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.2922001801760027e-05, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.818677365779877, "val_fraction": 0.00016647640791476407, "val_loss": 6.68098817552839, "within_bucket_gap": -0.13768919025148652}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004983552061078029, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004983552061078029, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.922955513000488, "val_fraction": 0.0001902587519025875, "val_loss": 8.5423104763031, "within_bucket_gap": 2.6193549633026123}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012376822837411528, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012376822837411528, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.799297332763672, "val_fraction": 0.0002616057838660578, "val_loss": 5.326187653975054, "within_bucket_gap": -0.4731096787886182}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.170259528327024e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.170259528327024e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.796464204788208, "val_fraction": 4.756468797564688e-05, "val_loss": 5.709468841552734, "within_bucket_gap": -1.0869953632354736}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017281419270114812, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017281419270114812, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.42068076133728, "val_fraction": 7.134703196347032e-05, "val_loss": 5.998517036437988, "within_bucket_gap": -2.422163724899292}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.7085315952562305e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.7085315952562305e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.126996994018555, "val_fraction": 4.756468797564688e-05, "val_loss": 7.347315311431885, "within_bucket_gap": -0.7796816825866699}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.8248819325068225e-05, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.8248819325068225e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.9984917640686035, "val_fraction": 9.512937595129375e-05, "val_loss": 7.806660175323486, "within_bucket_gap": -0.1918315887451172}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013440459558408555, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013440459558408555, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.7425787448883057, "val_fraction": 9.512937595129375e-05, "val_loss": 5.155439853668213, "within_bucket_gap": 1.4128611087799072}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020675769861067448, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020675769861067448, "step": 800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.429586791992188, "val_fraction": 7.134703196347032e-05, "val_loss": 3.531670888264974, "within_bucket_gap": -2.8979159037272137}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00026044870983338606, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00026044870983338606, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.7767043113708496, "val_fraction": 0.0001902587519025875, "val_loss": 5.145622730255127, "within_bucket_gap": 1.3689184188842773}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0892488005680216e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0892488005680216e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.612623691558838, "val_fraction": 2.378234398782344e-05, "val_loss": 4.154616355895996, "within_bucket_gap": -0.4580073356628418}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00025534994129085834, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00025534994129085834, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.3902398347854614, "val_fraction": 7.134703196347032e-05, "val_loss": 4.969224611918132, "within_bucket_gap": 3.5789847771326704}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020561344592720046, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020561344592720046, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.170558452606201, "val_fraction": 0.00011891171993911719, "val_loss": 4.899685287475586, "within_bucket_gap": 1.7291268348693851}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018259687873689367, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018259687873689367, "step": 800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.608832677205403, "val_fraction": 7.134703196347032e-05, "val_loss": 8.168110529581705, "within_bucket_gap": 2.559277852376302}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018661146022413419, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018661146022413419, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.329645872116089, "val_fraction": 9.512937595129375e-05, "val_loss": 7.2913055419921875, "within_bucket_gap": 1.9616596698760986}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.333210945129395, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.159167039339944e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.159167039339944e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.09212064743042, "val_fraction": 4.756468797564688e-05, "val_loss": 5.79721736907959, "within_bucket_gap": -1.29490327835083}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016669247475207486, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016669247475207486, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.306713581085205, "val_fraction": 4.756468797564688e-05, "val_loss": 2.802170991897583, "within_bucket_gap": -3.504542589187622}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.963485479354858, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.447939914836912e-05, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.447939914836912e-05, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.720035552978516, "val_fraction": 0.0001902587519025875, "val_loss": 4.4862518310546875, "within_bucket_gap": -0.23378372192382812}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00027409341055871506, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00027409341055871506, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.127180099487305, "val_fraction": 7.134703196347032e-05, "val_loss": 7.2854868570963545, "within_bucket_gap": -3.8416932423909502}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018512108894787967, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018512108894787967, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.320378541946411, "val_fraction": 4.756468797564688e-05, "val_loss": 3.428392767906189, "within_bucket_gap": -3.891985774040222}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.610661542216029e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.610661542216029e-05, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.170597076416016, "val_fraction": 9.512937595129375e-05, "val_loss": 5.180869817733765, "within_bucket_gap": 1.010272741317749}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013260568295076739, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013260568295076739, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.51355504989624, "val_fraction": 9.512937595129375e-05, "val_loss": 5.119604110717773, "within_bucket_gap": -1.3939509391784668}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0260339695931931e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0260339695931931e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.494149684906006, "val_fraction": 2.378234398782344e-05, "val_loss": 3.06272292137146, "within_bucket_gap": -0.4314267635345459}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.55517043987547e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.55517043987547e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.322408199310303, "val_fraction": 2.378234398782344e-05, "val_loss": 6.237766265869141, "within_bucket_gap": 1.915358066558838}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002571005224879171, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002571005224879171, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.924978892008464, "val_fraction": 4.756468797564688e-05, "val_loss": 10.330260276794434, "within_bucket_gap": 5.40528138478597}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011000227710427759, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011000227710427759, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.392256736755371, "val_fraction": 2.378234398782344e-05, "val_loss": 10.017632484436035, "within_bucket_gap": 4.625375747680664}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.877811137218272e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.877811137218272e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.269705772399902, "val_fraction": 4.756468797564688e-05, "val_loss": 5.874736785888672, "within_bucket_gap": 0.6050310134887695}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010760998866510899, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010760998866510899, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.279590368270874, "val_fraction": 2.378234398782344e-05, "val_loss": 6.804375171661377, "within_bucket_gap": 4.524784803390503}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015789847395735788, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015789847395735788, "step": 800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.221732711791992, "val_fraction": 2.378234398782344e-05, "val_loss": 11.861047744750977, "within_bucket_gap": 6.639315032958985}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.7894807098663014e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.7894807098663014e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.977899491786957, "val_fraction": 4.756468797564688e-05, "val_loss": 4.984839916229248, "within_bucket_gap": 1.0069404244422913}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00036449301732729554, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00036449301732729554, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.47998332977295, "val_fraction": 7.134703196347032e-05, "val_loss": 6.371249198913574, "within_bucket_gap": -5.108734130859375}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.3076789379119873, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.695103645324707, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.159136389851388e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.159136389851388e-05, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.693374872207642, "val_fraction": 2.378234398782344e-05, "val_loss": 6.8626885414123535, "within_bucket_gap": 2.169313669204712}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.985992908477783, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.319504737854004, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.210086619291494e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.210086619291494e-05, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.613527297973633, "val_fraction": 2.378234398782344e-05, "val_loss": 5.96330451965332, "within_bucket_gap": 1.3497772216796875}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.158703088760376, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.5108877420425415, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.709254264831543, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003233105227736759, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003233105227736759, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.7414292097091675, "val_fraction": 4.756468797564688e-05, "val_loss": 8.53870964050293, "within_bucket_gap": 6.797280430793762}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.5952611622745044e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.5952611622745044e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.51895809173584, "val_fraction": 2.378234398782344e-05, "val_loss": 7.189733505249023, "within_bucket_gap": 0.6707754135131836}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014374720973148373, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014374720973148373, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.205986738204956, "val_fraction": 7.134703196347032e-05, "val_loss": 3.220747629801432, "within_bucket_gap": 2.014760891596476}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.277334213256836, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025090346144033166, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025090346144033166, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.68975830078125, "val_fraction": 0.00014269406392694063, "val_loss": 3.9314268430074057, "within_bucket_gap": -1.7583314577738443}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.340438698827163e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.340438698827163e-05, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.72803791364034, "val_fraction": 4.756468797564688e-05, "val_loss": 7.691771745681763, "within_bucket_gap": 1.9637338320414228}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.757903943867444e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.757903943867444e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.874868392944336, "val_fraction": 2.378234398782344e-05, "val_loss": 7.295951843261719, "within_bucket_gap": 2.421083450317383}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001541336819733659, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001541336819733659, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.65277099609375, "val_fraction": 2.378234398782344e-05, "val_loss": 3.171757936477661, "within_bucket_gap": -6.481013059616089}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.2934071832595893e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.2934071832595893e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.504895210266113, "val_fraction": 2.378234398782344e-05, "val_loss": 5.961043357849121, "within_bucket_gap": -0.5438518524169922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.105298519134521, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.3060684204101562, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.306843177005762e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.306843177005762e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.132511138916016, "val_fraction": 4.756468797564688e-05, "val_loss": 5.407261848449707, "within_bucket_gap": 0.2747507095336914}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.658666610717773, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.9938262701034546, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6151340007781982, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.135488510131836, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.398907836713747e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.398907836713747e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.6844210624694824, "val_fraction": 2.378234398782344e-05, "val_loss": 4.534073829650879, "within_bucket_gap": 1.8496527671813965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.584529399871826, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.0706264972686768, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.5608023405075073, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7032573223114014, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.5465245297691774e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.5465245297691774e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.050136566162109, "val_fraction": 4.756468797564688e-05, "val_loss": 5.5147552490234375, "within_bucket_gap": -0.5353813171386719}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.18142032623291, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.053520560264587, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.402185440063477, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.341986369506046e-06, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.341986369506046e-06, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.850250244140625, "val_fraction": 2.378234398782344e-05, "val_loss": 4.583582401275635, "within_bucket_gap": -0.26666784286499023}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.931208757867914e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.931208757867914e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.7453813552856445, "val_fraction": 2.378234398782344e-05, "val_loss": 9.239336013793945, "within_bucket_gap": 2.493954658508301}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.811722167187811e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.811722167187811e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.765984058380127, "val_fraction": 2.378234398782344e-05, "val_loss": 4.742751121520996, "within_bucket_gap": -2.023232936859131}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.4628400802612305, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.877050399780273, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.943204879760742, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001261434553239262, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001261434553239262, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.402056694030762, "val_fraction": 4.756468797564688e-05, "val_loss": 11.054096698760986, "within_bucket_gap": 2.6520400047302246}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.10848110914230347, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.792840003967285, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.276094913482666, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.6413116455078125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.779868125915527, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.3983986876689494e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.3983986876689494e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.6399085521698, "val_fraction": 2.378234398782344e-05, "val_loss": 2.3699898719787598, "within_bucket_gap": -2.26991868019104}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.545042991638184, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.985953688621521, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.68351885082152e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.68351885082152e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.5562729835510254, "val_fraction": 2.378234398782344e-05, "val_loss": 1.0074269771575928, "within_bucket_gap": -1.5488460063934326}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.319167315143428e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.319167315143428e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.323356866836548, "val_fraction": 2.378234398782344e-05, "val_loss": 3.7190003395080566, "within_bucket_gap": 1.3956434726715088}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.764409065246582, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.672867298126221, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.902940988540649, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.404419898986816, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.631697654724121, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.83525276184082, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.761362075805664, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.1136990658587335e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.1136990658587335e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.0726444721221924, "val_fraction": 2.378234398782344e-05, "val_loss": 2.5409326553344727, "within_bucket_gap": 0.4682881832122803}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.960627555847168, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.057509201665266e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.057509201665266e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.260769844055176, "val_fraction": 2.378234398782344e-05, "val_loss": 4.872748374938965, "within_bucket_gap": -3.388021469116211}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018869816240080111, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018869816240080111, "step": 800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 8.939491907755533, "val_fraction": 2.378234398782344e-05, "val_loss": 1.005111575126648, "within_bucket_gap": -7.934380332628885}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.965069652873813e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.965069652873813e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.534615993499756, "val_fraction": 2.378234398782344e-05, "val_loss": 6.360888481140137, "within_bucket_gap": 0.8262724876403809}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.222186088562012, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.391378532987388e-06, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.391378532987388e-06, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.015239953994751, "val_fraction": 2.378234398782344e-05, "val_loss": 8.326032638549805, "within_bucket_gap": 0.3107926845550537}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.435507774353027, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.0494630336761475, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.0354691743850708, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.040955543518066, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.657990455627441, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.257180690765381, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.464332580566406, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.006852429360151291, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.851539611816406, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.247470855712891, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.998917659849518e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.998917659849518e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.906446933746338, "val_fraction": 2.378234398782344e-05, "val_loss": 5.167431831359863, "within_bucket_gap": 1.2609848976135254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.326689720153809, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.21978759765625, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.2319230246308005e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.2319230246308005e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.2623948454856873, "val_fraction": 2.378234398782344e-05, "val_loss": 2.7803938388824463, "within_bucket_gap": 0.517998993396759}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002179753639178189, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002179753639178189, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.8481035828590393, "val_fraction": 2.378234398782344e-05, "val_loss": 10.013531684875488, "within_bucket_gap": 9.165428102016449}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.260066509246826, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.111320495605469, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.005541558377444744, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010917864773915782, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010917864773915782, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.116799592971802, "val_fraction": 2.378234398782344e-05, "val_loss": 9.70754337310791, "within_bucket_gap": 4.590743780136108}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011222154650513984, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011222154650513984, "step": 800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.015476584434509, "val_fraction": 4.756468797564688e-05, "val_loss": 3.656130790710449, "within_bucket_gap": -2.35934579372406}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.777128219604492, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.637845993041992, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.5187458992004395, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.1825850009918213, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.07458622008562088, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.002915301127359271, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.0807905197143555, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.022564679384231567, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.2815667704723014e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.2815667704723014e-05, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.420116424560547, "val_fraction": 2.378234398782344e-05, "val_loss": 3.8812432289123535, "within_bucket_gap": -0.5388731956481934}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.9595701694488525, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.154397964477539, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.455150586647893e-06, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.455150586647893e-06, "step": 800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.8685606718063354, "val_fraction": 2.378234398782344e-05, "val_loss": 0.971794843673706, "within_bucket_gap": 0.1032341718673706}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.6841439142755175e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.6841439142755175e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5369900465011597, "val_fraction": 4.756468797564688e-05, "val_loss": 0.7624356299638748, "within_bucket_gap": -0.7745544165372849}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.01160358439116e-05, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.01160358439116e-05, "step": 800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.079689025878906, "val_fraction": 9.512937595129375e-05, "val_loss": 5.711628794670105, "within_bucket_gap": 0.6319397687911987}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.019719393183672308, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.019719393183672308, "step": 900, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 5.800204419461346, "val_fraction": 0.05995528919330289, "val_loss": 5.471302775436334, "within_bucket_gap": -0.3289016440250112}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.002320202211007186, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.002320202211007186, "step": 900, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 5.8428450434430985, "val_fraction": 0.019882039573820395, "val_loss": 5.726146643241627, "within_bucket_gap": -0.11669840020147149}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0033018818312631794, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0033018818312631794, "step": 900, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 5.782542146381579, "val_fraction": 0.011582001522070015, "val_loss": 5.4974548214514884, "within_bucket_gap": -0.2850873249300907}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0015374879206244588, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0015374879206244588, "step": 900, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.643861576548794, "val_fraction": 0.007301179604261796, "val_loss": 5.433280820566979, "within_bucket_gap": -0.21058075598181514}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012978965429568203, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012978965429568203, "step": 900, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 5.662417267005481, "val_fraction": 0.004827815829528158, "val_loss": 5.689300988108067, "within_bucket_gap": 0.02688372110258541}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001321523122139832, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001321523122139832, "step": 900, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 5.921971097905585, "val_fraction": 0.003305745814307458, "val_loss": 5.52220559977799, "within_bucket_gap": -0.39976549812759465}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0012225078492448158, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0012225078492448158, "step": 900, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 5.8074337292268785, "val_fraction": 0.002877663622526636, "val_loss": 5.3826071999289775, "within_bucket_gap": -0.424826529297901}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.7291909388082115e-05, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.7291909388082115e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.4607037135532925, "val_fraction": 0.002449581430745814, "val_loss": 5.4373152649518355, "within_bucket_gap": -0.023388448601457057}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004358312124344672, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004358312124344672, "step": 900, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 5.389746863562782, "val_fraction": 0.0025684931506849314, "val_loss": 5.220063244854963, "within_bucket_gap": -0.16968361870781923}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00043864598705403246, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00043864598705403246, "step": 900, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 5.247857074348294, "val_fraction": 0.0014745053272450532, "val_loss": 5.545343952794229, "within_bucket_gap": 0.2974868784459348}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008698304046823665, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008698304046823665, "step": 900, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 5.666942439667166, "val_fraction": 0.0020214992389649923, "val_loss": 5.236652688419118, "within_bucket_gap": -0.4302897512480488}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005708508497161074, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005708508497161074, "step": 900, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 5.0953174167209205, "val_fraction": 0.0016647640791476407, "val_loss": 4.752415466308594, "within_bucket_gap": -0.34290195041232696}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006194330924668792, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006194330924668792, "step": 900, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.35451395371381, "val_fraction": 0.0018074581430745813, "val_loss": 5.697223462556538, "within_bucket_gap": 0.3427095088427281}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.254358229294281e-05, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.254358229294281e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.12735860487994, "val_fraction": 0.0015220700152207, "val_loss": 5.086267471313477, "within_bucket_gap": -0.041091133566463434}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007516290523287385, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007516290523287385, "step": 900, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 6.069217412798237, "val_fraction": 0.0014745053272450532, "val_loss": 5.5594674387285785, "within_bucket_gap": -0.509749974069658}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005953297404575206, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005953297404575206, "step": 900, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 5.398418515227561, "val_fraction": 0.0010226407914764079, "val_loss": 5.980567932128906, "within_bucket_gap": 0.5821494169013448}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00026906035625045514, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00026906035625045514, "step": 900, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.779170989990234, "val_fraction": 0.0004280821917808219, "val_loss": 5.150645997789171, "within_bucket_gap": -0.6285249922010632}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018140540098380182, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018140540098380182, "step": 900, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.438770759396437, "val_fraction": 0.0007372526636225266, "val_loss": 5.19271481421686, "within_bucket_gap": -0.2460559451795774}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.004889637627136e-05, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.004889637627136e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.727251461574009, "val_fraction": 0.000594558599695586, "val_loss": 5.828248901367187, "within_bucket_gap": 0.10099743979317832}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.3384937817424685e-05, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.3384937817424685e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.132607680100661, "val_fraction": 0.0007372526636225266, "val_loss": 6.1008886521862395, "within_bucket_gap": -0.031719027914421716}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004158253907841281, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004158253907841281, "step": 900, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.588043828164378, "val_fraction": 0.0006183409436834094, "val_loss": 6.2605294447678785, "within_bucket_gap": 0.6724856166035007}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.066609735358276e-05, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.066609735358276e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.822994232177734, "val_fraction": 0.0006659056316590563, "val_loss": 5.914097377232143, "within_bucket_gap": 0.09110314505440886}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006647446508912008, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006647446508912008, "step": 900, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 4.538966953754425, "val_fraction": 0.00047564687975646877, "val_loss": 5.936526107788086, "within_bucket_gap": 1.3975591540336607}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.091220986353213e-05, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.091220986353213e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.435452423095703, "val_fraction": 0.0003091704718417047, "val_loss": 4.535436776968149, "within_bucket_gap": 0.09998435387244609}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00030011210487728226, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00030011210487728226, "step": 900, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.329213414873395, "val_fraction": 0.000594558599695586, "val_loss": 4.833977966308594, "within_bucket_gap": 0.5047645514351986}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003523612947768814, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003523612947768814, "step": 900, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.768448193868001, "val_fraction": 0.0004280821917808219, "val_loss": 5.945332209269206, "within_bucket_gap": -0.823115984598795}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.932415007890283e-05, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.932415007890283e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 5.192148358214135, "val_fraction": 0.000594558599695586, "val_loss": 5.3087460327148435, "within_bucket_gap": 0.11659767450070824}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.5301299939037465e-05, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.5301299939037465e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.643380251797763, "val_fraction": 0.00045186453576864533, "val_loss": 5.765764939157586, "within_bucket_gap": 0.12238468735982355}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0011225619073550236, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0011225619073550236, "step": 900, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.5227875248078377, "val_fraction": 0.00040429984779299846, "val_loss": 5.299345353070428, "within_bucket_gap": 2.77655782826259}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00028832120458305185, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00028832120458305185, "step": 900, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.327390882703993, "val_fraction": 0.00033295281582952814, "val_loss": 6.193343026297433, "within_bucket_gap": 0.8659521435934403}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00027775364436061765, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00027775364436061765, "step": 900, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.497918009757996, "val_fraction": 0.0002616057838660578, "val_loss": 5.436192079023882, "within_bucket_gap": -1.0617259307341138}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0011030653749608196, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0011030653749608196, "step": 900, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.816695213317871, "val_fraction": 0.0005232115677321156, "val_loss": 4.708436445756392, "within_bucket_gap": -2.1082587675614795}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00037253460927641, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00037253460927641, "step": 900, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.4081878662109375, "val_fraction": 0.0002616057838660578, "val_loss": 4.984157388860529, "within_bucket_gap": -1.4240304773504082}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.934741640986069e-05, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.934741640986069e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.5405623118082685, "val_fraction": 0.0002616057838660578, "val_loss": 5.767420508644798, "within_bucket_gap": 0.2268581968365293}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00024572901620531013, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00024572901620531013, "step": 900, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.047077655792236, "val_fraction": 0.00014269406392694063, "val_loss": 6.76914660135905, "within_bucket_gap": 1.7220689455668134}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015126045685561997, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015126045685561997, "step": 900, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.2789366245269775, "val_fraction": 0.0002616057838660578, "val_loss": 6.700736652721059, "within_bucket_gap": -0.578199971805919}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.076405665562114e-05, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.076405665562114e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.3234244126539965, "val_fraction": 0.00028538812785388126, "val_loss": 5.536341667175293, "within_bucket_gap": 0.2129172545212965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004613790519342757, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004613790519342757, "step": 900, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.740974426269531, "val_fraction": 0.0003567351598173516, "val_loss": 5.4476366678873696, "within_bucket_gap": -1.2933377583821617}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002802258772271831, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002802258772271831, "step": 900, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.354968706766765, "val_fraction": 0.0002616057838660578, "val_loss": 6.426144860007546, "within_bucket_gap": 1.0711761532407813}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.564990917214784e-05, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.564990917214784e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.386888590725985, "val_fraction": 9.512937595129375e-05, "val_loss": 4.9718804359436035, "within_bucket_gap": 0.5849918452176182}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00033418747932399245, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00033418747932399245, "step": 900, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.984774169921875, "val_fraction": 0.00021404109589041095, "val_loss": 6.546098073323567, "within_bucket_gap": 1.5613239034016928}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.705581700696513e-05, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.705581700696513e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 4.380034516900133, "val_fraction": 0.00014269406392694063, "val_loss": 4.990121682484944, "within_bucket_gap": 0.6100871655848117}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004736507916144636, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004736507916144636, "step": 900, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.6831673213413785, "val_fraction": 0.00023782343987823439, "val_loss": 5.674774169921875, "within_bucket_gap": 1.9916068485804965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007134798795120902, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007134798795120902, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.120561122894287, "val_fraction": 0.0002616057838660578, "val_loss": 4.847870393232866, "within_bucket_gap": 2.727309270338579}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011284335623653581, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011284335623653581, "step": 900, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 6.177744547526042, "val_fraction": 0.00016647640791476407, "val_loss": 5.499910627092634, "within_bucket_gap": -0.6778339204334083}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006932851204951967, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006932851204951967, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.378364562988281, "val_fraction": 0.000380517503805175, "val_loss": 4.556411266326904, "within_bucket_gap": -1.821953296661377}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011963373449839412, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011963373449839412, "step": 900, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.914804458618164, "val_fraction": 7.134703196347032e-05, "val_loss": 7.591590881347656, "within_bucket_gap": 1.6767864227294922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.280613173451416e-05, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.280613173451416e-05, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.2043304443359375, "val_fraction": 4.756468797564688e-05, "val_loss": 4.094134330749512, "within_bucket_gap": -1.1101961135864258}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010869795607990468, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010869795607990468, "step": 900, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 2.607869003758286, "val_fraction": 4.756468797564688e-05, "val_loss": 4.893134832382202, "within_bucket_gap": 2.285265828623916}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.593513557256736e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.593513557256736e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.764294537630948, "val_fraction": 0.00016647640791476407, "val_loss": 5.188025883265904, "within_bucket_gap": -0.5762686543650446}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000356027184556064, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000356027184556064, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.216833114624023, "val_fraction": 7.134703196347032e-05, "val_loss": 3.2267560958862305, "within_bucket_gap": -4.990077018737793}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013980027883564499, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013980027883564499, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.287219524383545, "val_fraction": 4.756468797564688e-05, "val_loss": 1.3480584621429443, "within_bucket_gap": -2.9391610622406006}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000195119435189988, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000195119435189988, "step": 900, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.6610199723924906, "val_fraction": 0.00014269406392694063, "val_loss": 5.293622970581055, "within_bucket_gap": -1.3673970018114359}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.0201614538102753e-06, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.0201614538102753e-06, "step": 900, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 3.2962536811828613, "val_fraction": 2.378234398782344e-05, "val_loss": 3.423245429992676, "within_bucket_gap": 0.12699174880981445}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.711353970444915e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.711353970444915e-05, "step": 900, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.678514671325684, "val_fraction": 0.00016647640791476407, "val_loss": 5.455578940255301, "within_bucket_gap": -0.22293573107038256}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001981742169759045, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001981742169759045, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.855700492858887, "val_fraction": 0.00014269406392694063, "val_loss": 6.244505405426025, "within_bucket_gap": 1.3888049125671387}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017299582044873245, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017299582044873245, "step": 900, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.993606150150299, "val_fraction": 0.00011891171993911719, "val_loss": 3.5387804985046385, "within_bucket_gap": -1.4548256516456606}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020641684702270107, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020641684702270107, "step": 900, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.532522231340408, "val_fraction": 9.512937595129375e-05, "val_loss": 6.702376127243042, "within_bucket_gap": 2.1698538959026337}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007005077911959512, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007005077911959512, "step": 900, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.178798357645671, "val_fraction": 0.00023782343987823439, "val_loss": 7.124293518066406, "within_bucket_gap": 2.9454951604207356}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025040023765970405, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025040023765970405, "step": 900, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.156187438964844, "val_fraction": 0.00021404109589041095, "val_loss": 4.986317528618707, "within_bucket_gap": -1.1698699103461374}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010045351321657313, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010045351321657313, "step": 900, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.7453773498535154, "val_fraction": 0.00011891171993911719, "val_loss": 5.900603485107422, "within_bucket_gap": -0.8447738647460934}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010928467392377112, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010928467392377112, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.049923419952393, "val_fraction": 7.134703196347032e-05, "val_loss": 6.581657409667969, "within_bucket_gap": 1.5317339897155762}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.3462997848221839e-05, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.3462997848221839e-05, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.5289894342422485, "val_fraction": 0.00016647640791476407, "val_loss": 6.60985973903111, "within_bucket_gap": 0.0808703047888617}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.775667081442905e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.775667081442905e-05, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.137490590413411, "val_fraction": 0.0001902587519025875, "val_loss": 7.54617965221405, "within_bucket_gap": 0.40868906180063913}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.2559671652371586e-05, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.2559671652371586e-05, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.821288585662842, "val_fraction": 0.0002616057838660578, "val_loss": 5.773278669877485, "within_bucket_gap": -0.048009915785356405}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.29438603559404e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.29438603559404e-05, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.696588754653931, "val_fraction": 4.756468797564688e-05, "val_loss": 6.16301703453064, "within_bucket_gap": -1.533571720123291}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002226999236477746, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002226999236477746, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.952417612075806, "val_fraction": 7.134703196347032e-05, "val_loss": 4.831055482228597, "within_bucket_gap": -3.121362129847209}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.8664033264148542e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.8664033264148542e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.79525899887085, "val_fraction": 4.756468797564688e-05, "val_loss": 7.402866363525391, "within_bucket_gap": -0.392392635345459}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.5107322581101044e-05, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.5107322581101044e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.732614517211914, "val_fraction": 9.512937595129375e-05, "val_loss": 6.996542692184448, "within_bucket_gap": 0.2639281749725342}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001682124272147633, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001682124272147633, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.337984085083008, "val_fraction": 9.512937595129375e-05, "val_loss": 5.1062331199646, "within_bucket_gap": 1.7682490348815918}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001768359673440547, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001768359673440547, "step": 900, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.855709838867187, "val_fraction": 7.134703196347032e-05, "val_loss": 3.3771769205729165, "within_bucket_gap": -2.478532918294271}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.65077047043195e-06, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.65077047043195e-06, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.36335563659668, "val_fraction": 0.0001902587519025875, "val_loss": 4.317887187004089, "within_bucket_gap": -0.04546844959259033}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.895468115262244e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.895468115262244e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.062485694885254, "val_fraction": 2.378234398782344e-05, "val_loss": 2.163079261779785, "within_bucket_gap": -2.8994064331054688}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021423496408005282, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021423496408005282, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.2186119556427, "val_fraction": 7.134703196347032e-05, "val_loss": 5.221329212188721, "within_bucket_gap": 3.0027172565460205}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.678653415479618e-05, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.678653415479618e-05, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.2255702018737793, "val_fraction": 0.00011891171993911719, "val_loss": 3.8713142395019533, "within_bucket_gap": 0.645744037628174}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016732556997731756, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016732556997731756, "step": 900, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.026100158691406, "val_fraction": 7.134703196347032e-05, "val_loss": 7.371335347493489, "within_bucket_gap": 2.345235188802083}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014475809021808966, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014475809021808966, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.620556116104126, "val_fraction": 9.512937595129375e-05, "val_loss": 7.142253160476685, "within_bucket_gap": 1.5216970443725586}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.161462306976318, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.348135496746278e-07, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.348135496746278e-07, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.408079147338867, "val_fraction": 4.756468797564688e-05, "val_loss": 6.410913467407227, "within_bucket_gap": 0.002834320068359375}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002559014742051448, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002559014742051448, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.7445502281188965, "val_fraction": 4.756468797564688e-05, "val_loss": 2.3644776344299316, "within_bucket_gap": -5.380072593688965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.194107294082642, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020205496017972809, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020205496017972809, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.03807532787323, "val_fraction": 0.0001902587519025875, "val_loss": 3.976074457168579, "within_bucket_gap": -1.0620008707046509}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00024526297501777404, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00024526297501777404, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.433743476867676, "val_fraction": 7.134703196347032e-05, "val_loss": 6.996137619018555, "within_bucket_gap": -3.437605857849121}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022100583512670553, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022100583512670553, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.381669640541077, "val_fraction": 4.756468797564688e-05, "val_loss": 2.7352429628372192, "within_bucket_gap": -4.646426677703857}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.5127379801114856e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.5127379801114856e-05, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.570111910502116, "val_fraction": 9.512937595129375e-05, "val_loss": 4.8342509269714355, "within_bucket_gap": 0.26413901646931937}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014560017593012189, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014560017593012189, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.083382606506348, "val_fraction": 9.512937595129375e-05, "val_loss": 4.552833557128906, "within_bucket_gap": -1.5305490493774414}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.784928902824901e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.784928902824901e-05, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.141317844390869, "val_fraction": 2.378234398782344e-05, "val_loss": 2.8918447494506836, "within_bucket_gap": 0.7505269050598145}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.299772649958072e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.299772649958072e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.829603433609009, "val_fraction": 2.378234398782344e-05, "val_loss": 6.478531837463379, "within_bucket_gap": 2.64892840385437}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020001251826970966, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020001251826970966, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.730213801066081, "val_fraction": 4.756468797564688e-05, "val_loss": 8.935276985168457, "within_bucket_gap": 4.205063184102376}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011941567176734656, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011941567176734656, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.522703647613525, "val_fraction": 2.378234398782344e-05, "val_loss": 9.543893814086914, "within_bucket_gap": 5.021190166473389}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.0636487816385295e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.0636487816385295e-05, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.540306091308594, "val_fraction": 4.756468797564688e-05, "val_loss": 5.815127611160278, "within_bucket_gap": 1.2748215198516846}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010778146542189146, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010778146542189146, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.872265577316284, "val_fraction": 2.378234398782344e-05, "val_loss": 8.404260635375977, "within_bucket_gap": 4.531995058059692}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012228128300409882, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012228128300409882, "step": 900, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.658407402038574, "val_fraction": 2.378234398782344e-05, "val_loss": 9.800090789794922, "within_bucket_gap": 5.1416833877563475}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.0930473584925563e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.0930473584925563e-05, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.366898894309998, "val_fraction": 4.756468797564688e-05, "val_loss": 5.437661170959473, "within_bucket_gap": 1.070762276649475}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003732438206309779, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003732438206309779, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.403834342956543, "val_fraction": 7.134703196347032e-05, "val_loss": 5.172448952992757, "within_bucket_gap": -5.231385389963786}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.103710889816284, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.841464042663574, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.07153350692422e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.07153350692422e-05, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.898880084355672, "val_fraction": 2.378234398782344e-05, "val_loss": 7.451838493347168, "within_bucket_gap": 2.552958408991496}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.7845444679260254, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.084075927734375, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.7525381864115163e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.7525381864115163e-05, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.437485694885254, "val_fraction": 2.378234398782344e-05, "val_loss": 5.594872951507568, "within_bucket_gap": 1.1573872566223145}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.208432197570801, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.253644227981567, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.18336296081543, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003178139363204326, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003178139363204326, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5369725823402405, "val_fraction": 4.756468797564688e-05, "val_loss": 8.218692779541016, "within_bucket_gap": 6.681720197200775}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.4650557763318855e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.4650557763318855e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.479400634765625, "val_fraction": 2.378234398782344e-05, "val_loss": 6.442893981933594, "within_bucket_gap": -1.0365066528320312}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011397124788351074, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011397124788351074, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5401678085327148, "val_fraction": 7.134703196347032e-05, "val_loss": 3.1375888188680015, "within_bucket_gap": 1.5974210103352866}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.195010185241699, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000144330594728345, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000144330594728345, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.952300786972046, "val_fraction": 0.00014269406392694063, "val_loss": 3.940831979115804, "within_bucket_gap": -1.0114688078562417}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.802984364924243e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.802984364924243e-05, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.39606777826945, "val_fraction": 4.756468797564688e-05, "val_loss": 7.036567211151123, "within_bucket_gap": 1.6404994328816729}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.7395014327410693e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.7395014327410693e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.328178405761719, "val_fraction": 2.378234398782344e-05, "val_loss": 7.321043968200684, "within_bucket_gap": 1.9928655624389648}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014043534727401383, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014043534727401383, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.86279582977295, "val_fraction": 2.378234398782344e-05, "val_loss": 2.957770347595215, "within_bucket_gap": -5.905025482177734}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.8353882503654676e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.8353882503654676e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.196563720703125, "val_fraction": 2.378234398782344e-05, "val_loss": 6.163379669189453, "within_bucket_gap": -2.033184051513672}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.779484272003174, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.9824597835540771, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.5447662460930995e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.5447662460930995e-05, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.377274036407471, "val_fraction": 4.756468797564688e-05, "val_loss": 5.122525691986084, "within_bucket_gap": 0.7452516555786133}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.87484073638916, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.702993392944336, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.867281198501587, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.135146617889404, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.324291585424356e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.324291585424356e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.6076996326446533, "val_fraction": 2.378234398782344e-05, "val_loss": 5.107897758483887, "within_bucket_gap": 3.5001981258392334}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.2446746826171875, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.1597464084625244, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.005352258682251, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.964853763580322, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.4794546233283145e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.4794546233283145e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.230934143066406, "val_fraction": 4.756468797564688e-05, "val_loss": 5.289173603057861, "within_bucket_gap": -0.9417605400085449}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.671810626983643, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.023693323135376, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.9070608615875244, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.884553010060907e-06, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.884553010060907e-06, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.919550895690918, "val_fraction": 2.378234398782344e-05, "val_loss": 5.124936580657959, "within_bucket_gap": 0.20538568496704102}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.34604960090312e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.34604960090312e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.048330783843994, "val_fraction": 2.378234398782344e-05, "val_loss": 8.716717720031738, "within_bucket_gap": 2.668386936187744}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.011631398803386e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.011631398803386e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.11910343170166, "val_fraction": 2.378234398782344e-05, "val_loss": 4.1708526611328125, "within_bucket_gap": -2.9482507705688477}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.406744956970215, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.248169422149658, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.8630852699279785, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003094423216408972, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003094423216408972, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.625503063201904, "val_fraction": 4.756468797564688e-05, "val_loss": 11.131218433380127, "within_bucket_gap": 6.505715370178223}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.11169151961803436, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.336880683898926, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.9267959594726562, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.574862480163574, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.158163070678711, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.232409504632064e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.232409504632064e-05, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.90043306350708, "val_fraction": 2.378234398782344e-05, "val_loss": 2.7003095149993896, "within_bucket_gap": -2.2001235485076904}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.985745429992676, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.5674097239971161, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0415622496713785e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0415622496713785e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7809770107269287, "val_fraction": 2.378234398782344e-05, "val_loss": 2.3430209159851074, "within_bucket_gap": -0.4379560947418213}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.012381468007916e-06, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.012381468007916e-06, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.460299015045166, "val_fraction": 2.378234398782344e-05, "val_loss": 3.629011631011963, "within_bucket_gap": 0.16871261596679688}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.9800693988800049, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.561126708984375, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.075603246688843, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.267484664916992, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.660909652709961, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.228055000305176, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.185246229171753, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.063799638363506e-06, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.063799638363506e-06, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.8338013887405396, "val_fraction": 2.378234398782344e-05, "val_loss": 2.1308200359344482, "within_bucket_gap": 0.2970186471939087}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.4678955078125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.128996218357637e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.128996218357637e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.429851531982422, "val_fraction": 2.378234398782344e-05, "val_loss": 4.011771202087402, "within_bucket_gap": -3.4180803298950195}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016385137010412976, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016385137010412976, "step": 900, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 8.724252065022787, "val_fraction": 2.378234398782344e-05, "val_loss": 1.8346296548843384, "within_bucket_gap": -6.889622410138449}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.526186412509355e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.526186412509355e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.361428260803223, "val_fraction": 2.378234398782344e-05, "val_loss": 7.526039123535156, "within_bucket_gap": 3.1646108627319336}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.99717903137207, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.3273850060670524e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.3273850060670524e-05, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.143860816955566, "val_fraction": 2.378234398782344e-05, "val_loss": 7.701999664306641, "within_bucket_gap": 0.5581388473510742}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.560367584228516, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.0651464462280273, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.1878163814544678, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.592726707458496, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.052376747131348, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.1482839584350586, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.409711837768555, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.016496209427714348, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.558761119842529, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.041258811950684, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.123106569459993e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.123106569459993e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.109525442123413, "val_fraction": 2.378234398782344e-05, "val_loss": 5.684169292449951, "within_bucket_gap": 2.574643850326538}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.037784576416016, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.968471527099609, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0705029923621921e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0705029923621921e-05, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.2717865705490112, "val_fraction": 2.378234398782344e-05, "val_loss": 1.8216614723205566, "within_bucket_gap": -0.4501250982284546}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014593682990408016, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014593682990408016, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.204939603805542, "val_fraction": 2.378234398782344e-05, "val_loss": 7.341291427612305, "within_bucket_gap": 6.136351823806763}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.403052806854248, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.723232269287109, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.02025269716978073, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001179838845083884, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001179838845083884, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.257742643356323, "val_fraction": 2.378234398782344e-05, "val_loss": 9.218729019165039, "within_bucket_gap": 4.960986375808716}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015885786605570777, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015885786605570777, "step": 900, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.531481027603149, "val_fraction": 4.756468797564688e-05, "val_loss": 2.191653251647949, "within_bucket_gap": -3.3398277759552}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.697906494140625, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.390118598937988, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.855778455734253, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.46151483058929443, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.08970185369253159, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.05696175992488861, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.849730491638184, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.0002821285743266344, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.8985871129202877e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.8985871129202877e-05, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.38748836517334, "val_fraction": 2.378234398782344e-05, "val_loss": 4.1858062744140625, "within_bucket_gap": 0.7983179092407227}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.662900924682617, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.092049598693848, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.6076945312763458e-06, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.6076945312763458e-06, "step": 900, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.3483261168003082, "val_fraction": 2.378234398782344e-05, "val_loss": 0.415926456451416, "within_bucket_gap": 0.06760033965110779}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.1691603692336345e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.1691603692336345e-05, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.3907312154769897, "val_fraction": 4.756468797564688e-05, "val_loss": 0.3039669394493103, "within_bucket_gap": -1.0867642760276794}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001281837463560351, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001281837463560351, "step": 900, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.638167858123779, "val_fraction": 9.512937595129375e-05, "val_loss": 4.290700316429138, "within_bucket_gap": -1.3474675416946411}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.020313267114270808, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.020313267114270808, "step": 1000, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 5.698512157934196, "val_fraction": 0.05995528919330289, "val_loss": 5.359705233848175, "within_bucket_gap": -0.338806924086021}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0021257797943214566, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0021257797943214566, "step": 1000, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 5.702196066792468, "val_fraction": 0.019882039573820395, "val_loss": 5.595276462974731, "within_bucket_gap": -0.10691960381773757}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.003033006077229875, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.003033006077229875, "step": 1000, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 5.650175267269737, "val_fraction": 0.011582001522070015, "val_loss": 5.388302906827516, "within_bucket_gap": -0.2618723604422213}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0012782542389567087, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0012782542389567087, "step": 1000, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.498554002192982, "val_fraction": 0.007301179604261796, "val_loss": 5.323478972096397, "within_bucket_gap": -0.17507503009658532}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012108344231936963, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012108344231936963, "step": 1000, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 5.557561904348004, "val_fraction": 0.004827815829528158, "val_loss": 5.532481527093596, "within_bucket_gap": -0.025080377254408148}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0010695099870519594, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0010695099870519594, "step": 1000, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 5.726475059563387, "val_fraction": 0.003305745814307458, "val_loss": 5.402944441321942, "within_bucket_gap": -0.3235306182414446}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001641127344155754, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001641127344155754, "step": 1000, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 5.781242591108201, "val_fraction": 0.002877663622526636, "val_loss": 5.210944057496126, "within_bucket_gap": -0.5702985336120756}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002070293591887735, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002070293591887735, "step": 1000, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.400354930332729, "val_fraction": 0.002449581430745814, "val_loss": 5.315838711933025, "within_bucket_gap": -0.08451621839970436}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004531659761789995, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004531659761789995, "step": 1000, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 5.2750431095157655, "val_fraction": 0.0025684931506849314, "val_loss": 5.098610489456742, "within_bucket_gap": -0.17643262005902383}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000725487436596658, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000725487436596658, "step": 1000, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 5.054312414052535, "val_fraction": 0.0014745053272450532, "val_loss": 5.546333312988281, "within_bucket_gap": 0.4920208989357464}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006672006560009255, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006672006560009255, "step": 1000, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 5.5193807523544525, "val_fraction": 0.0020214992389649923, "val_loss": 5.189328361960018, "within_bucket_gap": -0.33005239039443435}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000622191728320639, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000622191728320639, "step": 1000, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 4.964477199978298, "val_fraction": 0.0016647640791476407, "val_loss": 4.590735517229352, "within_bucket_gap": -0.37374168274894615}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008593909752615479, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008593909752615479, "step": 1000, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 5.00049041299259, "val_fraction": 0.0018074581430745813, "val_loss": 5.475959777832031, "within_bucket_gap": 0.4754693648394417}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004056119931631114, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004056119931631114, "step": 1000, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 4.871058744542739, "val_fraction": 0.0015220700152207, "val_loss": 5.137545824050903, "within_bucket_gap": 0.2664870795081642}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006266055157740576, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006266055157740576, "step": 1000, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 5.843561069599326, "val_fraction": 0.0014745053272450532, "val_loss": 5.4186012514175905, "within_bucket_gap": -0.4249598181817351}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005549214201975085, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005549214201975085, "step": 1000, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 5.178483297658521, "val_fraction": 0.0010226407914764079, "val_loss": 5.721119015715843, "within_bucket_gap": 0.5426357180573218}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022403433860772217, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022403433860772217, "step": 1000, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.574662913446841, "val_fraction": 0.0004280821917808219, "val_loss": 5.051318698459202, "within_bucket_gap": -0.5233442149876391}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00024927956014335047, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00024927956014335047, "step": 1000, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.406533590177211, "val_fraction": 0.0007372526636225266, "val_loss": 5.0684140113092235, "within_bucket_gap": -0.33811957886798716}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002899585949075981, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002899585949075981, "step": 1000, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.4374574933733255, "val_fraction": 0.000594558599695586, "val_loss": 5.925144653320313, "within_bucket_gap": 0.4876871599469874}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001326978395058055, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001326978395058055, "step": 1000, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 6.059747842641977, "val_fraction": 0.0007372526636225266, "val_loss": 6.239737479917465, "within_bucket_gap": 0.17998963727548745}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003671252616885694, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003671252616885694, "step": 1000, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.586053663684476, "val_fraction": 0.0006183409436834094, "val_loss": 6.1797799330491285, "within_bucket_gap": 0.5937262693646526}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.6245367581441773e-05, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.6245367581441773e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.793866430010114, "val_fraction": 0.0006659056316590563, "val_loss": 5.833279473440988, "within_bucket_gap": 0.03941304343087371}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000594864804949753, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000594864804949753, "step": 1000, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 4.3361411690711975, "val_fraction": 0.00047564687975646877, "val_loss": 5.586784934997558, "within_bucket_gap": 1.2506437659263607}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.0107559700534929e-05, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.0107559700534929e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.596224594116211, "val_fraction": 0.0003091704718417047, "val_loss": 4.628917107215295, "within_bucket_gap": 0.03269251309908405}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003944767327587559, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003944767327587559, "step": 1000, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.21574456351144, "val_fraction": 0.000594558599695586, "val_loss": 4.879222869873047, "within_bucket_gap": 0.6634783063616068}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001651539922304896, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001651539922304896, "step": 1000, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.399255434672038, "val_fraction": 0.0004280821917808219, "val_loss": 6.013455708821614, "within_bucket_gap": -0.38579972585042377}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00045176734090239166, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00045176734090239166, "step": 1000, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 4.455508475210152, "val_fraction": 0.000594558599695586, "val_loss": 5.215345001220703, "within_bucket_gap": 0.7598365260105506}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.25725651618699e-05, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.25725651618699e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.406778162175959, "val_fraction": 0.00045186453576864533, "val_loss": 5.478862963224712, "within_bucket_gap": 0.07208480104875292}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009308429982248153, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009308429982248153, "step": 1000, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.5946507607736895, "val_fraction": 0.00040429984779299846, "val_loss": 4.897008783677045, "within_bucket_gap": 2.302358022903355}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003402862535925376, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003402862535925376, "step": 1000, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.870942857530382, "val_fraction": 0.00033295281582952814, "val_loss": 5.892968314034598, "within_bucket_gap": 1.022025456504216}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002630380814387556, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002630380814387556, "step": 1000, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.305841684341431, "val_fraction": 0.0002616057838660578, "val_loss": 5.300366661765358, "within_bucket_gap": -1.0054750225760722}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0011026986748479092, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0011026986748479092, "step": 1000, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.695660591125488, "val_fraction": 0.0005232115677321156, "val_loss": 4.588102687488902, "within_bucket_gap": -2.107557903636586}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003569153383684182, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003569153383684182, "step": 1000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.174418290456136, "val_fraction": 0.0002616057838660578, "val_loss": 4.810093186118386, "within_bucket_gap": -1.36432510433775}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.6191322206342076e-05, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.6191322206342076e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.580160564846462, "val_fraction": 0.0002616057838660578, "val_loss": 5.680278084494851, "within_bucket_gap": 0.10011751964838833}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002571148188328271, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002571148188328271, "step": 1000, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 4.763406753540039, "val_fraction": 0.00014269406392694063, "val_loss": 6.565267403920491, "within_bucket_gap": 1.8018606503804522}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010727903508462631, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010727903508462631, "step": 1000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.157857100168864, "val_fraction": 0.0002616057838660578, "val_loss": 6.7477781122381035, "within_bucket_gap": -0.4100789879307607}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.846827639557563e-05, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.846827639557563e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.172284052922175, "val_fraction": 0.00028538812785388126, "val_loss": 5.377156893412272, "within_bucket_gap": 0.20487284049009702}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005241715926195024, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005241715926195024, "step": 1000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.82766850789388, "val_fraction": 0.0003567351598173516, "val_loss": 5.358310699462891, "within_bucket_gap": -1.4693578084309893}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016084823706378684, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016084823706378684, "step": 1000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.59744930267334, "val_fraction": 0.0002616057838660578, "val_loss": 6.212299000133168, "within_bucket_gap": 0.6148496974598281}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.041486793556071e-05, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.041486793556071e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.109236803921786, "val_fraction": 9.512937595129375e-05, "val_loss": 4.9545578956604, "within_bucket_gap": 0.8453210917386142}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00034006381143718, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00034006381143718, "step": 1000, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.948466033935547, "val_fraction": 0.00021404109589041095, "val_loss": 6.537244160970052, "within_bucket_gap": 1.5887781270345052}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.119812421897303e-05, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.119812421897303e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 4.197937986035368, "val_fraction": 0.00014269406392694063, "val_loss": 4.486654440561931, "within_bucket_gap": 0.288716454526563}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004119418462635926, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004119418462635926, "step": 1000, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.55688898903983, "val_fraction": 0.00023782343987823439, "val_loss": 5.289022064208984, "within_bucket_gap": 1.7321330751691542}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005761702435836762, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005761702435836762, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.3633506298065186, "val_fraction": 0.0002616057838660578, "val_loss": 4.5657875754616475, "within_bucket_gap": 2.202436945655129}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.9291237507418024e-05, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.9291237507418024e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.712684631347656, "val_fraction": 0.00016647640791476407, "val_loss": 5.536736352103097, "within_bucket_gap": -0.17594827924455902}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004988995861245072, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004988995861245072, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.576043128967285, "val_fraction": 0.000380517503805175, "val_loss": 4.26493501663208, "within_bucket_gap": -1.311108112335205}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.997537354330155e-05, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.997537354330155e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.728400366646903, "val_fraction": 7.134703196347032e-05, "val_loss": 7.129655202229817, "within_bucket_gap": 1.4012548355829146}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.2428233070097734e-05, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.2428233070097734e-05, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.161498665809631, "val_fraction": 4.756468797564688e-05, "val_loss": 4.0592474937438965, "within_bucket_gap": -1.1022511720657349}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012159469150363131, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012159469150363131, "step": 1000, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 2.0124835390033144, "val_fraction": 4.756468797564688e-05, "val_loss": 4.568890333175659, "within_bucket_gap": 2.5564067941723447}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.797332040069669e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.797332040069669e-05, "step": 1000, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.7349443435668945, "val_fraction": 0.00016647640791476407, "val_loss": 5.146432604108538, "within_bucket_gap": -0.5885117394583563}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00039205866861561115, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00039205866861561115, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.370322227478027, "val_fraction": 7.134703196347032e-05, "val_loss": 2.875227928161621, "within_bucket_gap": -5.495094299316406}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013792302170300593, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013792302170300593, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.195031046867371, "val_fraction": 4.756468797564688e-05, "val_loss": 1.295337438583374, "within_bucket_gap": -2.8996936082839966}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001075149577165278, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001075149577165278, "step": 1000, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.332565750394549, "val_fraction": 0.00014269406392694063, "val_loss": 5.579100926717122, "within_bucket_gap": -0.7534648236774268}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.4531467240332104e-07, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.4531467240332104e-07, "step": 1000, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 3.076723575592041, "val_fraction": 2.378234398782344e-05, "val_loss": 3.082833766937256, "within_bucket_gap": 0.006110191345214844}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011365118364220878, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011365118364220878, "step": 1000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.748592948913574, "val_fraction": 0.00016647640791476407, "val_loss": 5.065906524658203, "within_bucket_gap": -0.6826864242553707}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017248945738808393, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017248945738808393, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.752447446187337, "val_fraction": 0.00014269406392694063, "val_loss": 5.961253563563029, "within_bucket_gap": 1.2088061173756923}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002885375404376054, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002885375404376054, "step": 1000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.124823808670044, "val_fraction": 0.00011891171993911719, "val_loss": 2.698338508605957, "within_bucket_gap": -2.426485300064087}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018629629896383851, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018629629896383851, "step": 1000, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.2958599627017975, "val_fraction": 9.512937595129375e-05, "val_loss": 6.254206657409668, "within_bucket_gap": 1.9583466947078705}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006931402065135815, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006931402065135815, "step": 1000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.011374791463216, "val_fraction": 0.00023782343987823439, "val_loss": 6.925890731811523, "within_bucket_gap": 2.9145159403483074}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002755429465295336, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002755429465295336, "step": 1000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.977011871337891, "val_fraction": 0.00021404109589041095, "val_loss": 4.68967522515191, "within_bucket_gap": -1.287336646185981}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010778184532211972, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010778184532211972, "step": 1000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.640202331542969, "val_fraction": 0.00011891171993911719, "val_loss": 5.7338001251220705, "within_bucket_gap": -0.9064022064208981}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.843110998290131e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.843110998290131e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.501215934753418, "val_fraction": 7.134703196347032e-05, "val_loss": 6.180026372273763, "within_bucket_gap": 0.6788104375203448}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.1480834386120145e-05, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.1480834386120145e-05, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.110431671142578, "val_fraction": 0.00016647640791476407, "val_loss": 6.179395403180804, "within_bucket_gap": 0.0689637320382257}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003052136351891667, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003052136351891667, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.112550735473633, "val_fraction": 0.0001902587519025875, "val_loss": 7.716753602027893, "within_bucket_gap": 1.6042028665542603}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001635077087665076, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001635077087665076, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.222280979156494, "val_fraction": 0.0002616057838660578, "val_loss": 5.84729662808505, "within_bucket_gap": 0.6250156489285557}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.309088143765291e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.309088143765291e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.467754244804382, "val_fraction": 4.756468797564688e-05, "val_loss": 5.1413315534591675, "within_bucket_gap": -1.3264226913452148}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002534795960743133, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002534795960743133, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.090163111686707, "val_fraction": 7.134703196347032e-05, "val_loss": 4.537393093109131, "within_bucket_gap": -3.5527700185775757}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.080124452233859e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.080124452233859e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.926633834838867, "val_fraction": 4.756468797564688e-05, "val_loss": 7.279068470001221, "within_bucket_gap": -0.6475653648376465}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012572440836164687, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012572440836164687, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.736157417297363, "val_fraction": 9.512937595129375e-05, "val_loss": 7.057772397994995, "within_bucket_gap": 1.3216149806976318}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00023059274780151505, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00023059274780151505, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.6641552448272705, "val_fraction": 9.512937595129375e-05, "val_loss": 5.088146209716797, "within_bucket_gap": 2.4239909648895264}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018421615186165822, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018421615186165822, "step": 1000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.585461807250977, "val_fraction": 7.134703196347032e-05, "val_loss": 3.003488222757975, "within_bucket_gap": -2.5819735844930016}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00024926923272090775, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00024926923272090775, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.993387460708618, "val_fraction": 0.0001902587519025875, "val_loss": 4.3035465478897095, "within_bucket_gap": 1.3101590871810913}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.48511352767683e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.48511352767683e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.929542064666748, "val_fraction": 2.378234398782344e-05, "val_loss": 1.7822015285491943, "within_bucket_gap": -3.1473405361175537}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002505100435680813, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002505100435680813, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.1165361404418945, "val_fraction": 7.134703196347032e-05, "val_loss": 5.627684911092122, "within_bucket_gap": 3.5111487706502276}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012237597341951167, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012237597341951167, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.881603479385376, "val_fraction": 0.00011891171993911719, "val_loss": 3.9107364654541015, "within_bucket_gap": 1.0291329860687255}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001919052281517598, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001919052281517598, "step": 1000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.9015960693359375, "val_fraction": 7.134703196347032e-05, "val_loss": 7.591339747111003, "within_bucket_gap": 2.6897436777750654}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.709779687849534e-05, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.709779687849534e-05, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.152045011520386, "val_fraction": 9.512937595129375e-05, "val_loss": 6.857377052307129, "within_bucket_gap": 0.7053320407867432}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.840099334716797, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.484795370784161e-06, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.484795370784161e-06, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.146705865859985, "val_fraction": 4.756468797564688e-05, "val_loss": 6.262018203735352, "within_bucket_gap": 0.11531233787536621}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002315818775825123, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002315818775825123, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.860406875610352, "val_fraction": 4.756468797564688e-05, "val_loss": 1.9916294813156128, "within_bucket_gap": -4.868777394294739}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.091156840324402, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00042806225552406485, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00042806225552406485, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.03115713596344, "val_fraction": 0.0001902587519025875, "val_loss": 3.781261920928955, "within_bucket_gap": -2.249895215034485}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00026907226357104387, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00026907226357104387, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.311872482299805, "val_fraction": 7.134703196347032e-05, "val_loss": 6.540555636088054, "within_bucket_gap": -3.771316846211751}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022505267127735973, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022505267127735973, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.3510777950286865, "val_fraction": 4.756468797564688e-05, "val_loss": 2.6195704340934753, "within_bucket_gap": -4.731507360935211}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.6148173604623635e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.6148173604623635e-05, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.467254002888997, "val_fraction": 9.512937595129375e-05, "val_loss": 4.742123603820801, "within_bucket_gap": 0.2748696009318037}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.5384729129901576e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.5384729129901576e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.58162260055542, "val_fraction": 9.512937595129375e-05, "val_loss": 5.3147783279418945, "within_bucket_gap": -0.2668442726135254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.3205275234384986e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.3205275234384986e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.4343528747558594, "val_fraction": 2.378234398782344e-05, "val_loss": 2.9896082878112793, "within_bucket_gap": 0.5552554130554199}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.7792026463164584e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.7792026463164584e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.458150863647461, "val_fraction": 2.378234398782344e-05, "val_loss": 3.6267499923706055, "within_bucket_gap": 1.1685991287231445}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018017538334136684, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018017538334136684, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.723789215087891, "val_fraction": 4.756468797564688e-05, "val_loss": 8.511796474456787, "within_bucket_gap": 3.7880072593688965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015797979755488703, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015797979755488703, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.8657779693603516, "val_fraction": 2.378234398782344e-05, "val_loss": 9.508512496948242, "within_bucket_gap": 6.642734527587891}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.613856642576475e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.613856642576475e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.66493558883667, "val_fraction": 4.756468797564688e-05, "val_loss": 6.475912809371948, "within_bucket_gap": 1.8109772205352783}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014456556585462852, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014456556585462852, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.371181011199951, "val_fraction": 2.378234398782344e-05, "val_loss": 8.449873924255371, "within_bucket_gap": 6.07869291305542}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013730277844578527, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013730277844578527, "step": 1000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.689332580566406, "val_fraction": 2.378234398782344e-05, "val_loss": 10.462639808654785, "within_bucket_gap": 5.773307228088379}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000117694406204572, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000117694406204572, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.5060296058654785, "val_fraction": 4.756468797564688e-05, "val_loss": 5.9804368019104, "within_bucket_gap": 2.474407196044922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00047842101328630606, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00047842101328630606, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.832433700561523, "val_fraction": 7.134703196347032e-05, "val_loss": 5.126884778340657, "within_bucket_gap": -6.705548922220866}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.4968690077463784, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.557470321655273, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.7321172567686416e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.7321172567686416e-05, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.416983564694722, "val_fraction": 2.378234398782344e-05, "val_loss": 5.986264228820801, "within_bucket_gap": 1.5692806641260786}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4755430221557617, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.532456398010254, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.582854454557282e-06, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.582854454557282e-06, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.422275066375732, "val_fraction": 2.378234398782344e-05, "val_loss": 4.699070930480957, "within_bucket_gap": 0.2767958641052246}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.974064588546753, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.622503042221069, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.530472755432129, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00027982451098218356, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00027982451098218356, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4849537760019302, "val_fraction": 4.756468797564688e-05, "val_loss": 7.367984294891357, "within_bucket_gap": 5.883030518889427}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.3636989930993347e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.3636989930993347e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.309192657470703, "val_fraction": 2.378234398782344e-05, "val_loss": 5.894824504852295, "within_bucket_gap": -1.4143681526184082}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012347768408159867, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012347768408159867, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.2933776378631592, "val_fraction": 7.134703196347032e-05, "val_loss": 3.024040857950846, "within_bucket_gap": 1.730663220087687}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.698462009429932, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012544781264467686, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012544781264467686, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.836491107940674, "val_fraction": 0.00014269406392694063, "val_loss": 3.9573528369267783, "within_bucket_gap": -0.8791382710138955}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.7324194733955e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.7324194733955e-05, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.719444751739502, "val_fraction": 4.756468797564688e-05, "val_loss": 6.134868621826172, "within_bucket_gap": 1.41542387008667}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.452449417912615e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.452449417912615e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.932826042175293, "val_fraction": 2.378234398782344e-05, "val_loss": 7.2254719734191895, "within_bucket_gap": 2.2926459312438965}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016899097886622586, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016899097886622586, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.542052268981934, "val_fraction": 2.378234398782344e-05, "val_loss": 2.436319589614868, "within_bucket_gap": -7.105732679367065}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.823962831424614e-07, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.823962831424614e-07, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.525315284729004, "val_fraction": 2.378234398782344e-05, "val_loss": 7.537189483642578, "within_bucket_gap": 0.011874198913574219}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.509821891784668, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.361161470413208, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.4352834750891093e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.4352834750891093e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.461669921875, "val_fraction": 4.756468797564688e-05, "val_loss": 4.763423919677734, "within_bucket_gap": 0.3017539978027344}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.7849555015563965, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.284977912902832, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7316410541534424, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.125590244928996, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.433313250360242e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.433313250360242e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.8600678443908691, "val_fraction": 2.378234398782344e-05, "val_loss": 4.565147399902344, "within_bucket_gap": 2.7050795555114746}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.8184142112731934, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.579141855239868, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.3103102743625641, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.0175089836120605, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.166440052166014e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.166440052166014e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.585406303405762, "val_fraction": 4.756468797564688e-05, "val_loss": 5.919693946838379, "within_bucket_gap": -0.6657123565673828}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.478853225708008, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.4365622997283936, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5144933462142944, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.744567683479739e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.744567683479739e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.329655647277832, "val_fraction": 2.378234398782344e-05, "val_loss": 2.7551398277282715, "within_bucket_gap": -1.5745158195495605}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.6736663628930915e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.6736663628930915e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.873812675476074, "val_fraction": 2.378234398782344e-05, "val_loss": 7.998035907745361, "within_bucket_gap": 1.124223232269287}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.552760747651168e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.552760747651168e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.157273292541504, "val_fraction": 2.378234398782344e-05, "val_loss": 3.5610084533691406, "within_bucket_gap": -3.5962648391723633}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.034982681274414, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.260396480560303, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.886946678161621, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003180316503370915, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003180316503370915, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.147945404052734, "val_fraction": 4.756468797564688e-05, "val_loss": 10.834242820739746, "within_bucket_gap": 6.686297416687012}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.38502267003059387, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.680224895477295, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.4819607734680176, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.245635986328125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.685606002807617, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.646106536121673e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.646106536121673e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.833530128002167, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4594552516937256, "within_bucket_gap": -2.374074876308441}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.545272827148438, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.17005512863397598, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.5840512704631508e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.5840512704631508e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.110492706298828, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4444308280944824, "within_bucket_gap": -0.6660618782043457}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.2772143919536873e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.2772143919536873e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.443307876586914, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4857847690582275, "within_bucket_gap": -0.9575231075286865}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.7364463806152344, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.944421768188477, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.947484970092773, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.606748580932617, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.208271026611328, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.235645294189453, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.11749267578125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0778359172913944e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0778359172913944e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.623330593109131, "val_fraction": 2.378234398782344e-05, "val_loss": 2.1701221466064453, "within_bucket_gap": -0.45320844650268555}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.153224945068359, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.8194381109292826e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.8194381109292826e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.333676815032959, "val_fraction": 2.378234398782344e-05, "val_loss": 4.307199478149414, "within_bucket_gap": -2.026477336883545}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016615405217576665, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016615405217576665, "step": 1000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 8.775683085123697, "val_fraction": 2.378234398782344e-05, "val_loss": 1.7892374992370605, "within_bucket_gap": -6.986445585886637}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.194609968266712e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.194609968266712e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.385375499725342, "val_fraction": 2.378234398782344e-05, "val_loss": 6.149125099182129, "within_bucket_gap": 1.763749599456787}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.755681991577148, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.351306560286285e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.351306560286285e-05, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.612965703010559, "val_fraction": 2.378234398782344e-05, "val_loss": 7.601643085479736, "within_bucket_gap": 0.9886773824691772}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.397263526916504, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.1861580610275269, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.40096113085746765, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.63367223739624, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.374662399291992, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.1450271606445312, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.82611083984375, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.02133985236287117, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.479318141937256, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.8937158584594727, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.82590271330678e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.82590271330678e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.910837411880493, "val_fraction": 2.378234398782344e-05, "val_loss": 5.7809929847717285, "within_bucket_gap": 2.8701555728912354}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.793330192565918, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.880273818969727, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.626533594396379e-06, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.626533594396379e-06, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.059987723827362, "val_fraction": 2.378234398782344e-05, "val_loss": 1.823403239250183, "within_bucket_gap": -0.23658448457717896}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016047516369928507, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016047516369928507, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.0886454582214355, "val_fraction": 2.378234398782344e-05, "val_loss": 7.836305141448975, "within_bucket_gap": 6.747659683227539}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.816539764404297, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.308200836181641, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.026534274220466614, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014323170780228335, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014323170780228335, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.810379981994629, "val_fraction": 2.378234398782344e-05, "val_loss": 9.832986831665039, "within_bucket_gap": 6.02260684967041}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001333461368464988, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001333461368464988, "step": 1000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.385237455368042, "val_fraction": 4.756468797564688e-05, "val_loss": 2.581768274307251, "within_bucket_gap": -2.803469181060791}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.005246639251709, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.4133830070495605, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.003237247467041, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.4215850830078125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.16256588697433472, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.10953116416931152, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.37414026260376, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.0012537722941488028, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.719913262029035e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.719913262029035e-05, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.3251633644104004, "val_fraction": 2.378234398782344e-05, "val_loss": 4.468832492828369, "within_bucket_gap": 1.1436691284179688}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.554049491882324, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.322991371154785, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.1965517618068276e-06, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.1965517618068276e-06, "step": 1000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.3372252583503723, "val_fraction": 2.378234398782344e-05, "val_loss": 0.3875378668308258, "within_bucket_gap": 0.05031260848045349}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.406746782555246e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.406746782555246e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.458146572113037, "val_fraction": 4.756468797564688e-05, "val_loss": 0.32143212854862213, "within_bucket_gap": -1.136714443564415}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.4141281355098683e-05, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.4141281355098683e-05, "step": 1000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.81780481338501, "val_fraction": 9.512937595129375e-05, "val_loss": 4.669151663780212, "within_bucket_gap": -0.14865314960479736}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.020123276219742907, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.020123276219742907, "step": 1200, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 5.546307293318723, "val_fraction": 0.05995528919330289, "val_loss": 5.210669245525089, "within_bucket_gap": -0.3356380477936334}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0008854159054987501, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0008854159054987501, "step": 1200, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 5.49661443212923, "val_fraction": 0.019882039573820395, "val_loss": 5.452080977590461, "within_bucket_gap": -0.04453345453876967}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0022188247366516064, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0022188247366516064, "step": 1200, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 5.456412674753289, "val_fraction": 0.011582001522070015, "val_loss": 5.264837433425308, "within_bucket_gap": -0.19157524132798098}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007136266550943549, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007136266550943549, "step": 1200, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.228523120545504, "val_fraction": 0.007301179604261796, "val_loss": 5.130781838482288, "within_bucket_gap": -0.09774128206321642}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003337804552987939, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003337804552987939, "step": 1200, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 5.204970434698135, "val_fraction": 0.004827815829528158, "val_loss": 5.2741073833897785, "within_bucket_gap": 0.06913694869164377}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005001389743665705, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005001389743665705, "step": 1200, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 5.4832776658078455, "val_fraction": 0.003305745814307458, "val_loss": 5.331983827000899, "within_bucket_gap": -0.15129383880694647}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0018193206992173562, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0018193206992173562, "step": 1200, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 5.664101484883038, "val_fraction": 0.002877663622526636, "val_loss": 5.031880024050878, "within_bucket_gap": -0.6322214608321604}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016175700701120228, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016175700701120228, "step": 1200, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.205205372401646, "val_fraction": 0.002449581430745814, "val_loss": 5.139170822587985, "within_bucket_gap": -0.06603454981366053}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007393892016648755, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007393892016648755, "step": 1200, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 5.161836916261965, "val_fraction": 0.0025684931506849314, "val_loss": 4.873968053747107, "within_bucket_gap": -0.2878688625148582}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007060486091407829, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007060486091407829, "step": 1200, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 4.861505002391581, "val_fraction": 0.0014745053272450532, "val_loss": 5.340342613958543, "within_bucket_gap": 0.4788376115669619}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0014131707919056303, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0014131707919056303, "step": 1200, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 5.494841275149828, "val_fraction": 0.0020214992389649923, "val_loss": 4.795770622702205, "within_bucket_gap": -0.6990706524476229}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00035153825573765456, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00035153825573765456, "step": 1200, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 4.763759528266059, "val_fraction": 0.0016647640791476407, "val_loss": 4.552595520019532, "within_bucket_gap": -0.21116400824652715}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008371829431564383, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008371829431564383, "step": 1200, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 4.93518133724437, "val_fraction": 0.0018074581430745813, "val_loss": 5.398363816110711, "within_bucket_gap": 0.463182478866341}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004339541218276303, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004339541218276303, "step": 1200, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 4.602014653822955, "val_fraction": 0.0015220700152207, "val_loss": 4.8871225118637085, "within_bucket_gap": 0.2851078580407531}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007581284743689396, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007581284743689396, "step": 1200, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 5.56687351084349, "val_fraction": 0.0014745053272450532, "val_loss": 5.052715670677923, "within_bucket_gap": -0.5141578401655673}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003353428078568691, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003353428078568691, "step": 1200, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 5.30309242425963, "val_fraction": 0.0010226407914764079, "val_loss": 5.631010898323947, "within_bucket_gap": 0.32791847406431707}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002272331724885708, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002272331724885708, "step": 1200, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.479929053265115, "val_fraction": 0.0004280821917808219, "val_loss": 4.949112362331814, "within_bucket_gap": -0.5308166909333014}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001460275693571184, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001460275693571184, "step": 1200, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.2907867431640625, "val_fraction": 0.0007372526636225266, "val_loss": 5.0927168323147685, "within_bucket_gap": -0.19806991084929404}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00022758951742355118, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00022758951742355118, "step": 1200, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.174288749694824, "val_fraction": 0.000594558599695586, "val_loss": 5.557076110839843, "within_bucket_gap": 0.3827873611450192}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00031594873633126195, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00031594873633126195, "step": 1200, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.590940622182993, "val_fraction": 0.0007372526636225266, "val_loss": 6.0194894113848285, "within_bucket_gap": 0.4285487892018356}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00027994986175583557, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00027994986175583557, "step": 1200, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 5.3374481201171875, "val_fraction": 0.0006183409436834094, "val_loss": 5.790191650390625, "within_bucket_gap": 0.4527435302734375}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00028369252540205183, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00028369252540205183, "step": 1200, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.5389202662876675, "val_fraction": 0.0006659056316590563, "val_loss": 5.964945384434292, "within_bucket_gap": 0.42602511814662414}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006165998949579995, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006165998949579995, "step": 1200, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 4.207074344158173, "val_fraction": 0.00047564687975646877, "val_loss": 5.503413963317871, "within_bucket_gap": 1.2963396191596983}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001892092075522087, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001892092075522087, "step": 1200, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.032895965576172, "val_fraction": 0.0003091704718417047, "val_loss": 4.644885870126577, "within_bucket_gap": 0.6119899045504056}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00035318687872462596, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00035318687872462596, "step": 1200, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.105661528451102, "val_fraction": 0.000594558599695586, "val_loss": 4.699693603515625, "within_bucket_gap": 0.5940320750645229}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020134603233279349, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020134603233279349, "step": 1200, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.113260904947917, "val_fraction": 0.0004280821917808219, "val_loss": 5.642916573418511, "within_bucket_gap": -0.4703443315294056}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000558755347063218, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000558755347063218, "step": 1200, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 4.193698583864698, "val_fraction": 0.000594558599695586, "val_loss": 5.133480377197266, "within_bucket_gap": 0.9397817933325676}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.852811000877289e-05, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.852811000877289e-05, "step": 1200, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.228532964533025, "val_fraction": 0.00045186453576864533, "val_loss": 5.313797699777703, "within_bucket_gap": 0.08526473524467804}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009776165704052432, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009776165704052432, "step": 1200, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.5207198358351186, "val_fraction": 0.00040429984779299846, "val_loss": 4.938768162446864, "within_bucket_gap": 2.418048326611745}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004657884348110157, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004657884348110157, "step": 1200, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.5117645263671875, "val_fraction": 0.00033295281582952814, "val_loss": 5.910726819719587, "within_bucket_gap": 1.3989622933523993}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00044606671395004254, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00044606671395004254, "step": 1200, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.6292924880981445, "val_fraction": 0.0002616057838660578, "val_loss": 4.924182198264382, "within_bucket_gap": -1.7051102898337627}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0011336380518492796, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0011336380518492796, "step": 1200, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.489208539326985, "val_fraction": 0.0005232115677321156, "val_loss": 4.322517048228871, "within_bucket_gap": -2.1666914910981143}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00030064113336066684, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00030064113336066684, "step": 1200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.889466285705566, "val_fraction": 0.0002616057838660578, "val_loss": 4.7402518879283555, "within_bucket_gap": -1.149214397777211}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.534940372198003e-05, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.534940372198003e-05, "step": 1200, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.175804773966472, "val_fraction": 0.0002616057838660578, "val_loss": 5.463831294666637, "within_bucket_gap": 0.2880265207001651}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003027010091000677, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003027010091000677, "step": 1200, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 4.785229365030925, "val_fraction": 0.00014269406392694063, "val_loss": 6.906558036804199, "within_bucket_gap": 2.1213286717732744}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011153383258936189, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011153383258936189, "step": 1200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.079286495844523, "val_fraction": 0.0002616057838660578, "val_loss": 6.652943351052024, "within_bucket_gap": -0.426343144792499}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.7690897109622816e-06, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.7690897109622816e-06, "step": 1200, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.019169440636268, "val_fraction": 0.00028538812785388126, "val_loss": 5.0288723309834795, "within_bucket_gap": 0.009702890347211834}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006896153976928155, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006896153976928155, "step": 1200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.1159617106119795, "val_fraction": 0.0003567351598173516, "val_loss": 5.182831827799479, "within_bucket_gap": -1.9331298828125005}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002082636771711343, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002082636771711343, "step": 1200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.221822341283162, "val_fraction": 0.0002616057838660578, "val_loss": 6.0179197138006035, "within_bucket_gap": 0.7960973725174414}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013202378920405602, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013202378920405602, "step": 1200, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 3.465087890625, "val_fraction": 9.512937595129375e-05, "val_loss": 4.852921962738037, "within_bucket_gap": 1.387834072113037}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00032584551988308483, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00032584551988308483, "step": 1200, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.774823760986328, "val_fraction": 0.00021404109589041095, "val_loss": 6.2971740298801, "within_bucket_gap": 1.5223502688937725}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.596514456152747e-05, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.596514456152747e-05, "step": 1200, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 3.756697056097386, "val_fraction": 0.00014269406392694063, "val_loss": 4.21898078918457, "within_bucket_gap": 0.4622837330871845}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00047209286694942427, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00047209286694942427, "step": 1200, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.1921352658952986, "val_fraction": 0.00023782343987823439, "val_loss": 5.177191352844238, "within_bucket_gap": 1.9850560869489393}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006114565618506305, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006114565618506305, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.080059051513672, "val_fraction": 0.0002616057838660578, "val_loss": 4.417379552667791, "within_bucket_gap": 2.337320501154119}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001927464368801803, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001927464368801803, "step": 1200, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.221420923868815, "val_fraction": 0.00016647640791476407, "val_loss": 6.37922123500279, "within_bucket_gap": 1.1578003111339745}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005234667699630946, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005234667699630946, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.4519429206848145, "val_fraction": 0.000380517503805175, "val_loss": 4.076272249221802, "within_bucket_gap": -1.3756706714630127}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010313529129986348, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010313529129986348, "step": 1200, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.503848075866699, "val_fraction": 7.134703196347032e-05, "val_loss": 6.949392318725586, "within_bucket_gap": 1.4455442428588867}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.797138059520286e-05, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.797138059520286e-05, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.087948322296143, "val_fraction": 4.756468797564688e-05, "val_loss": 3.4486780166625977, "within_bucket_gap": -1.639270305633545}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015608167254222636, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015608167254222636, "step": 1200, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 1.4983013037479285, "val_fraction": 4.756468797564688e-05, "val_loss": 4.779762387275696, "within_bucket_gap": 3.281461083527767}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.28989688373954e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.28989688373954e-05, "step": 1200, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.242258071899414, "val_fraction": 0.00016647640791476407, "val_loss": 4.9845700945172995, "within_bucket_gap": -0.2576879773821146}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003655136674869369, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003655136674869369, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.142290115356445, "val_fraction": 7.134703196347032e-05, "val_loss": 3.0192505518595376, "within_bucket_gap": -5.123039563496908}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.456928930507584e-05, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.456928930507584e-05, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.0041333436965942, "val_fraction": 4.756468797564688e-05, "val_loss": 1.2261486053466797, "within_bucket_gap": -1.7779847383499146}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013173480906365198, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013173480906365198, "step": 1200, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.331622873033796, "val_fraction": 0.00014269406392694063, "val_loss": 5.408425331115723, "within_bucket_gap": -0.9231975419180731}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.5470932767452894e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.5470932767452894e-05, "step": 1200, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 2.951728343963623, "val_fraction": 2.378234398782344e-05, "val_loss": 4.443210124969482, "within_bucket_gap": 1.4914817810058594}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.454325536614688e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.454325536614688e-05, "step": 1200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.555439186096192, "val_fraction": 0.00016647640791476407, "val_loss": 5.047599928719657, "within_bucket_gap": -0.5078392573765349}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010850202275193445, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010850202275193445, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.949003219604492, "val_fraction": 0.00014269406392694063, "val_loss": 5.709385395050049, "within_bucket_gap": 0.7603821754455566}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003279111542295284, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003279111542295284, "step": 1200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.136963367462158, "val_fraction": 0.00011891171993911719, "val_loss": 2.3793617248535157, "within_bucket_gap": -2.7576016426086425}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011516912716980937, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011516912716980937, "step": 1200, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.2620693147182465, "val_fraction": 9.512937595129375e-05, "val_loss": 5.472727179527283, "within_bucket_gap": 1.2106578648090363}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004948577047422535, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004948577047422535, "step": 1200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.621956507364909, "val_fraction": 0.00023782343987823439, "val_loss": 6.702734184265137, "within_bucket_gap": 2.0807776769002277}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00031104807938977825, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00031104807938977825, "step": 1200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.899815368652344, "val_fraction": 0.00021404109589041095, "val_loss": 4.4465987417432995, "within_bucket_gap": -1.453216626909044}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011651097730233062, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011651097730233062, "step": 1200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.82971420288086, "val_fraction": 0.00011891171993911719, "val_loss": 5.84990348815918, "within_bucket_gap": -0.9798107147216797}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014151975263017858, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014151975263017858, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.757299423217773, "val_fraction": 7.134703196347032e-05, "val_loss": 6.7408402760823565, "within_bucket_gap": 1.983540852864583}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.9837617172861045e-05, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.9837617172861045e-05, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.403069615364075, "val_fraction": 0.00016647640791476407, "val_loss": 6.043633597237723, "within_bucket_gap": -0.3594360181263516}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.853427994378262e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.853427994378262e-05, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.323732852935791, "val_fraction": 0.0001902587519025875, "val_loss": 6.8416290283203125, "within_bucket_gap": 0.5178961753845215}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017447862348872237, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017447862348872237, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.652366995811462, "val_fraction": 0.0002616057838660578, "val_loss": 5.319319464943626, "within_bucket_gap": 0.6669524691321636}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.854976926764396e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.854976926764396e-05, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.763046622276306, "val_fraction": 4.756468797564688e-05, "val_loss": 4.742336273193359, "within_bucket_gap": -1.0207103490829468}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00028137320899255744, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00028137320899255744, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 8.09815502166748, "val_fraction": 7.134703196347032e-05, "val_loss": 4.154428124427795, "within_bucket_gap": -3.943726897239685}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.6180342270721944e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.6180342270721944e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.629734516143799, "val_fraction": 4.756468797564688e-05, "val_loss": 7.180150032043457, "within_bucket_gap": 0.5504155158996582}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018725588441439415, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018725588441439415, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.327205181121826, "val_fraction": 9.512937595129375e-05, "val_loss": 7.2956390380859375, "within_bucket_gap": 1.9684338569641113}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002102419121624672, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002102419121624672, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.524414539337158, "val_fraction": 9.512937595129375e-05, "val_loss": 4.734477519989014, "within_bucket_gap": 2.2100629806518555}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018527684690745455, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018527684690745455, "step": 1200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.392557525634766, "val_fraction": 7.134703196347032e-05, "val_loss": 2.795717239379883, "within_bucket_gap": -2.596840286254883}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00024475735616466223, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00024475735616466223, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.911968231201172, "val_fraction": 0.0001902587519025875, "val_loss": 4.198412895202637, "within_bucket_gap": 1.2864446640014648}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.497144618444428e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.497144618444428e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.192978858947754, "val_fraction": 2.378234398782344e-05, "val_loss": 1.7224994897842407, "within_bucket_gap": -1.4704793691635132}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00025729600856111835, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00025729600856111835, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.161228895187378, "val_fraction": 7.134703196347032e-05, "val_loss": 5.767489751180013, "within_bucket_gap": 3.606260855992635}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001883639796683777, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001883639796683777, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.3111586570739746, "val_fraction": 0.00011891171993911719, "val_loss": 3.895224380493164, "within_bucket_gap": 1.5840657234191893}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021294277597961542, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021294277597961542, "step": 1200, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.768496513366699, "val_fraction": 7.134703196347032e-05, "val_loss": 7.753102461496989, "within_bucket_gap": 2.98460594813029}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014474475401903033, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014474475401903033, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.612217664718628, "val_fraction": 9.512937595129375e-05, "val_loss": 7.133774518966675, "within_bucket_gap": 1.5215568542480469}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.549324989318848, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.2408720848222846e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.2408720848222846e-05, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.095527172088623, "val_fraction": 4.756468797564688e-05, "val_loss": 5.834646224975586, "within_bucket_gap": -0.2608809471130371}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018870571976927316, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018870571976927316, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.4529852867126465, "val_fraction": 4.756468797564688e-05, "val_loss": 2.4856362342834473, "within_bucket_gap": -3.967349052429199}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.87545108795166, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002373611409914548, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002373611409914548, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.8785400390625, "val_fraction": 0.0001902587519025875, "val_loss": 3.6309698820114136, "within_bucket_gap": -1.2475701570510864}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00029397738850824365, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00029397738850824365, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.907365798950195, "val_fraction": 7.134703196347032e-05, "val_loss": 5.786978721618652, "within_bucket_gap": -4.120387077331543}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00023057111829541952, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00023057111829541952, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.43164587020874, "val_fraction": 4.756468797564688e-05, "val_loss": 2.58411867916584, "within_bucket_gap": -4.8475271910429}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.670280965072497e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.670280965072497e-05, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.634931882222493, "val_fraction": 9.512937595129375e-05, "val_loss": 4.143991947174072, "within_bucket_gap": -0.49093993504842093}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.003313346177657e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.003313346177657e-05, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.5693464279174805, "val_fraction": 9.512937595129375e-05, "val_loss": 4.622918128967285, "within_bucket_gap": -0.9464282989501953}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.7085114355500972e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.7085114355500972e-05, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.729711651802063, "val_fraction": 2.378234398782344e-05, "val_loss": 2.868586540222168, "within_bucket_gap": 1.138874888420105}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.5663973114806224e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.5663973114806224e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.1068274974823, "val_fraction": 2.378234398782344e-05, "val_loss": 3.606426239013672, "within_bucket_gap": 1.499598741531372}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020550795462698817, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020550795462698817, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.613989194234212, "val_fraction": 4.756468797564688e-05, "val_loss": 8.934588432312012, "within_bucket_gap": 4.3205992380778}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015976524357563465, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015976524357563465, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.8231866359710693, "val_fraction": 2.378234398782344e-05, "val_loss": 9.540995597839355, "within_bucket_gap": 6.717808961868286}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.92950771206953e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.92950771206953e-05, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.888739585876465, "val_fraction": 4.756468797564688e-05, "val_loss": 6.135359287261963, "within_bucket_gap": 1.246619701385498}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013989369460073962, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013989369460073962, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.117530584335327, "val_fraction": 2.378234398782344e-05, "val_loss": 7.999780654907227, "within_bucket_gap": 5.882250070571899}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001601838448639148, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001601838448639148, "step": 1200, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 3.8785146713256835, "val_fraction": 2.378234398782344e-05, "val_loss": 10.613924980163574, "within_bucket_gap": 6.73541030883789}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.112305020632808e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.112305020632808e-05, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.8656210899353027, "val_fraction": 4.756468797564688e-05, "val_loss": 5.7813920974731445, "within_bucket_gap": 1.9157710075378418}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005783113669405607, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005783113669405607, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.684605598449707, "val_fraction": 7.134703196347032e-05, "val_loss": 4.578993479410808, "within_bucket_gap": -8.105612119038899}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.679741382598877, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.72467041015625, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.3456622068196125e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.3456622068196125e-05, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.660973787307739, "val_fraction": 2.378234398782344e-05, "val_loss": 5.64727783203125, "within_bucket_gap": 0.9863040447235107}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.532806158065796, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.772634506225586, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.1035211418317332e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.1035211418317332e-05, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.385286569595337, "val_fraction": 2.378234398782344e-05, "val_loss": 3.9212779998779297, "within_bucket_gap": -0.4640085697174072}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.7720715999603271, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.144903540611267, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.97574520111084, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000322639758597997, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000322639758597997, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4851589649915695, "val_fraction": 4.756468797564688e-05, "val_loss": 8.26833724975586, "within_bucket_gap": 6.78317828476429}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.1603877312153624e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.1603877312153624e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.91105318069458, "val_fraction": 2.378234398782344e-05, "val_loss": 5.5821733474731445, "within_bucket_gap": -1.3288798332214355}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016034259269498798, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016034259269498798, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.8673375844955444, "val_fraction": 7.134703196347032e-05, "val_loss": 3.114699363708496, "within_bucket_gap": 2.2473617792129517}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.465424537658691, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016729625528805875, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016729625528805875, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.930534839630127, "val_fraction": 0.00014269406392694063, "val_loss": 2.758122682571411, "within_bucket_gap": -1.1724121570587158}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010286563395970003, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010286563395970003, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.342790524164836, "val_fraction": 4.756468797564688e-05, "val_loss": 6.505437612533569, "within_bucket_gap": 2.1626470883687334}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.090715822382423e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.090715822382423e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.4764657020568848, "val_fraction": 2.378234398782344e-05, "val_loss": 6.037489891052246, "within_bucket_gap": 2.5610241889953613}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000138690868106244, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000138690868106244, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.775854110717773, "val_fraction": 2.378234398782344e-05, "val_loss": 2.944180488586426, "within_bucket_gap": -5.831673622131348}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.9149052759283753e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.9149052759283753e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.995146751403809, "val_fraction": 2.378234398782344e-05, "val_loss": 8.220806121826172, "within_bucket_gap": 1.2256593704223633}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.593701362609863, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.360461950302124, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.726580947864364e-06, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.726580947864364e-06, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.410151958465576, "val_fraction": 4.756468797564688e-05, "val_loss": 4.572595596313477, "within_bucket_gap": 0.1624436378479004}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.89780330657959, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.0607823133468628, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.4265151023864746, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.071905215581258, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.371909527473797e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.371909527473797e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.4300529956817627, "val_fraction": 2.378234398782344e-05, "val_loss": 4.950273513793945, "within_bucket_gap": 3.5202205181121826}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.453427791595459, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.377126455307007, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.23429831862449646, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.959590911865234, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.856054409090967e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.856054409090967e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.2153167724609375, "val_fraction": 4.756468797564688e-05, "val_loss": 4.143179893493652, "within_bucket_gap": -2.072136878967285}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.340048313140869, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.62143087387085, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.039469838142395, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.1636095732858735e-06, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.1636095732858735e-06, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.440838098526001, "val_fraction": 2.378234398782344e-05, "val_loss": 3.6579575538635254, "within_bucket_gap": 0.21711945533752441}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.5760484783253894e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.5760484783253894e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.415833950042725, "val_fraction": 2.378234398782344e-05, "val_loss": 7.919490814208984, "within_bucket_gap": 1.5036568641662598}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.474981531886387e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.474981531886387e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.061431884765625, "val_fraction": 2.378234398782344e-05, "val_loss": 3.918351650238037, "within_bucket_gap": -3.143080234527588}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.255090713500977, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.432911396026611, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.7971062660217285, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00038195411636404795, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00038195411636404795, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7272791862487793, "val_fraction": 4.756468797564688e-05, "val_loss": 11.757482528686523, "within_bucket_gap": 8.030203342437744}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.10235799103975296, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.946655511856079, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.1495554447174072, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.753562569618225, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.370952606201172, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.692162603956379e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.692162603956379e-05, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.764119446277618, "val_fraction": 2.378234398782344e-05, "val_loss": 2.79115891456604, "within_bucket_gap": -1.9729605317115784}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.654621124267578, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.15397129580378532, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.902170710730589e-06, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.902170710730589e-06, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7280192375183105, "val_fraction": 2.378234398782344e-05, "val_loss": 2.8920977115631104, "within_bucket_gap": 0.1640784740447998}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.120783446586295e-06, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.120783446586295e-06, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.757723331451416, "val_fraction": 2.378234398782344e-05, "val_loss": 3.5844526290893555, "within_bucket_gap": -0.17327070236206055}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.7014186382293701, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.526666641235352, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.810699224472046, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.622002601623535, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.646450996398926, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.544050216674805, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.1187111139297485, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.6859183980994026e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.6859183980994026e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.9296865463256836, "val_fraction": 2.378234398782344e-05, "val_loss": 2.2207915782928467, "within_bucket_gap": -0.7088949680328369}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.513233184814453, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.73393227012553e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.73393227012553e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.553454399108887, "val_fraction": 2.378234398782344e-05, "val_loss": 3.721970558166504, "within_bucket_gap": -2.831483840942383}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015656835007855032, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015656835007855032, "step": 1200, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 8.511116663614908, "val_fraction": 2.378234398782344e-05, "val_loss": 1.927730679512024, "within_bucket_gap": -6.583385984102884}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.95707139685818e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.95707139685818e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.93125057220459, "val_fraction": 2.378234398782344e-05, "val_loss": 6.595119953155518, "within_bucket_gap": 1.6638693809509277}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.121999740600586, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.2832134134693232e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.2832134134693232e-05, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.569963693618774, "val_fraction": 2.378234398782344e-05, "val_loss": 7.5300092697143555, "within_bucket_gap": 0.960045576095581}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.461400985717773, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.8585259914398193, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.13566282391548157, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.441589832305908, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.927359580993652, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.1157207489013672, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.675447463989258, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.023975595831871033, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.731324195861816, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.297699451446533, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.011638233650764e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.011638233650764e-05, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.9504358768463135, "val_fraction": 2.378234398782344e-05, "val_loss": 6.319169521331787, "within_bucket_gap": 3.3687336444854736}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.320935249328613, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.196579933166504, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.9137426111251794e-06, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.9137426111251794e-06, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.022733986377716, "val_fraction": 2.378234398782344e-05, "val_loss": 2.1452510356903076, "within_bucket_gap": 0.12251704931259155}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001694724471216151, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001694724471216151, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.742805540561676, "val_fraction": 2.378234398782344e-05, "val_loss": 7.868782997131348, "within_bucket_gap": 7.125977456569672}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.490786075592041, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.425649642944336, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.016918379813432693, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014864041822958208, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014864041822958208, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.8629132509231567, "val_fraction": 2.378234398782344e-05, "val_loss": 10.112945556640625, "within_bucket_gap": 6.250032305717468}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014709394362782052, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014709394362782052, "step": 1200, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.453514337539673, "val_fraction": 4.756468797564688e-05, "val_loss": 2.361011266708374, "within_bucket_gap": -3.092503070831299}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.8017477989196777, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.500373840332031, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.260899782180786, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.12502175569534302, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.02519860677421093, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.03791103884577751, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.441746711730957, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.00018499570433050394, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.4216376856944695e-06, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.4216376856944695e-06, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.922327995300293, "val_fraction": 2.378234398782344e-05, "val_loss": 3.982105016708374, "within_bucket_gap": 0.059777021408081055}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7585713863372803, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.838134765625, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.5187822849329567e-07, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.5187822849329567e-07, "step": 1200, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.17235426604747772, "val_fraction": 2.378234398782344e-05, "val_loss": 0.16596809029579163, "within_bucket_gap": -0.006386175751686096}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010104706444277156, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010104706444277156, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.3403377532958984, "val_fraction": 4.756468797564688e-05, "val_loss": 0.21592427045106888, "within_bucket_gap": -2.1244134828448296}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012193032777109827, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012193032777109827, "step": 1200, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.753927707672119, "val_fraction": 9.512937595129375e-05, "val_loss": 4.472196102142334, "within_bucket_gap": -1.2817316055297852}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.018169461347317863, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.018169461347317863, "step": 1400, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 5.381263697050088, "val_fraction": 0.05995528919330289, "val_loss": 5.078213514292443, "within_bucket_gap": -0.3030501827576444}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006076842114372458, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006076842114372458, "step": 1400, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 5.296909344931984, "val_fraction": 0.019882039573820395, "val_loss": 5.266344864402662, "within_bucket_gap": -0.030564480529322147}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0010314662829142946, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0010314662829142946, "step": 1400, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 5.194803916529605, "val_fraction": 0.011582001522070015, "val_loss": 5.105746228102541, "within_bucket_gap": -0.08905768842706419}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005232291443299713, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005232291443299713, "step": 1400, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 5.026575563665022, "val_fraction": 0.007301179604261796, "val_loss": 4.954911918515879, "within_bucket_gap": -0.07166364514914214}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005962885921279803, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005962885921279803, "step": 1400, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 4.959499019602831, "val_fraction": 0.004827815829528158, "val_loss": 5.083010067493458, "within_bucket_gap": 0.12351104789062717}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00046544523502784336, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00046544523502784336, "step": 1400, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 5.325029305532469, "val_fraction": 0.003305745814307458, "val_loss": 5.184230447673111, "within_bucket_gap": -0.14079885785935797}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0018740430208275755, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0018740430208275755, "step": 1400, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 5.55439255554552, "val_fraction": 0.002877663622526636, "val_loss": 4.903154861828512, "within_bucket_gap": -0.6512376937170075}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.660338448037933e-06, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.660338448037933e-06, "step": 1400, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 5.0724062238420755, "val_fraction": 0.002449581430745814, "val_loss": 5.075941659871814, "within_bucket_gap": 0.0035354360297388254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005039356183356462, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005039356183356462, "step": 1400, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 4.996235752964879, "val_fraction": 0.0025684931506849314, "val_loss": 4.800036818892868, "within_bucket_gap": -0.1961989340720116}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007739786488091599, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007739786488091599, "step": 1400, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 4.65877797652264, "val_fraction": 0.0014745053272450532, "val_loss": 5.183685302734375, "within_bucket_gap": 0.5249073262117347}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0012022847093893267, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0012022847093893267, "step": 1400, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 5.273149046179366, "val_fraction": 0.0020214992389649923, "val_loss": 4.678400017233455, "within_bucket_gap": -0.5947490289459108}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001820078481580334, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001820078481580334, "step": 1400, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 4.447417195638021, "val_fraction": 0.0016647640791476407, "val_loss": 4.338087681361607, "within_bucket_gap": -0.10932951427641413}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007024522284085009, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007024522284085009, "step": 1400, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 4.785643745871151, "val_fraction": 0.0018074581430745813, "val_loss": 5.174284684030633, "within_bucket_gap": 0.3886409381594822}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006777503430891389, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006777503430891389, "step": 1400, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 4.247499353745404, "val_fraction": 0.0015220700152207, "val_loss": 4.692781329154968, "within_bucket_gap": 0.44528197540956427}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005105088552561971, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005105088552561971, "step": 1400, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 5.231651717696447, "val_fraction": 0.0014745053272450532, "val_loss": 4.885427905667212, "within_bucket_gap": -0.34622381202923513}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002478809298627092, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002478809298627092, "step": 1400, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 5.190976165061773, "val_fraction": 0.0010226407914764079, "val_loss": 5.433369126430778, "within_bucket_gap": 0.24239296136900457}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014830140510232404, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014830140510232404, "step": 1400, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 5.237132528553838, "val_fraction": 0.0004280821917808219, "val_loss": 4.890700446234809, "within_bucket_gap": -0.346432082319029}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00019805361531813445, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00019805361531813445, "step": 1400, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.10398715879859, "val_fraction": 0.0007372526636225266, "val_loss": 4.835349790511593, "within_bucket_gap": -0.26863736828699736}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003097831528973234, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003097831528973234, "step": 1400, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.876349585396903, "val_fraction": 0.000594558599695586, "val_loss": 5.397380065917969, "within_bucket_gap": 0.5210304805210662}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006369867679065347, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006369867679065347, "step": 1400, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 5.074707764845628, "val_fraction": 0.0007372526636225266, "val_loss": 5.9387083976499495, "within_bucket_gap": 0.8640006328043217}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004479216974426707, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004479216974426707, "step": 1400, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 4.998114432058027, "val_fraction": 0.0006183409436834094, "val_loss": 5.722507183368389, "within_bucket_gap": 0.7243927513103623}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003577349211345885, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003577349211345885, "step": 1400, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.275705065046038, "val_fraction": 0.0006659056316590563, "val_loss": 5.812920706612723, "within_bucket_gap": 0.5372156415666849}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006803796148917082, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006803796148917082, "step": 1400, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 4.050986409187317, "val_fraction": 0.00047564687975646877, "val_loss": 5.481416511535644, "within_bucket_gap": 1.4304301023483275}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019692855701417507, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019692855701417507, "step": 1400, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 3.8583904266357423, "val_fraction": 0.0003091704718417047, "val_loss": 4.495348270122822, "within_bucket_gap": 0.6369578434870795}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005679498725311732, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005679498725311732, "step": 1400, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.7392257962908064, "val_fraction": 0.000594558599695586, "val_loss": 4.694472045898437, "within_bucket_gap": 0.9552462496076308}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002847706709096783, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002847706709096783, "step": 1400, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.110761642456055, "val_fraction": 0.0004280821917808219, "val_loss": 5.445537355211046, "within_bucket_gap": -0.6652242872450085}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009250844798960572, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009250844798960572, "step": 1400, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 3.418707155713848, "val_fraction": 0.000594558599695586, "val_loss": 4.974625244140625, "within_bucket_gap": 1.5559180884267767}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.205280060225088e-05, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.205280060225088e-05, "step": 1400, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.015072562477806, "val_fraction": 0.00045186453576864533, "val_loss": 5.218790255094829, "within_bucket_gap": 0.20371769261702344}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009854697069656094, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009854697069656094, "step": 1400, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.4856876250236266, "val_fraction": 0.00040429984779299846, "val_loss": 4.923159991993623, "within_bucket_gap": 2.4374723669699967}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00040000312202940043, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00040000312202940043, "step": 1400, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.381532457139757, "val_fraction": 0.00033295281582952814, "val_loss": 5.582913262503488, "within_bucket_gap": 1.2013808053637307}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004109748195476908, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004109748195476908, "step": 1400, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.457890510559082, "val_fraction": 0.0002616057838660578, "val_loss": 4.886920582164418, "within_bucket_gap": -1.570969928394664}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0013578306261746787, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0013578306261746787, "step": 1400, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.314995129903157, "val_fraction": 0.0005232115677321156, "val_loss": 3.7198104858398438, "within_bucket_gap": -2.5951846440633135}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00023913601595759075, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00023913601595759075, "step": 1400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.379976113637288, "val_fraction": 0.0002616057838660578, "val_loss": 4.46586782282049, "within_bucket_gap": -0.9141082908167979}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011633970341100998, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011633970341100998, "step": 1400, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.913738992479113, "val_fraction": 0.0002616057838660578, "val_loss": 5.358452796936035, "within_bucket_gap": 0.44471380445692255}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003303880696971667, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003303880696971667, "step": 1400, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 4.252371311187744, "val_fraction": 0.00014269406392694063, "val_loss": 6.567730903625488, "within_bucket_gap": 2.315359592437744}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.7854813953634193e-06, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.7854813953634193e-06, "step": 1400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.992560068766276, "val_fraction": 0.0002616057838660578, "val_loss": 6.981912439519709, "within_bucket_gap": -0.01064762924656737}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.883007012936595e-05, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.883007012936595e-05, "step": 1400, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 4.762403194720928, "val_fraction": 0.00028538812785388126, "val_loss": 4.45114262898763, "within_bucket_gap": -0.3112605657332983}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007033060720159219, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007033060720159219, "step": 1400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 7.14786942799886, "val_fraction": 0.0003567351598173516, "val_loss": 5.176361846923828, "within_bucket_gap": -1.9715075810750324}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003435047582464736, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003435047582464736, "step": 1400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.846970240275065, "val_fraction": 0.0002616057838660578, "val_loss": 6.160032792524858, "within_bucket_gap": 1.313062552249793}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013071970770067736, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013071970770067736, "step": 1400, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 3.3209456530484287, "val_fraction": 9.512937595129375e-05, "val_loss": 4.695071220397949, "within_bucket_gap": 1.3741255673495205}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002892687730774668, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002892687730774668, "step": 1400, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.835842590332032, "val_fraction": 0.00021404109589041095, "val_loss": 6.187306298149957, "within_bucket_gap": 1.351463707817925}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014224161578901563, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014224161578901563, "step": 1400, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 3.2587071538487553, "val_fraction": 0.00014269406392694063, "val_loss": 4.255536397298177, "within_bucket_gap": 0.9968292434494215}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00044597109573564985, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00044597109573564985, "step": 1400, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.0946095330374583, "val_fraction": 0.00023782343987823439, "val_loss": 4.969828796386719, "within_bucket_gap": 1.8752192633492606}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000689249488566201, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000689249488566201, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 1.4974607229232788, "val_fraction": 0.0002616057838660578, "val_loss": 4.13214822248979, "within_bucket_gap": 2.6346874995665113}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.804805430627282e-05, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.804805430627282e-05, "step": 1400, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 5.0344958835177955, "val_fraction": 0.00016647640791476407, "val_loss": 5.38318225315639, "within_bucket_gap": 0.34868636963859423}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005034127554755596, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005034127554755596, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.179555892944336, "val_fraction": 0.000380517503805175, "val_loss": 3.8565871715545654, "within_bucket_gap": -1.3229687213897705}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015761377138012156, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015761377138012156, "step": 1400, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.113273075648716, "val_fraction": 7.134703196347032e-05, "val_loss": 7.3223876953125, "within_bucket_gap": 2.209114619663784}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010879314540002262, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010879314540002262, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.099379181861877, "val_fraction": 4.756468797564688e-05, "val_loss": 2.8121120929718018, "within_bucket_gap": -2.2872670888900757}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011849004013595211, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011849004013595211, "step": 1400, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 1.7676760355631511, "val_fraction": 4.756468797564688e-05, "val_loss": 4.258810639381409, "within_bucket_gap": 2.4911346038182574}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.8215410785587892e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.8215410785587892e-05, "step": 1400, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.019135475158691, "val_fraction": 0.00016647640791476407, "val_loss": 5.128552845546177, "within_bucket_gap": 0.10941737038748567}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004370613094515633, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004370613094515633, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.163527488708496, "val_fraction": 7.134703196347032e-05, "val_loss": 2.0376761754353843, "within_bucket_gap": -6.125851313273111}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.23421861861848e-05, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.23421861861848e-05, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.083878442645073, "val_fraction": 4.756468797564688e-05, "val_loss": 1.1424763202667236, "within_bucket_gap": -1.9414021223783493}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011132917949935088, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011132917949935088, "step": 1400, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 6.154641440936497, "val_fraction": 0.00014269406392694063, "val_loss": 5.374446551005046, "within_bucket_gap": -0.780194889931451}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.958625585702638e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.958625585702638e-05, "step": 1400, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 2.7044718265533447, "val_fraction": 2.378234398782344e-05, "val_loss": 4.36899471282959, "within_bucket_gap": 1.6645228862762451}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.4391816529690484e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.4391816529690484e-05, "step": 1400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.2481037139892575, "val_fraction": 0.00016647640791476407, "val_loss": 5.041516985212054, "within_bucket_gap": -0.20658672877720363}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.6762780965372446e-05, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.6762780965372446e-05, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.454996744791667, "val_fraction": 0.00014269406392694063, "val_loss": 5.782710313796997, "within_bucket_gap": 0.3277135690053301}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00041373176400973555, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00041373176400973555, "step": 1400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.332684707641602, "val_fraction": 0.00011891171993911719, "val_loss": 1.8533660650253296, "within_bucket_gap": -3.4793186426162723}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.393086099461333e-05, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.393086099461333e-05, "step": 1400, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.30921533703804, "val_fraction": 9.512937595129375e-05, "val_loss": 5.0863765478134155, "within_bucket_gap": 0.7771612107753754}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005198048492907751, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005198048492907751, "step": 1400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.377162933349609, "val_fraction": 0.00023782343987823439, "val_loss": 6.562838363647461, "within_bucket_gap": 2.1856754302978514}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005637240484130075, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005637240484130075, "step": 1400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.2801820755004885, "val_fraction": 0.00021404109589041095, "val_loss": 3.6464633213149176, "within_bucket_gap": -2.633718754185571}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.33901640487043e-05, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.33901640487043e-05, "step": 1400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.631177520751953, "val_fraction": 0.00011891171993911719, "val_loss": 5.929899597167969, "within_bucket_gap": -0.7012779235839837}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010784148398781115, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010784148398781115, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.171048641204834, "val_fraction": 7.134703196347032e-05, "val_loss": 6.682554880777995, "within_bucket_gap": 1.511506239573161}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00021692204517361957, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00021692204517361957, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.566168010234833, "val_fraction": 0.00016647640791476407, "val_loss": 5.263148273740496, "within_bucket_gap": -1.3030197364943366}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.838283174960888e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.838283174960888e-05, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.243521054585774, "val_fraction": 0.0001902587519025875, "val_loss": 6.708061218261719, "within_bucket_gap": 0.4645401636759443}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014523158198622262, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014523158198622262, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.532339572906494, "val_fraction": 0.0002616057838660578, "val_loss": 5.087493896484375, "within_bucket_gap": 0.5551543235778809}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.225708357366071e-06, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.225708357366071e-06, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.093089431524277, "val_fraction": 4.756468797564688e-05, "val_loss": 5.287050724029541, "within_bucket_gap": 0.19396129250526428}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002749320233022423, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002749320233022423, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.775895595550537, "val_fraction": 7.134703196347032e-05, "val_loss": 3.9224483569463096, "within_bucket_gap": -3.8534472386042276}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.340430356959048e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.340430356959048e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.1427001953125, "val_fraction": 4.756468797564688e-05, "val_loss": 6.65064811706543, "within_bucket_gap": -0.4920520782470703}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021132618143496809, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021132618143496809, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.634851455688477, "val_fraction": 9.512937595129375e-05, "val_loss": 6.856312274932861, "within_bucket_gap": 2.2214608192443848}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014184342762469518, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014184342762469518, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.5974743366241455, "val_fraction": 9.512937595129375e-05, "val_loss": 4.088532447814941, "within_bucket_gap": 1.491058111190796}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015305174993779193, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015305174993779193, "step": 1400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.3948322296142575, "val_fraction": 7.134703196347032e-05, "val_loss": 3.2496589024861655, "within_bucket_gap": -2.145173327128092}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00040564120178716184, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00040564120178716184, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.6679936647415161, "val_fraction": 0.0001902587519025875, "val_loss": 3.800043821334839, "within_bucket_gap": 2.1320501565933228}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.6665770630132475e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.6665770630132475e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.79846715927124, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4157848358154297, "within_bucket_gap": -2.3826823234558105}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021911010463669602, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021911010463669602, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.0312349796295166, "val_fraction": 7.134703196347032e-05, "val_loss": 5.102282206217448, "within_bucket_gap": 3.0710472265879316}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001669074061413152, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001669074061413152, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.301044762134552, "val_fraction": 0.00011891171993911719, "val_loss": 3.7046692848205565, "within_bucket_gap": 1.4036245226860045}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018562664542931214, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018562664542931214, "step": 1400, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 5.308104515075684, "val_fraction": 7.134703196347032e-05, "val_loss": 7.909847577412923, "within_bucket_gap": 2.6017430623372393}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016288128164079454, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016288128164079454, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.880920648574829, "val_fraction": 9.512937595129375e-05, "val_loss": 6.593128681182861, "within_bucket_gap": 1.7122080326080322}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.360518455505371, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.0265573860122004e-06, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.0265573860122004e-06, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.9058239459991455, "val_fraction": 4.756468797564688e-05, "val_loss": 5.842193603515625, "within_bucket_gap": -0.06363034248352051}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015116137093786416, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015116137093786416, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.815529823303223, "val_fraction": 4.756468797564688e-05, "val_loss": 2.6375131607055664, "within_bucket_gap": -3.1780166625976562}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.603488087654114, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012687501842028473, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012687501842028473, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.9433627128601074, "val_fraction": 0.0001902587519025875, "val_loss": 3.276507616043091, "within_bucket_gap": -0.6668550968170166}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00043085167578547695, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00043085167578547695, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.053927421569824, "val_fraction": 7.134703196347032e-05, "val_loss": 6.015110333760579, "within_bucket_gap": -6.038817087809245}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022096376188518975, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022096376188518975, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.912631273269653, "val_fraction": 4.756468797564688e-05, "val_loss": 2.267089143395424, "within_bucket_gap": -4.645542129874229}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.434931169603758e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.434931169603758e-05, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.240557034810384, "val_fraction": 9.512937595129375e-05, "val_loss": 4.391396999359131, "within_bucket_gap": 0.15083996454874704}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.500798273304282e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.500798273304282e-05, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.833083152770996, "val_fraction": 9.512937595129375e-05, "val_loss": 4.14971923828125, "within_bucket_gap": -0.6833639144897461}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.0783411585577002e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.0783411585577002e-05, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.20474910736084, "val_fraction": 2.378234398782344e-05, "val_loss": 3.0786499977111816, "within_bucket_gap": 0.8739008903503418}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.9764400406696662e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.9764400406696662e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.2422893047332764, "val_fraction": 2.378234398782344e-05, "val_loss": 3.4938228130340576, "within_bucket_gap": 1.2515335083007812}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020926973046777453, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020926973046777453, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.642627716064453, "val_fraction": 4.756468797564688e-05, "val_loss": 9.042314529418945, "within_bucket_gap": 4.399686813354492}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014685410751055363, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014685410751055363, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6434073448181152, "val_fraction": 2.378234398782344e-05, "val_loss": 9.818328857421875, "within_bucket_gap": 6.17492151260376}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.219847725769519e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.219847725769519e-05, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.876184940338135, "val_fraction": 4.756468797564688e-05, "val_loss": 6.1326457262039185, "within_bucket_gap": 0.2564607858657837}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016927972288527263, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016927972288527263, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.3633865118026733, "val_fraction": 2.378234398782344e-05, "val_loss": 8.481260299682617, "within_bucket_gap": 7.117873787879944}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016702764654812747, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016702764654812747, "step": 1400, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.038489913940429, "val_fraction": 2.378234398782344e-05, "val_loss": 11.061668395996094, "within_bucket_gap": 7.023178482055664}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.891687632967892e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.891687632967892e-05, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.5340667366981506, "val_fraction": 4.756468797564688e-05, "val_loss": 5.61369514465332, "within_bucket_gap": 2.0796284079551697}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005637727779884861, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005637727779884861, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.121927261352539, "val_fraction": 7.134703196347032e-05, "val_loss": 4.220088005065918, "within_bucket_gap": -7.901839256286621}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.193313777446747, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.520403861999512, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.4447653240385776e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.4447653240385776e-05, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.365153948465983, "val_fraction": 2.378234398782344e-05, "val_loss": 5.393128871917725, "within_bucket_gap": 1.0279749234517412}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.122638463973999, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.161097526550293, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.69505543541872e-06, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.69505543541872e-06, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.186618089675903, "val_fraction": 2.378234398782344e-05, "val_loss": 3.905104398727417, "within_bucket_gap": -0.28151369094848633}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.3051912784576416, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.473909378051758, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.61734390258789, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003068766725367244, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003068766725367244, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.3080390989780426, "val_fraction": 4.756468797564688e-05, "val_loss": 7.759814262390137, "within_bucket_gap": 6.451775163412094}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.1550011475155156e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.1550011475155156e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.408321380615234, "val_fraction": 2.378234398782344e-05, "val_loss": 6.24074649810791, "within_bucket_gap": -2.167574882507324}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010478523201231178, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010478523201231178, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.0529253482818604, "val_fraction": 7.134703196347032e-05, "val_loss": 2.5215951601664224, "within_bucket_gap": 1.468669811884562}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.43485689163208, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022976450576720533, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022976450576720533, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.8596306443214417, "val_fraction": 0.00014269406392694063, "val_loss": 2.2494409879048667, "within_bucket_gap": -1.610189656416575}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.667856310782004e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.667856310782004e-05, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.194404125213623, "val_fraction": 4.756468797564688e-05, "val_loss": 5.806494235992432, "within_bucket_gap": 1.6120901107788086}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.84752521213694e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.84752521213694e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.687457323074341, "val_fraction": 2.378234398782344e-05, "val_loss": 5.725744724273682, "within_bucket_gap": 2.038287401199341}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015457582346743462, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015457582346743462, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.029053688049316, "val_fraction": 2.378234398782344e-05, "val_loss": 2.529449462890625, "within_bucket_gap": -6.499604225158691}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.739435658970198e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.739435658970198e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.40711784362793, "val_fraction": 2.378234398782344e-05, "val_loss": 8.399955749511719, "within_bucket_gap": 1.992837905883789}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.075778007507324, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.7357609272003174, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.7395279904660206e-05, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.7395279904660206e-05, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.083057880401611, "val_fraction": 4.756468797564688e-05, "val_loss": 4.4487762451171875, "within_bucket_gap": 0.36571836471557617}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.065104007720947, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.7760978937149048, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.792072057723999, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.797731081644694, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.193006257579025e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.193006257579025e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.5166500806808472, "val_fraction": 2.378234398782344e-05, "val_loss": 4.541165351867676, "within_bucket_gap": 3.0245152711868286}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.4399142265319824, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.1986148357391357, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.34250548481941223, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.443787574768066, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.9136296755647e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.9136296755647e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.325339317321777, "val_fraction": 4.756468797564688e-05, "val_loss": 4.451337814331055, "within_bucket_gap": -1.8740015029907227}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.646580696105957, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.359126329421997, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.0048624007031321526, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.4770883766302052e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.4770883766302052e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.9257988929748535, "val_fraction": 2.378234398782344e-05, "val_loss": 3.3047127723693848, "within_bucket_gap": -0.6210861206054688}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.6164154291515844e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.6164154291515844e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.595073223114014, "val_fraction": 2.378234398782344e-05, "val_loss": 8.115703582763672, "within_bucket_gap": 1.5206303596496582}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.652913696145358e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.652913696145358e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.065155029296875, "val_fraction": 2.378234398782344e-05, "val_loss": 3.688217878341675, "within_bucket_gap": -2.3769371509552}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.136007785797119, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.021602869033813, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.3972859382629395, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003999282621538984, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003999282621538984, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.987931489944458, "val_fraction": 4.756468797564688e-05, "val_loss": 11.396023273468018, "within_bucket_gap": 8.40809178352356}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.2690083980560303, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.613632917404175, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.7577383518218994, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.210895299911499, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.171189308166504, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.6495710798231616e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.6495710798231616e-05, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.2780373096466064, "val_fraction": 2.378234398782344e-05, "val_loss": 2.7434656620025635, "within_bucket_gap": -1.534571647644043}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.630376815795898, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.19689758867025375, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.734764273036016e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.734764273036016e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.3236225843429565, "val_fraction": 2.378234398782344e-05, "val_loss": 2.8940162658691406, "within_bucket_gap": 1.570393681526184}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.2755388311780933e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.2755388311780933e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.9277150630950928, "val_fraction": 2.378234398782344e-05, "val_loss": 3.391376495361328, "within_bucket_gap": -0.5363385677337646}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.8788352012634277, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.539588451385498, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.573365926742554, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.786911964416504, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.747958183288574, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.642435073852539, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.9045321345329285, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.8481919300973504e-06, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.8481919300973504e-06, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.08654522895813, "val_fraction": 2.378234398782344e-05, "val_loss": 2.1642580032348633, "within_bucket_gap": 0.0777127742767334}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.480955123901367, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.694760618144519e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.694760618144519e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.169620513916016, "val_fraction": 2.378234398782344e-05, "val_loss": 2.9341275691986084, "within_bucket_gap": -3.2354929447174072}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013978955657696252, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013978955657696252, "step": 1400, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.839866638183594, "val_fraction": 2.378234398782344e-05, "val_loss": 1.9619953632354736, "within_bucket_gap": -5.87787127494812}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.154332008173114e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.154332008173114e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.265457630157471, "val_fraction": 2.378234398782344e-05, "val_loss": 6.591791152954102, "within_bucket_gap": 1.3263335227966309}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.7314863204956055, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.3674928309166267e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.3674928309166267e-05, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.736274003982544, "val_fraction": 2.378234398782344e-05, "val_loss": 7.311277389526367, "within_bucket_gap": 0.5750033855438232}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.346090316772461, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.21319104731082916, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.1569979041814804, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.854758262634277, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.826519012451172, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.7953336238861084, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.448773384094238, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.10218650102615356, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.213837623596191, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.904439926147461, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.891854496307024e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.891854496307024e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.1962995529174805, "val_fraction": 2.378234398782344e-05, "val_loss": 5.673706531524658, "within_bucket_gap": 2.4774069786071777}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.436831951141357, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.871367454528809, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.985066733585282e-06, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.985066733585282e-06, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.745811939239502, "val_fraction": 2.378234398782344e-05, "val_loss": 2.039520025253296, "within_bucket_gap": 0.29370808601379395}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014391527730897862, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014391527730897862, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.6073485016822815, "val_fraction": 2.378234398782344e-05, "val_loss": 6.658698081970215, "within_bucket_gap": 6.051349580287933}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.493273735046387, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.42068338394165, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.03833331912755966, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016960902267122921, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016960902267122921, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.530840277671814, "val_fraction": 2.378234398782344e-05, "val_loss": 9.66256046295166, "within_bucket_gap": 7.131720185279846}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014302249617954003, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014302249617954003, "step": 1400, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.539737343788147, "val_fraction": 4.756468797564688e-05, "val_loss": 2.532832384109497, "within_bucket_gap": -3.00690495967865}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.038118362426758, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.743629455566406, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.1756017208099365, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.10403671115636826, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.06729208678007126, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.022862069308757782, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.6736836433410645, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.002226970624178648, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.7045113956910108e-05, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.7045113956910108e-05, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6983680725097656, "val_fraction": 2.378234398782344e-05, "val_loss": 4.415081024169922, "within_bucket_gap": 0.7167129516601562}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.090306282043457, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.99301528930664, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.462105210853494e-06, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.462105210853494e-06, "step": 1400, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.2255626767873764, "val_fraction": 2.378234398782344e-05, "val_loss": 0.1640840768814087, "within_bucket_gap": -0.06147859990596771}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.544029851357596e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.544029851357596e-05, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.7786461114883423, "val_fraction": 4.756468797564688e-05, "val_loss": 0.19258927553892136, "within_bucket_gap": -1.586056835949421}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011728830987097102, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011728830987097102, "step": 1400, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.281866073608398, "val_fraction": 9.512937595129375e-05, "val_loss": 4.048931360244751, "within_bucket_gap": -1.2329347133636475}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.01910074874924339, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.01910074874924339, "step": 1600, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 5.184180943438392, "val_fraction": 0.05995528919330289, "val_loss": 4.8655977290757635, "within_bucket_gap": -0.3185832143626284}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007549783190694031, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007549783190694031, "step": 1600, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 5.063625606544459, "val_fraction": 0.019882039573820395, "val_loss": 5.025652725730787, "within_bucket_gap": -0.03797288081367256}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0009781548817035727, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0009781548817035727, "step": 1600, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 4.961237664473685, "val_fraction": 0.011582001522070015, "val_loss": 4.876782928404133, "within_bucket_gap": -0.084454736069552}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005309455509150691, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005309455509150691, "step": 1600, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 4.787883557771382, "val_fraction": 0.007301179604261796, "val_loss": 4.715163041403705, "within_bucket_gap": -0.07272051636767696}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006720210500685225, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006720210500685225, "step": 1600, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 4.70257568359375, "val_fraction": 0.004827815829528158, "val_loss": 4.841773423068042, "within_bucket_gap": 0.13919773947429182}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00043095592874043206, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00043095592874043206, "step": 1600, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 5.095466343223626, "val_fraction": 0.003305745814307458, "val_loss": 4.965100624578462, "within_bucket_gap": -0.13036571864516322}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0017938453297145994, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0017938453297145994, "step": 1600, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 5.287117665902728, "val_fraction": 0.002877663622526636, "val_loss": 4.6637490012428975, "within_bucket_gap": -0.6233686646598304}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003286290878989346, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003286290878989346, "step": 1600, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 4.972440447126116, "val_fraction": 0.002449581430745814, "val_loss": 4.838283205495297, "within_bucket_gap": -0.13415724163081943}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00021707838518969055, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00021707838518969055, "step": 1600, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 4.741466625316723, "val_fraction": 0.0025684931506849314, "val_loss": 4.656950774016203, "within_bucket_gap": -0.08451585130051953}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006333117908220287, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006333117908220287, "step": 1600, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 4.5369866818797835, "val_fraction": 0.0014745053272450532, "val_loss": 4.966494652532762, "within_bucket_gap": 0.4295079706529785}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0013733089861252686, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0013733089861252686, "step": 1600, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 4.9679241441700555, "val_fraction": 0.0020214992389649923, "val_loss": 4.288572423598346, "within_bucket_gap": -0.6793517205717094}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00035418981064722825, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00035418981064722825, "step": 1600, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 4.301014031304254, "val_fraction": 0.0016647640791476407, "val_loss": 4.088257271902902, "within_bucket_gap": -0.2127567594013522}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006228078179056755, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006228078179056755, "step": 1600, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 4.673589369829963, "val_fraction": 0.0018074581430745813, "val_loss": 5.018165989925987, "within_bucket_gap": 0.3445766200960243}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008332221080456587, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008332221080456587, "step": 1600, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 3.9842527052935433, "val_fraction": 0.0015220700152207, "val_loss": 4.531679630279541, "within_bucket_gap": 0.5474269249859978}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00012302822103231587, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00012302822103231587, "step": 1600, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 4.765264613994424, "val_fraction": 0.0014745053272450532, "val_loss": 4.681827668220766, "within_bucket_gap": -0.08343694577365834}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000328467498268348, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000328467498268348, "step": 1600, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 4.7872841413630995, "val_fraction": 0.0010226407914764079, "val_loss": 5.108479521995367, "within_bucket_gap": 0.3211953806322674}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010524806019551594, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010524806019551594, "step": 1600, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 4.962878019913383, "val_fraction": 0.0004280821917808219, "val_loss": 4.717018551296658, "within_bucket_gap": -0.24585946861672525}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00022166596757598742, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00022166596757598742, "step": 1600, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.024568883384147, "val_fraction": 0.0007372526636225266, "val_loss": 4.7239040251701105, "within_bucket_gap": -0.3006648582140361}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00026318662242802334, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00026318662242802334, "step": 1600, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.793941497802734, "val_fraction": 0.000594558599695586, "val_loss": 5.236600341796875, "within_bucket_gap": 0.442658843994141}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007467023132417118, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007467023132417118, "step": 1600, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 4.9493560791015625, "val_fraction": 0.0007372526636225266, "val_loss": 5.9621734619140625, "within_bucket_gap": 1.0128173828125}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00046398287675877874, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00046398287675877874, "step": 1600, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 4.737934112548828, "val_fraction": 0.0006183409436834094, "val_loss": 5.488301497239333, "within_bucket_gap": 0.750367384690505}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003944287924280269, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003944287924280269, "step": 1600, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 5.182967867170062, "val_fraction": 0.0006659056316590563, "val_loss": 5.775287219456264, "within_bucket_gap": 0.5923193522862027}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006813182009318829, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006813182009318829, "step": 1600, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 3.833530604839325, "val_fraction": 0.00047564687975646877, "val_loss": 5.265933990478516, "within_bucket_gap": 1.4324033856391907}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003450779228994292, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003450779228994292, "step": 1600, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 3.3801778411865233, "val_fraction": 0.0003091704718417047, "val_loss": 4.496319110576923, "within_bucket_gap": 1.1161412693904}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006647852256936232, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006647852256936232, "step": 1600, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.6029468263898576, "val_fraction": 0.000594558599695586, "val_loss": 4.721062393188476, "within_bucket_gap": 1.1181155667986187}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00039834249055911784, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00039834249055911784, "step": 1600, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.891603310902913, "val_fraction": 0.0004280821917808219, "val_loss": 4.961075252956814, "within_bucket_gap": -0.9305280579460993}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008855096756349971, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008855096756349971, "step": 1600, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 3.3017392999985637, "val_fraction": 0.000594558599695586, "val_loss": 4.791095733642578, "within_bucket_gap": 1.4893564336440144}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002009792167549559, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002009792167549559, "step": 1600, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.705310474742543, "val_fraction": 0.00045186453576864533, "val_loss": 5.150088059274774, "within_bucket_gap": 0.4447775845322308}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009566026861364328, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009566026861364328, "step": 1600, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 2.319998833440965, "val_fraction": 0.00040429984779299846, "val_loss": 4.686071171480067, "within_bucket_gap": 2.3660723380391016}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004889090064615037, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004889090064615037, "step": 1600, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.168509748246935, "val_fraction": 0.00033295281582952814, "val_loss": 5.636913027082171, "within_bucket_gap": 1.4684032788352361}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005310727467638535, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005310727467638535, "step": 1600, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.643503665924072, "val_fraction": 0.0002616057838660578, "val_loss": 4.613453951748935, "within_bucket_gap": -2.0300497141751377}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0015388381686808192, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0015388381686808192, "step": 1600, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 6.250768343607585, "val_fraction": 0.0005232115677321156, "val_loss": 3.3096289201216265, "within_bucket_gap": -2.9411394234859585}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016084413565336158, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016084413565336158, "step": 1600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.951882203420003, "val_fraction": 0.0002616057838660578, "val_loss": 4.337048183787953, "within_bucket_gap": -0.6148340196320499}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003200811307966133, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003200811307966133, "step": 1600, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.195410410563151, "val_fraction": 0.0002616057838660578, "val_loss": 5.418935082175515, "within_bucket_gap": 1.2235246716123633}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00032564870661615057, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00032564870661615057, "step": 1600, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 4.008802890777588, "val_fraction": 0.00014269406392694063, "val_loss": 6.290949026743571, "within_bucket_gap": 2.2821461359659834}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.1505550581570713e-05, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.1505550581570713e-05, "step": 1600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.577933073043823, "val_fraction": 0.0002616057838660578, "val_loss": 6.621913563121449, "within_bucket_gap": 0.043980490077625944}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.284685241298152e-05, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.284685241298152e-05, "step": 1600, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 4.3292322892409105, "val_fraction": 0.00028538812785388126, "val_loss": 4.214136918385823, "within_bucket_gap": -0.11509537085508725}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006146420876366541, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006146420876366541, "step": 1600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.810796578725179, "val_fraction": 0.0003567351598173516, "val_loss": 5.08783187866211, "within_bucket_gap": -1.722964700063069}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004884995376471031, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004884995376471031, "step": 1600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.172868172327678, "val_fraction": 0.0002616057838660578, "val_loss": 6.040179859508168, "within_bucket_gap": 1.8673116871804902}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001659382011479026, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001659382011479026, "step": 1600, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 3.0032162232832476, "val_fraction": 9.512937595129375e-05, "val_loss": 4.74755859375, "within_bucket_gap": 1.7443423704667524}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003394936408444989, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003394936408444989, "step": 1600, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.6899971008300785, "val_fraction": 0.00021404109589041095, "val_loss": 6.2761113908555775, "within_bucket_gap": 1.586114290025499}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001584425798774098, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001584425798774098, "step": 1600, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 2.75420585235992, "val_fraction": 0.00014269406392694063, "val_loss": 3.864571452140808, "within_bucket_gap": 1.110365599780888}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004235266291283866, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004235266291283866, "step": 1600, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 2.9753577368600026, "val_fraction": 0.00023782343987823439, "val_loss": 4.756202507019043, "within_bucket_gap": 1.7808447701590402}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007072191010689989, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007072191010689989, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 1.1692299842834473, "val_fraction": 0.0002616057838660578, "val_loss": 3.8726071444424717, "within_bucket_gap": 2.7033771601590244}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.542973527806713e-05, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.542973527806713e-05, "step": 1600, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.8487091064453125, "val_fraction": 0.00016647640791476407, "val_loss": 5.421941893441336, "within_bucket_gap": 0.5732327869960239}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000354377238173463, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000354377238173463, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.78560733795166, "val_fraction": 0.000380517503805175, "val_loss": 3.8543039560317993, "within_bucket_gap": -0.9313033819198608}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012694745381051082, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012694745381051082, "step": 1600, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.065747669764927, "val_fraction": 7.134703196347032e-05, "val_loss": 6.845043182373047, "within_bucket_gap": 1.7792955126081198}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010147719690788826, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010147719690788826, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.6235328912734985, "val_fraction": 4.756468797564688e-05, "val_loss": 2.4900763034820557, "within_bucket_gap": -2.133456587791443}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010866755701690509, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010866755701690509, "step": 1600, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 1.4275025165442266, "val_fraction": 4.756468797564688e-05, "val_loss": 3.712129235267639, "within_bucket_gap": 2.2846267187234126}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.8056484445160467e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.8056484445160467e-05, "step": 1600, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 5.0297253348610615, "val_fraction": 0.00016647640791476407, "val_loss": 5.198256628853934, "within_bucket_gap": 0.16853129399287248}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004892390635278489, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004892390635278489, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.679327011108398, "val_fraction": 7.134703196347032e-05, "val_loss": 1.8221522967020671, "within_bucket_gap": -6.857174714406331}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.693528525324332e-05, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.693528525324332e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.7974863182753325, "val_fraction": 4.756468797564688e-05, "val_loss": 1.390238881111145, "within_bucket_gap": -1.4072474371641874}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.233994637463087e-05, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.233994637463087e-05, "step": 1600, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.923777673925672, "val_fraction": 0.00014269406392694063, "val_loss": 5.276659329732259, "within_bucket_gap": -0.6471183441934132}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.2316754113775045e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.2316754113775045e-05, "step": 1600, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 2.558294892311096, "val_fraction": 2.378234398782344e-05, "val_loss": 3.4966697692871094, "within_bucket_gap": 0.9383748769760132}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.730820324867281e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.730820324867281e-05, "step": 1600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.494907569885254, "val_fraction": 0.00016647640791476407, "val_loss": 4.9704596655709405, "within_bucket_gap": -0.5244479043143135}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.486392085773569e-05, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.486392085773569e-05, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.916365305582683, "val_fraction": 0.00014269406392694063, "val_loss": 5.300851662953694, "within_bucket_gap": 0.3844863573710118}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00037141841704442616, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00037141841704442616, "step": 1600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.0285866260528564, "val_fraction": 0.00011891171993911719, "val_loss": 1.9051063060760498, "within_bucket_gap": -3.1234803199768066}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.9402157647061745e-05, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.9402157647061745e-05, "step": 1600, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 4.040852665901184, "val_fraction": 9.512937595129375e-05, "val_loss": 4.560168147087097, "within_bucket_gap": 0.5193154811859131}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005449250960398422, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005449250960398422, "step": 1600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 3.8614834944407144, "val_fraction": 0.00023782343987823439, "val_loss": 6.152784538269043, "within_bucket_gap": 2.2913010438283288}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000617801588272486, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000617801588272486, "step": 1600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.370895767211914, "val_fraction": 0.00021404109589041095, "val_loss": 3.48452674680286, "within_bucket_gap": -2.8863690204090546}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001273041582543012, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001273041582543012, "step": 1600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.692527770996094, "val_fraction": 0.00011891171993911719, "val_loss": 5.621950721740722, "within_bucket_gap": -1.0705770492553714}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.029059716374178e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.029059716374178e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.440010070800781, "val_fraction": 7.134703196347032e-05, "val_loss": 6.285043080647786, "within_bucket_gap": 0.8450330098470049}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00024131747087409864, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00024131747087409864, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.272804826498032, "val_fraction": 0.00016647640791476407, "val_loss": 4.823245252881732, "within_bucket_gap": -1.4495595736163}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.9234378799939802e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.9234378799939802e-05, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.617387930552165, "val_fraction": 0.0001902587519025875, "val_loss": 6.516292035579681, "within_bucket_gap": -0.10109589497248361}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015396544854390562, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015396544854390562, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.883829355239868, "val_fraction": 0.0002616057838660578, "val_loss": 4.472369280728427, "within_bucket_gap": 0.5885399254885586}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.548298276178369e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.548298276178369e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.570097327232361, "val_fraction": 4.756468797564688e-05, "val_loss": 4.193383097648621, "within_bucket_gap": -1.3767142295837402}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025190017505051336, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025190017505051336, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 7.164538025856018, "val_fraction": 7.134703196347032e-05, "val_loss": 3.6339051723480225, "within_bucket_gap": -3.5306328535079956}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.587278782686323e-06, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.587278782686323e-06, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.113539218902588, "val_fraction": 4.756468797564688e-05, "val_loss": 7.038120269775391, "within_bucket_gap": -0.07541894912719727}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00025787945025950625, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00025787945025950625, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.402950763702393, "val_fraction": 9.512937595129375e-05, "val_loss": 7.113779544830322, "within_bucket_gap": 2.7108287811279297}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011279621080720806, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011279621080720806, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.488954544067383, "val_fraction": 9.512937595129375e-05, "val_loss": 3.674668312072754, "within_bucket_gap": 1.185713768005371}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010590474355166362, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010590474355166362, "step": 1600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.671195220947266, "val_fraction": 7.134703196347032e-05, "val_loss": 3.1868343353271484, "within_bucket_gap": -1.4843608856201174}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004254108921396496, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004254108921396496, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.52310049533844, "val_fraction": 0.0001902587519025875, "val_loss": 3.7590601444244385, "within_bucket_gap": 2.2359596490859985}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.1994391759599543e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.1994391759599543e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.9930384159088135, "val_fraction": 2.378234398782344e-05, "val_loss": 3.068218231201172, "within_bucket_gap": -0.9248201847076416}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019365911977294734, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019365911977294734, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.0887675285339355, "val_fraction": 7.134703196347032e-05, "val_loss": 4.8030937512715655, "within_bucket_gap": 2.71432622273763}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.716081242673835e-05, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.716081242673835e-05, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.6913241147994995, "val_fraction": 0.00011891171993911719, "val_loss": 3.3402156829833984, "within_bucket_gap": 0.6488915681838989}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002876726210208063, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002876726210208063, "step": 1600, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.692478656768799, "val_fraction": 7.134703196347032e-05, "val_loss": 8.72449811299642, "within_bucket_gap": 4.032019456227621}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019381706754548, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019381706754548, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.317363977432251, "val_fraction": 9.512937595129375e-05, "val_loss": 6.354768991470337, "within_bucket_gap": 2.037405014038086}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.96632194519043, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.890948880934461e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.890948880934461e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.940931558609009, "val_fraction": 4.756468797564688e-05, "val_loss": 5.333138465881348, "within_bucket_gap": -0.6077930927276611}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015742800097850177, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015742800097850177, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.492583751678467, "val_fraction": 4.756468797564688e-05, "val_loss": 2.1828174591064453, "within_bucket_gap": -3.3097662925720215}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.658222079277039, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.450713217167731e-05, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.450713217167731e-05, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.0164616107940674, "val_fraction": 0.0001902587519025875, "val_loss": 3.3029510974884033, "within_bucket_gap": 0.28648948669433594}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00034237727863059195, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00034237727863059195, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.488503456115723, "val_fraction": 7.134703196347032e-05, "val_loss": 5.689743518829346, "within_bucket_gap": -4.798759937286377}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002054073964777194, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002054073964777194, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.57819128036499, "val_fraction": 4.756468797564688e-05, "val_loss": 2.259706176817417, "within_bucket_gap": -4.318485103547573}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.971737338709141e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.971737338709141e-05, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.759108543395996, "val_fraction": 9.512937595129375e-05, "val_loss": 4.026239514350891, "within_bucket_gap": -0.732869029045105}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.50777775620761e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.50777775620761e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.77503776550293, "val_fraction": 9.512937595129375e-05, "val_loss": 3.7755801677703857, "within_bucket_gap": -0.999457597732544}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.0036443297721475e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.0036443297721475e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4093248844146729, "val_fraction": 2.378234398782344e-05, "val_loss": 3.0927772521972656, "within_bucket_gap": 1.6834523677825928}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.376300938053218e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.376300938053218e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.64805269241333, "val_fraction": 2.378234398782344e-05, "val_loss": 3.4881997108459473, "within_bucket_gap": 1.8401470184326172}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019592097391034672, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019592097391034672, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.394011815388997, "val_fraction": 4.756468797564688e-05, "val_loss": 8.513054370880127, "within_bucket_gap": 4.11904255549113}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00015691990993882965, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00015691990993882965, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.412571907043457, "val_fraction": 2.378234398782344e-05, "val_loss": 10.010740280151367, "within_bucket_gap": 6.59816837310791}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.1679505808349794e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.1679505808349794e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.761078834533691, "val_fraction": 4.756468797564688e-05, "val_loss": 6.4271087646484375, "within_bucket_gap": 0.6660299301147461}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017540612721552042, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017540612721552042, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.9105415344238281, "val_fraction": 2.378234398782344e-05, "val_loss": 8.286018371582031, "within_bucket_gap": 7.375476837158203}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001820254951851553, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001820254951851553, "step": 1600, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.067202186584472, "val_fraction": 2.378234398782344e-05, "val_loss": 11.721010208129883, "within_bucket_gap": 7.6538080215454105}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.662608473404357e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.662608473404357e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.0354168713092804, "val_fraction": 4.756468797564688e-05, "val_loss": 4.8566436767578125, "within_bucket_gap": 1.821226805448532}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006088529638322339, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006088529638322339, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.61686897277832, "val_fraction": 7.134703196347032e-05, "val_loss": 4.0831858317057295, "within_bucket_gap": -8.53368314107259}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.1112863620122275, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.929386138916016, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.0095126567789294e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.0095126567789294e-05, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.521255095799764, "val_fraction": 2.378234398782344e-05, "val_loss": 4.945734977722168, "within_bucket_gap": 0.42447988192240427}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.2086237668991089, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.628490447998047, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.0413969647213022e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.0413969647213022e-05, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.8471916913986206, "val_fraction": 2.378234398782344e-05, "val_loss": 3.4093050956726074, "within_bucket_gap": -0.4378865957260132}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.2761203050613403, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.201326847076416, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.900017738342285, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002983332114773299, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002983332114773299, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.1132070943713188, "val_fraction": 4.756468797564688e-05, "val_loss": 7.385364532470703, "within_bucket_gap": 6.272157438099384}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.387940688764668e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.387940688764668e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.651472091674805, "val_fraction": 2.378234398782344e-05, "val_loss": 4.965470790863037, "within_bucket_gap": -2.6860013008117676}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010091285511873992, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010091285511873992, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.8192983865737915, "val_fraction": 7.134703196347032e-05, "val_loss": 2.2336929639180503, "within_bucket_gap": 1.4143945773442588}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.749722480773926, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00020715966678963043, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00020715966678963043, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.4528378546237946, "val_fraction": 0.00014269406392694063, "val_loss": 2.0010629097620645, "within_bucket_gap": -1.45177494486173}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.319187283334485e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.319187283334485e-05, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.121855020523071, "val_fraction": 4.756468797564688e-05, "val_loss": 5.6606409549713135, "within_bucket_gap": 1.5387859344482422}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.605352547615086e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.605352547615086e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.6671066284179688, "val_fraction": 2.378234398782344e-05, "val_loss": 6.28548526763916, "within_bucket_gap": 3.6183786392211914}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015400904634771826, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015400904634771826, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.591323852539062, "val_fraction": 2.378234398782344e-05, "val_loss": 2.115551471710205, "within_bucket_gap": -6.475772380828857}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.9080472569487406e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.9080472569487406e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.963197231292725, "val_fraction": 2.378234398782344e-05, "val_loss": 8.606452941894531, "within_bucket_gap": 1.6432557106018066}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.0392165184021, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.5229007005691528, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.5360540871932265e-06, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.5360540871932265e-06, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.9058079719543457, "val_fraction": 4.756468797564688e-05, "val_loss": 4.001173973083496, "within_bucket_gap": 0.09536600112915039}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.829782485961914, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.3832836151123047, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.1392297744750977, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.8739470640818277, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.585959522147156e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.585959522147156e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.1846153736114502, "val_fraction": 2.378234398782344e-05, "val_loss": 5.215319633483887, "within_bucket_gap": 4.0307042598724365}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.032401084899902, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.8294700384140015, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.028453104197978973, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.413623809814453, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010697870221856522, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010697870221856522, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.823856353759766, "val_fraction": 4.756468797564688e-05, "val_loss": 4.57473611831665, "within_bucket_gap": -2.2491202354431152}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.3227014541625977, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.266050934791565, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.011367950588464737, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.0391813525689066e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.0391813525689066e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.376025676727295, "val_fraction": 2.378234398782344e-05, "val_loss": 3.8129806518554688, "within_bucket_gap": 0.43695497512817383}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.067259534853234e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.067259534853234e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.876500606536865, "val_fraction": 2.378234398782344e-05, "val_loss": 8.427661895751953, "within_bucket_gap": 2.551161289215088}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.924621205170223e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.924621205170223e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.142243385314941, "val_fraction": 2.378234398782344e-05, "val_loss": 3.230578660964966, "within_bucket_gap": -2.9116647243499756}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.996956825256348, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.212335348129272, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.107361793518066, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00040412411842172003, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00040412411842172003, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.8285160064697266, "val_fraction": 4.756468797564688e-05, "val_loss": 11.324821472167969, "within_bucket_gap": 8.496305465698242}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.09767154604196548, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.341063022613525, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.20117449760437, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.066474676132202, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.876626968383789, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.080958278846886e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.080958278846886e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.182125449180603, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4661641120910645, "within_bucket_gap": -1.7159613370895386}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.59402847290039, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.19755205512046814, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.7135165903938415e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.7135165903938415e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.9564631581306458, "val_fraction": 2.378234398782344e-05, "val_loss": 1.6769626140594482, "within_bucket_gap": 0.7204994559288025}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.095752710983996e-06, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.095752710983996e-06, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.2844090461730957, "val_fraction": 2.378234398782344e-05, "val_loss": 2.9019508361816406, "within_bucket_gap": -0.3824582099914551}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.661834955215454, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.713563919067383, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.250598073005676, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.437734603881836, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.5470733642578125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.857312202453613, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.149411678314209, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.85629996489535e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.85629996489535e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7689361572265625, "val_fraction": 2.378234398782344e-05, "val_loss": 1.9883991479873657, "within_bucket_gap": -0.7805370092391968}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.597492694854736, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.186986003655095e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.186986003655095e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.965512275695801, "val_fraction": 2.378234398782344e-05, "val_loss": 2.5230484008789062, "within_bucket_gap": -3.4424638748168945}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014945601960355894, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014945601960355894, "step": 1600, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.459071795145671, "val_fraction": 2.378234398782344e-05, "val_loss": 1.1747450828552246, "within_bucket_gap": -6.284326712290446}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.5669240073344842e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.5669240073344842e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.724761962890625, "val_fraction": 2.378234398782344e-05, "val_loss": 6.383622169494629, "within_bucket_gap": 0.6588602066040039}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.149839878082275, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.0300413806507393e-05, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.0300413806507393e-05, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.634041428565979, "val_fraction": 2.378234398782344e-05, "val_loss": 7.487633228302002, "within_bucket_gap": 0.853591799736023}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.57603931427002, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.135956272482872, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.05148407444357872, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.652565002441406, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.571736335754395, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.027827262878418, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.682211875915527, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.01636596955358982, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.679182052612305, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.779015302658081, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.246595092560058e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.246595092560058e-05, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.6801064014434814, "val_fraction": 2.378234398782344e-05, "val_loss": 4.886194705963135, "within_bucket_gap": 2.2060883045196533}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.896736145019531, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.306541919708252, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.4580627463905415e-05, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.4580627463905415e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4348747730255127, "val_fraction": 2.378234398782344e-05, "val_loss": 2.0479609966278076, "within_bucket_gap": 0.6130862236022949}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016039525636413687, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016039525636413687, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.47317805886268616, "val_fraction": 2.378234398782344e-05, "val_loss": 7.217477798461914, "within_bucket_gap": 6.744299739599228}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.96585750579834, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.738405227661133, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.040943484753370285, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00019988109736137738, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00019988109736137738, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.958770513534546, "val_fraction": 2.378234398782344e-05, "val_loss": 10.363370895385742, "within_bucket_gap": 8.404600381851196}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001380568883371861, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001380568883371861, "step": 1600, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.344443082809448, "val_fraction": 4.756468797564688e-05, "val_loss": 2.4419350624084473, "within_bucket_gap": -2.902508020401001}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.5067710876464844, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.426513195037842, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.862800121307373, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.19997365772724152, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.06894221156835556, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.17372342944145203, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.915204048156738, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.00046588058467023075, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.11546216163461e-06, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.11546216163461e-06, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6255593299865723, "val_fraction": 2.378234398782344e-05, "val_loss": 4.008846282958984, "within_bucket_gap": 0.3832869529724121}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7502663135528564, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.709528923034668, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.2633713119106206e-07, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.2633713119106206e-07, "step": 1600, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.1783922016620636, "val_fraction": 2.378234398782344e-05, "val_loss": 0.19211402535438538, "within_bucket_gap": 0.013721823692321777}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.73890393085585e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.73890393085585e-05, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5924649238586426, "val_fraction": 4.756468797564688e-05, "val_loss": 0.17567776143550873, "within_bucket_gap": -1.4167871624231339}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00016037655974450538, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00016037655974450538, "step": 1600, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.779168605804443, "val_fraction": 9.512937595129375e-05, "val_loss": 4.093290209770203, "within_bucket_gap": -1.6858783960342407}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.01971097400203556, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.01971097400203556, "step": 1800, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 5.05931093024597, "val_fraction": 0.05995528919330289, "val_loss": 4.730549710556327, "within_bucket_gap": -0.32876121968964345}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00042177920439021726, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00042177920439021726, "step": 1800, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 4.876287448936214, "val_fraction": 0.019882039573820395, "val_loss": 4.89750153025942, "within_bucket_gap": 0.02121408132320557}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0011402222232196676, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0011402222232196676, "step": 1800, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 4.801633429276316, "val_fraction": 0.011582001522070015, "val_loss": 4.7031856591696615, "within_bucket_gap": -0.09844777010665418}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005149507297209382, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005149507297209382, "step": 1800, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 4.642508737664474, "val_fraction": 0.007301179604261796, "val_loss": 4.57197893869605, "within_bucket_gap": -0.07052979896842348}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009291612492171385, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009291612492171385, "step": 1800, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 4.505706307775688, "val_fraction": 0.004827815829528158, "val_loss": 4.69816626938693, "within_bucket_gap": 0.19245996161124257}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002722734562703681, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002722734562703681, "step": 1800, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 4.852145796971963, "val_fraction": 0.003305745814307458, "val_loss": 4.934509496894671, "within_bucket_gap": 0.0823636999227082}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0015525411979934454, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0015525411979934454, "step": 1800, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 5.073677018887735, "val_fraction": 0.002877663622526636, "val_loss": 4.534162537125517, "within_bucket_gap": -0.5395144817622182}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00048514092505094666, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00048514092505094666, "step": 1800, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 4.878898893083845, "val_fraction": 0.002449581430745814, "val_loss": 4.680848353117415, "within_bucket_gap": -0.19805053996642918}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.3112090106581307e-06, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.3112090106581307e-06, "step": 1800, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 4.52736285785297, "val_fraction": 0.0025684931506849314, "val_loss": 4.526852360478154, "within_bucket_gap": -0.0005104973748162323}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006127293866111643, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006127293866111643, "step": 1800, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 4.418251660405373, "val_fraction": 0.0014745053272450532, "val_loss": 4.8338007773122476, "within_bucket_gap": 0.4155491169068748}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001096693052342495, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001096693052342495, "step": 1800, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 4.755018469405501, "val_fraction": 0.0020214992389649923, "val_loss": 4.212503769818475, "within_bucket_gap": -0.5425146995870263}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005718695821041138, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005718695821041138, "step": 1800, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 4.259468587239583, "val_fraction": 0.0016647640791476407, "val_loss": 3.9159546988351006, "within_bucket_gap": -0.3435138884044826}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008381785632610024, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008381785632610024, "step": 1800, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 4.439345864688649, "val_fraction": 0.0018074581430745813, "val_loss": 4.903079183478105, "within_bucket_gap": 0.4637333187894557}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007616217363295893, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007616217363295893, "step": 1800, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 3.8690617505241844, "val_fraction": 0.0015220700152207, "val_loss": 4.369447231292725, "within_bucket_gap": 0.5003854807685402}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011076188726962503, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011076188726962503, "step": 1800, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 4.404315616085322, "val_fraction": 0.0014745053272450532, "val_loss": 4.4794336134387605, "within_bucket_gap": 0.0751179973534386}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005412625214099158, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005412625214099158, "step": 1800, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 4.5745715651401255, "val_fraction": 0.0010226407914764079, "val_loss": 5.103850786076036, "within_bucket_gap": 0.5292792209359103}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00017020997564999794, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00017020997564999794, "step": 1800, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 4.91804371709409, "val_fraction": 0.0004280821917808219, "val_loss": 4.520433213975695, "within_bucket_gap": -0.3976105031183952}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000426595587283788, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000426595587283788, "step": 1800, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.027560536454364, "val_fraction": 0.0007372526636225266, "val_loss": 4.4489317863218245, "within_bucket_gap": -0.5786287501325393}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017367258658225625, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017367258658225625, "step": 1800, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.778490475245884, "val_fraction": 0.000594558599695586, "val_loss": 5.0705938720703125, "within_bucket_gap": 0.29210339682442843}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008433243978806538, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008433243978806538, "step": 1800, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 4.823791063748873, "val_fraction": 0.0007372526636225266, "val_loss": 5.967665395429058, "within_bucket_gap": 1.143874331680185}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005293538487641772, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005293538487641772, "step": 1800, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 4.609517743510585, "val_fraction": 0.0006183409436834094, "val_loss": 5.465605075542744, "within_bucket_gap": 0.8560873320321587}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004795491241791842, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004795491241791842, "step": 1800, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.903674261910575, "val_fraction": 0.0006659056316590563, "val_loss": 5.62382003239223, "within_bucket_gap": 0.720145770481655}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000625062767909542, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000625062767909542, "step": 1800, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 3.78600937128067, "val_fraction": 0.00047564687975646877, "val_loss": 5.100141334533691, "within_bucket_gap": 1.314131963253021}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003071658125025315, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003071658125025315, "step": 1800, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 3.1822003936767578, "val_fraction": 0.0003091704718417047, "val_loss": 4.175716400146484, "within_bucket_gap": 0.9935160064697266}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007015031692151738, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007015031692151738, "step": 1800, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.406806332724435, "val_fraction": 0.000594558599695586, "val_loss": 4.58667854309082, "within_bucket_gap": 1.1798722103663852}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0003540265959328894, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0003540265959328894, "step": 1800, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.6723833084106445, "val_fraction": 0.0004280821917808219, "val_loss": 4.845377180311415, "within_bucket_gap": -0.8270061280992298}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009179839011459951, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009179839011459951, "step": 1800, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 3.2036226309982, "val_fraction": 0.000594558599695586, "val_loss": 4.747598114013672, "within_bucket_gap": 1.5439754830154722}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00043026906863046714, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00043026906863046714, "step": 1800, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.308643514459783, "val_fraction": 0.00045186453576864533, "val_loss": 5.260851609079461, "within_bucket_gap": 0.952208094619678}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.001068429788105874, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.001068429788105874, "step": 1800, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 1.9575374357161983, "val_fraction": 0.00040429984779299846, "val_loss": 4.600204243379481, "within_bucket_gap": 2.642666807663282}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003378429344094346, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003378429344094346, "step": 1800, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.2074372503492565, "val_fraction": 0.00033295281582952814, "val_loss": 5.222124372209821, "within_bucket_gap": 1.0146871218605646}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005158435927542377, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005158435927542377, "step": 1800, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.345116019248962, "val_fraction": 0.0002616057838660578, "val_loss": 4.3732804385098545, "within_bucket_gap": -1.971835580739108}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0013120871912338356, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0013120871912338356, "step": 1800, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.746355374654134, "val_fraction": 0.0005232115677321156, "val_loss": 3.2385989102450283, "within_bucket_gap": -2.5077564644091055}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002195749607623259, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002195749607623259, "step": 1800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.993893623352051, "val_fraction": 0.0002616057838660578, "val_loss": 4.154558355158025, "within_bucket_gap": -0.8393352681940254}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00030716800072784657, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00030716800072784657, "step": 1800, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 3.9747543334960938, "val_fraction": 0.0002616057838660578, "val_loss": 5.148917978460139, "within_bucket_gap": 1.174163644964045}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002831927581464864, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002831927581464864, "step": 1800, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 3.9088169733683267, "val_fraction": 0.00014269406392694063, "val_loss": 5.893431822458903, "within_bucket_gap": 1.9846148490905766}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.26279711194113e-05, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.26279711194113e-05, "step": 1800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.518842180569966, "val_fraction": 0.0002616057838660578, "val_loss": 6.279443914240057, "within_bucket_gap": -0.23939826632990968}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.345925952717987e-05, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.345925952717987e-05, "step": 1800, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 3.9456211970402646, "val_fraction": 0.00028538812785388126, "val_loss": 4.273102442423503, "within_bucket_gap": 0.3274812453832383}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006987086328560177, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006987086328560177, "step": 1800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.856922626495361, "val_fraction": 0.0003567351598173516, "val_loss": 4.898302586873372, "within_bucket_gap": -1.958620039621989}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00044803656467564025, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00044803656467564025, "step": 1800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.180000066757202, "val_fraction": 0.0002616057838660578, "val_loss": 5.8926402005282315, "within_bucket_gap": 1.7126401337710293}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001349789026665361, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001349789026665361, "step": 1800, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 3.1711034774780273, "val_fraction": 9.512937595129375e-05, "val_loss": 4.590001702308655, "within_bucket_gap": 1.4188982248306274}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003286694591266741, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003286694591266741, "step": 1800, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.448566589355469, "val_fraction": 0.00021404109589041095, "val_loss": 5.98411030239529, "within_bucket_gap": 1.5355437130398215}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018968826237928326, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018968826237928326, "step": 1800, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 2.5955318104137075, "val_fraction": 0.00014269406392694063, "val_loss": 3.9248671531677246, "within_bucket_gap": 1.329335342754017}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003693272890435999, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003693272890435999, "step": 1800, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.081052916390555, "val_fraction": 0.00023782343987823439, "val_loss": 4.634000301361084, "within_bucket_gap": 1.5529473849705289}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006760523981971828, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006760523981971828, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 1.0380299091339111, "val_fraction": 0.0002616057838660578, "val_loss": 3.622270930897106, "within_bucket_gap": 2.584241021763195}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018859863704621396, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018859863704621396, "step": 1800, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.428935368855794, "val_fraction": 0.00016647640791476407, "val_loss": 5.561820438929966, "within_bucket_gap": 1.1328850700741722}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00038740751042939394, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00038740751042939394, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.505051136016846, "val_fraction": 0.000380517503805175, "val_loss": 3.4869441986083984, "within_bucket_gap": -1.0181069374084473}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001414002776534745, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001414002776534745, "step": 1800, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 4.877665110996792, "val_fraction": 7.134703196347032e-05, "val_loss": 6.859531402587891, "within_bucket_gap": 1.981866291591099}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011087822213292665, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011087822213292665, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.4973196387290955, "val_fraction": 4.756468797564688e-05, "val_loss": 2.1662158966064453, "within_bucket_gap": -2.33110374212265}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013436209658712958, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013436209658712958, "step": 1800, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 1.2712689891006008, "val_fraction": 4.756468797564688e-05, "val_loss": 4.096097707748413, "within_bucket_gap": 2.8248287186478125}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010233839773547846, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010233839773547846, "step": 1800, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.753119121898305, "val_fraction": 0.00016647640791476407, "val_loss": 5.367851257324219, "within_bucket_gap": 0.6147321354259141}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00046847642738162294, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00046847642738162294, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.41876220703125, "val_fraction": 7.134703196347032e-05, "val_loss": 1.852596600850423, "within_bucket_gap": -6.566165606180827}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.88592215224284e-05, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.88592215224284e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.557339381426573, "val_fraction": 4.756468797564688e-05, "val_loss": 1.319883108139038, "within_bucket_gap": -1.2374562732875347}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.923701911419478e-05, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.923701911419478e-05, "step": 1800, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.820987292698452, "val_fraction": 0.00014269406392694063, "val_loss": 5.335774262746175, "within_bucket_gap": -0.485213029952277}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.3057735333703968e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.3057735333703968e-05, "step": 1800, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 2.238303542137146, "val_fraction": 2.378234398782344e-05, "val_loss": 3.2078351974487305, "within_bucket_gap": 0.9695316553115845}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.140450009472295e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.140450009472295e-05, "step": 1800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.3209222793579105, "val_fraction": 0.00016647640791476407, "val_loss": 4.89200564793178, "within_bucket_gap": -0.4289166314261301}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.412058116094163e-05, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.412058116094163e-05, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.75568167368571, "val_fraction": 0.00014269406392694063, "val_loss": 5.205038706461589, "within_bucket_gap": 0.4493570327758789}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004080627893226695, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004080627893226695, "step": 1800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.087615871429444, "val_fraction": 0.00011891171993911719, "val_loss": 1.6559710383415223, "within_bucket_gap": -3.4316448330879217}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.293596152664138e-05, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.293596152664138e-05, "step": 1800, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 3.891862392425537, "val_fraction": 9.512937595129375e-05, "val_loss": 4.553445219993591, "within_bucket_gap": 0.6615828275680542}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004949145536928356, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004949145536928356, "step": 1800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 3.931027809778849, "val_fraction": 0.00023782343987823439, "val_loss": 6.012044525146484, "within_bucket_gap": 2.081016715367635}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006812078456565896, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006812078456565896, "step": 1800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.234092903137207, "val_fraction": 0.00021404109589041095, "val_loss": 3.05148984822962, "within_bucket_gap": -3.182603054907587}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.828964687736433e-05, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.828964687736433e-05, "step": 1800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.5657398223876955, "val_fraction": 0.00011891171993911719, "val_loss": 5.739163208007812, "within_bucket_gap": -0.8265766143798832}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.206319369863341e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.206319369863341e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.316330909729004, "val_fraction": 7.134703196347032e-05, "val_loss": 5.76572863260905, "within_bucket_gap": 0.44939772288004587}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002560718821381507, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002560718821381507, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.3149144649505615, "val_fraction": 0.00016647640791476407, "val_loss": 4.7767272506441385, "within_bucket_gap": -1.538187214306423}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.485951765245262e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.485951765245262e-05, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.286732991536458, "val_fraction": 0.0001902587519025875, "val_loss": 6.680194616317749, "within_bucket_gap": 0.393461624781291}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.52005609629542e-05, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.52005609629542e-05, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 3.939656615257263, "val_fraction": 0.0002616057838660578, "val_loss": 4.188888723200018, "within_bucket_gap": 0.2492321079427544}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.120261884534014e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.120261884534014e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.678343415260315, "val_fraction": 4.756468797564688e-05, "val_loss": 4.181379556655884, "within_bucket_gap": -1.4969638586044312}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00025705639499734345, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00025705639499734345, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.579897880554199, "val_fraction": 7.134703196347032e-05, "val_loss": 2.9769954482714334, "within_bucket_gap": -3.602902432282766}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.2402115427014308e-05, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.2402115427014308e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.28240442276001, "val_fraction": 4.756468797564688e-05, "val_loss": 6.811422348022461, "within_bucket_gap": -0.47098207473754883}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00029051351492807745, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00029051351492807745, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.6386773586273193, "val_fraction": 9.512937595129375e-05, "val_loss": 6.6925554275512695, "within_bucket_gap": 3.05387806892395}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012091171886641503, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012091171886641503, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.5324771404266357, "val_fraction": 9.512937595129375e-05, "val_loss": 3.8035011291503906, "within_bucket_gap": 1.2710239887237549}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001261953530972044, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001261953530972044, "step": 1800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.67978515625, "val_fraction": 7.134703196347032e-05, "val_loss": 2.9110310872395835, "within_bucket_gap": -1.7687540690104169}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006204002410672389, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006204002410672389, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.7816420197486877, "val_fraction": 0.0001902587519025875, "val_loss": 4.042465686798096, "within_bucket_gap": 3.260823667049408}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.9999649063818713e-05, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.9999649063818713e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.3944177627563477, "val_fraction": 2.378234398782344e-05, "val_loss": 2.5534725189208984, "within_bucket_gap": -0.8409452438354492}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012811470657723135, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012811470657723135, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.2510781288146973, "val_fraction": 7.134703196347032e-05, "val_loss": 4.046733856201172, "within_bucket_gap": 1.7956557273864746}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010676655629181968, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010676655629181968, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.5329686403274536, "val_fraction": 0.00011891171993911719, "val_loss": 3.4308326721191404, "within_bucket_gap": 0.8978640317916868}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003367748616130021, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003367748616130021, "step": 1800, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.399898529052734, "val_fraction": 7.134703196347032e-05, "val_loss": 9.120134989420572, "within_bucket_gap": 4.720236460367838}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00021066835300381686, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00021066835300381686, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.107681751251221, "val_fraction": 9.512937595129375e-05, "val_loss": 6.322227478027344, "within_bucket_gap": 2.214545726776123}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.0023088455200195, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.837252657888868e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.837252657888868e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.754147291183472, "val_fraction": 4.756468797564688e-05, "val_loss": 4.316683292388916, "within_bucket_gap": -1.4374639987945557}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014946633550130069, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014946633550130069, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.2818522453308105, "val_fraction": 4.756468797564688e-05, "val_loss": 2.139472007751465, "within_bucket_gap": -3.1423802375793457}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.288137793540955, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001045954236519754, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001045954236519754, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.8727670907974243, "val_fraction": 0.0001902587519025875, "val_loss": 3.422520637512207, "within_bucket_gap": 0.5497535467147827}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0004745367319072218, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0004745367319072218, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.236374855041504, "val_fraction": 7.134703196347032e-05, "val_loss": 5.585268020629883, "within_bucket_gap": -6.651106834411621}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001884172082264855, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001884172082264855, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.160139560699463, "val_fraction": 4.756468797564688e-05, "val_loss": 2.1988561749458313, "within_bucket_gap": -3.9612833857536316}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.420587195411424e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.420587195411424e-05, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.8561312357584634, "val_fraction": 9.512937595129375e-05, "val_loss": 4.005463361740112, "within_bucket_gap": 0.1493321259816489}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.288574604138936e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.288574604138936e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.290496826171875, "val_fraction": 9.512937595129375e-05, "val_loss": 3.83968186378479, "within_bucket_gap": -0.45081496238708496}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.2958063475073196e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.2958063475073196e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.137852668762207, "val_fraction": 2.378234398782344e-05, "val_loss": 3.364633321762085, "within_bucket_gap": 2.226780652999878}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.6870034738036954e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.6870034738036954e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.4090298414230347, "val_fraction": 2.378234398782344e-05, "val_loss": 3.3798210620880127, "within_bucket_gap": 1.970791220664978}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00023006706416818588, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00023006706416818588, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.424079259236653, "val_fraction": 4.756468797564688e-05, "val_loss": 9.261009216308594, "within_bucket_gap": 4.83692995707194}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001267542781895154, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001267542781895154, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.861438274383545, "val_fraction": 2.378234398782344e-05, "val_loss": 9.191202163696289, "within_bucket_gap": 5.329763889312744}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.680379471278082e-06, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.680379471278082e-06, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.183969497680664, "val_fraction": 4.756468797564688e-05, "val_loss": 5.980449199676514, "within_bucket_gap": -0.2035202980041504}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012878871478900155, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012878871478900155, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.7642298936843872, "val_fraction": 2.378234398782344e-05, "val_loss": 6.179537773132324, "within_bucket_gap": 5.415307879447937}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017076548602845934, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017076548602845934, "step": 1800, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.001866626739502, "val_fraction": 2.378234398782344e-05, "val_loss": 11.18221378326416, "within_bucket_gap": 7.180347156524658}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.325777671108506e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.325777671108506e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.3707101345062256, "val_fraction": 4.756468797564688e-05, "val_loss": 4.700641632080078, "within_bucket_gap": 1.3299314975738525}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006041810528872763, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006041810528872763, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.204085350036621, "val_fraction": 7.134703196347032e-05, "val_loss": 3.7358837127685547, "within_bucket_gap": -8.468201637268066}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.006230334440867, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.378238677978516, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.728030810919828e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.728030810919828e-05, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.418158213297526, "val_fraction": 2.378234398782344e-05, "val_loss": 5.144760608673096, "within_bucket_gap": 0.7266023953755694}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.9566925168037415, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.396449089050293, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.4540185100048825e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.4540185100048825e-05, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.022075533866882, "val_fraction": 2.378234398782344e-05, "val_loss": 3.4106898307800293, "within_bucket_gap": -0.611385703086853}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.1307506561279297, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.567850708961487, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.518944263458252, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002975260030515095, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002975260030515095, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.014419425278902, "val_fraction": 4.756468797564688e-05, "val_loss": 7.269606113433838, "within_bucket_gap": 6.255186688154936}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -3.697517890726957e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -3.697517890726957e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.6540117263793945, "val_fraction": 2.378234398782344e-05, "val_loss": 6.099279403686523, "within_bucket_gap": -1.554732322692871}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.271168744999523e-05, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.271168744999523e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.720234751701355, "val_fraction": 7.134703196347032e-05, "val_loss": 1.7393617630004883, "within_bucket_gap": 1.0191270112991333}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.565777778625488, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00023403481964469862, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00023403481964469862, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.4313358068466187, "val_fraction": 0.00014269406392694063, "val_loss": 1.7912197907765706, "within_bucket_gap": -1.640116016070048}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.836466299707305e-05, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.836466299707305e-05, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.921117067337036, "val_fraction": 4.756468797564688e-05, "val_loss": 5.9891357421875, "within_bucket_gap": 2.068018674850464}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.22708929238254e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.22708929238254e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.6608364582061768, "val_fraction": 2.378234398782344e-05, "val_loss": 6.1201629638671875, "within_bucket_gap": 3.4593265056610107}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001399603699258836, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001399603699258836, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.11559009552002, "val_fraction": 2.378234398782344e-05, "val_loss": 2.230536460876465, "within_bucket_gap": -5.885053634643555}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.905453729121471e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.905453729121471e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.3669939041137695, "val_fraction": 2.378234398782344e-05, "val_loss": 9.009159088134766, "within_bucket_gap": 1.642165184020996}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.492615222930908, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.7896937131881714, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.354143713162914e-06, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.354143713162914e-06, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.941333293914795, "val_fraction": 4.756468797564688e-05, "val_loss": 4.053898811340332, "within_bucket_gap": 0.11256551742553711}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.546274662017822, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.5480151176452637, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.286350965499878, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.8637688954671225, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.65363733028168e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.65363733028168e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.3357517719268799, "val_fraction": 2.378234398782344e-05, "val_loss": 5.394913196563721, "within_bucket_gap": 4.059161424636841}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.439817428588867, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4287196397781372, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.05728016421198845, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.787650108337402, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.911445532759575e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.911445532759575e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.645981311798096, "val_fraction": 4.756468797564688e-05, "val_loss": 4.772439002990723, "within_bucket_gap": -1.873542308807373}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.767610788345337, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.251162648200989, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.02887685038149357, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.083996568822062e-06, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.083996568822062e-06, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.9757137298583984, "val_fraction": 2.378234398782344e-05, "val_loss": 3.8880858421325684, "within_bucket_gap": -0.08762788772583008}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.4368053440633973e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.4368053440633973e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.153316497802734, "val_fraction": 2.378234398782344e-05, "val_loss": 7.757464408874512, "within_bucket_gap": 0.6041479110717773}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.101948486796978e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.101948486796978e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.100771903991699, "val_fraction": 2.378234398782344e-05, "val_loss": 3.2735846042633057, "within_bucket_gap": -3.8271872997283936}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.5220232009887695, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.141436100006104, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.551214694976807, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003590284076999856, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003590284076999856, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.5569026470184326, "val_fraction": 4.756468797564688e-05, "val_loss": 11.10511589050293, "within_bucket_gap": 7.548213243484497}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.022735759615898132, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.187772274017334, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.317314863204956, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.3340349197387695, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.426138401031494, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.8725257044150586e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.8725257044150586e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.7500447034835815, "val_fraction": 2.378234398782344e-05, "val_loss": 2.5422050952911377, "within_bucket_gap": -1.2078396081924438}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.9533109664917, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.2414763942360878, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.6945279579002926e-06, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.6945279579002926e-06, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.5868405103683472, "val_fraction": 2.378234398782344e-05, "val_loss": 1.6580920219421387, "within_bucket_gap": 0.0712515115737915}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.3460919737271522e-05, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.3460919737271522e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.329014539718628, "val_fraction": 2.378234398782344e-05, "val_loss": 2.895019292831421, "within_bucket_gap": 0.566004753112793}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.2640423774719238, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.390445709228516, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.248815298080444, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.672542095184326, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.422726631164551, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.347297668457031, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4123237133026123, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.561374307223106e-06, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.561374307223106e-06, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.6982369422912598, "val_fraction": 2.378234398782344e-05, "val_loss": 2.2962002754211426, "within_bucket_gap": -0.4020366668701172}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.163413047790527, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.68217114752649e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.68217114752649e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.236013412475586, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4262940883636475, "within_bucket_gap": -2.8097193241119385}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00015074887678201039, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00015074887678201039, "step": 1800, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 7.164120356241862, "val_fraction": 2.378234398782344e-05, "val_loss": 0.8254315853118896, "within_bucket_gap": -6.338688770929973}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.8786565263158834e-05, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.8786565263158834e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.695039749145508, "val_fraction": 2.378234398782344e-05, "val_loss": 6.4849772453308105, "within_bucket_gap": 0.7899374961853027}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.322898864746094, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.075646349647091e-08, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.075646349647091e-08, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.202578544616699, "val_fraction": 2.378234398782344e-05, "val_loss": 6.206394672393799, "within_bucket_gap": 0.0038161277770996094}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.679725646972656, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.11288794130086899, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.05347486585378647, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.2298383712768555, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.622516632080078, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.8747609853744507, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.952859878540039, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.011736702173948288, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.112921714782715, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.635455846786499, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.259191873048175e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.259191873048175e-05, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.791889190673828, "val_fraction": 2.378234398782344e-05, "val_loss": 4.582794189453125, "within_bucket_gap": 1.7909049987792969}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.700697898864746, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.497308731079102, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.412391582399017e-06, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.412391582399017e-06, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.5605415105819702, "val_fraction": 2.378234398782344e-05, "val_loss": 1.914265751838684, "within_bucket_gap": 0.35372424125671387}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00016272995377149393, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00016272995377149393, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.4243060350418091, "val_fraction": 2.378234398782344e-05, "val_loss": 7.266775131225586, "within_bucket_gap": 6.842469096183777}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.867633819580078, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.442134857177734, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.026738569140434265, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020068530220101774, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020068530220101774, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.7531022429466248, "val_fraction": 2.378234398782344e-05, "val_loss": 10.19151782989502, "within_bucket_gap": 8.438415586948395}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014156759737833448, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014156759737833448, "step": 1800, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.263612508773804, "val_fraction": 4.756468797564688e-05, "val_loss": 2.287295341491699, "within_bucket_gap": -2.9763171672821045}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.550844669342041, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.289978504180908, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.1366400718688965, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.24226021766662598, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.06030256301164627, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.09630965441465378, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.615011692047119, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.0010659971740096807, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.453571449131726e-06, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.453571449131726e-06, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.200704574584961, "val_fraction": 2.378234398782344e-05, "val_loss": 3.09753680229187, "within_bucket_gap": -0.10316777229309082}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.191169738769531, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.526597499847412, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.783975995476025e-07, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.783975995476025e-07, "step": 1800, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.30486947298049927, "val_fraction": 2.378234398782344e-05, "val_loss": 0.2637298107147217, "within_bucket_gap": -0.04113966226577759}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.465900999804337e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.465900999804337e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.6072008609771729, "val_fraction": 4.756468797564688e-05, "val_loss": 0.24780983477830887, "within_bucket_gap": -1.359391026198864}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.54419254930052e-05, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.54419254930052e-05, "step": 1800, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.090518951416016, "val_fraction": 9.512937595129375e-05, "val_loss": 4.297473430633545, "within_bucket_gap": -0.7930455207824707}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.01998544238062852, "frequency": 1, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.01998544238062852, "step": 2000, "support_mode": "paired", "train_fraction": 0.824747907153729, "train_loss": 4.993345441405461, "val_fraction": 0.05995528919330289, "val_loss": 4.660006337390916, "within_bucket_gap": -0.33333910401454503}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.371336026526467e-05, "frequency": 2, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.371336026526467e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.07167998477929985, "train_loss": 4.810261585258377, "val_fraction": 0.019882039573820395, "val_loss": 4.8119572543641596, "within_bucket_gap": 0.001695669105782116}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0010200591065567882, "frequency": 3, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0010200591065567882, "step": 2000, "support_mode": "paired", "train_fraction": 0.02259322678843227, "train_loss": 4.71515625, "val_fraction": 0.011582001522070015, "val_loss": 4.627083467017454, "within_bucket_gap": -0.08807278298254584}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005267766936063632, "frequency": 4, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005267766936063632, "step": 2000, "support_mode": "paired", "train_fraction": 0.01355593607305936, "train_loss": 4.568778483072917, "val_fraction": 0.007301179604261796, "val_loss": 4.496628950783795, "within_bucket_gap": -0.07214953228912169}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0009018701943380949, "frequency": 5, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0009018701943380949, "step": 2000, "support_mode": "paired", "train_fraction": 0.009084855403348554, "train_loss": 4.457990556487238, "val_fraction": 0.004827815829528158, "val_loss": 4.644797639893781, "within_bucket_gap": 0.18680708340654295}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003367239405858509, "frequency": 6, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003367239405858509, "step": 2000, "support_mode": "paired", "train_fraction": 0.00670662100456621, "train_loss": 4.748458321212877, "val_fraction": 0.003305745814307458, "val_loss": 4.850318524477293, "within_bucket_gap": 0.10186020326441625}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0014702910233158127, "frequency": 7, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0014702910233158127, "step": 2000, "support_mode": "paired", "train_fraction": 0.004114345509893455, "train_loss": 4.973965330620032, "val_fraction": 0.002877663622526636, "val_loss": 4.4630331244350465, "within_bucket_gap": -0.5109322061849859}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0009319823114580108, "frequency": 8, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0009319823114580108, "step": 2000, "support_mode": "paired", "train_fraction": 0.002663622526636225, "train_loss": 4.88864939553397, "val_fraction": 0.002449581430745814, "val_loss": 4.50818345153216, "within_bucket_gap": -0.3804659440018101}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.160204815243599e-05, "frequency": 9, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.160204815243599e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.0026398401826484017, "train_loss": 4.476404275980082, "val_fraction": 0.0025684931506849314, "val_loss": 4.51206800672743, "within_bucket_gap": 0.03566373074734841}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006229121031900548, "frequency": 10, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006229121031900548, "step": 2000, "support_mode": "paired", "train_fraction": 0.002330669710806697, "train_loss": 4.35981392373844, "val_fraction": 0.0014745053272450532, "val_loss": 4.782268893334173, "within_bucket_gap": 0.4224549695957327}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0012070190140727821, "frequency": 11, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0012070190140727821, "step": 2000, "support_mode": "paired", "train_fraction": 0.001736111111111111, "train_loss": 4.677358339910638, "val_fraction": 0.0020214992389649923, "val_loss": 4.080267333984375, "within_bucket_gap": -0.5970910059262629}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00046926324548564894, "frequency": 12, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00046926324548564894, "step": 2000, "support_mode": "paired", "train_fraction": 0.0010702054794520547, "train_loss": 4.076618279351129, "val_fraction": 0.0016647640791476407, "val_loss": 3.794738551548549, "within_bucket_gap": -0.2818797278025795}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007994849185122197, "frequency": 13, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007994849185122197, "step": 2000, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 4.412471210255342, "val_fraction": 0.0018074581430745813, "val_loss": 4.854796760960629, "within_bucket_gap": 0.442325550705287}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0008055371300622786, "frequency": 14, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0008055371300622786, "step": 2000, "support_mode": "paired", "train_fraction": 0.0016171993911719938, "train_loss": 3.8294701295740463, "val_fraction": 0.0015220700152207, "val_loss": 4.358708024024963, "within_bucket_gap": 0.5292378944509171}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.303829328995518e-05, "frequency": 15, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.303829328995518e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.005731544901065449, "train_loss": 4.380030715119295, "val_fraction": 0.0014745053272450532, "val_loss": 4.429564814413747, "within_bucket_gap": 0.04953409929445218}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005324709723892097, "frequency": 16, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005324709723892097, "step": 2000, "support_mode": "paired", "train_fraction": 0.0020452815829528157, "train_loss": 4.505054828732512, "val_fraction": 0.0010226407914764079, "val_loss": 5.025737141453942, "within_bucket_gap": 0.52068231272143}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0002169128516990323, "frequency": 17, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0002169128516990323, "step": 2000, "support_mode": "paired", "train_fraction": 0.0010939878234398782, "train_loss": 4.872649151345958, "val_fraction": 0.0004280821917808219, "val_loss": 4.3659407297770185, "within_bucket_gap": -0.5067084215689395}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000487149427784761, "frequency": 18, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000487149427784761, "step": 2000, "support_mode": "paired", "train_fraction": 0.000975076103500761, "train_loss": 5.017522300161967, "val_fraction": 0.0007372526636225266, "val_loss": 4.356759102113785, "within_bucket_gap": -0.6607631980481816}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.765293610824315e-05, "frequency": 19, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.765293610824315e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.739611353193011, "val_fraction": 0.000594558599695586, "val_loss": 4.836578979492187, "within_bucket_gap": 0.09696762629917632}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000723176999891018, "frequency": 20, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000723176999891018, "step": 2000, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 4.756899907038762, "val_fraction": 0.0007372526636225266, "val_loss": 5.737807858374811, "within_bucket_gap": 0.9809079513360492}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005344204191619098, "frequency": 21, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005344204191619098, "step": 2000, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 4.527777887159778, "val_fraction": 0.0006183409436834094, "val_loss": 5.3920590327336235, "within_bucket_gap": 0.8642811455738455}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00048059252299130256, "frequency": 22, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00048059252299130256, "step": 2000, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 4.844208717346191, "val_fraction": 0.0006659056316590563, "val_loss": 5.565921374729702, "within_bucket_gap": 0.7217126573835104}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0006960692930303208, "frequency": 23, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0006960692930303208, "step": 2000, "support_mode": "paired", "train_fraction": 0.000380517503805175, "train_loss": 3.726817339658737, "val_fraction": 0.00047564687975646877, "val_loss": 5.190233421325684, "within_bucket_gap": 1.4634160816669466}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002888768847855986, "frequency": 24, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002888768847855986, "step": 2000, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 3.08965087890625, "val_fraction": 0.0003091704718417047, "val_loss": 4.024012052095854, "within_bucket_gap": 0.934361173189604}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007287455294593102, "frequency": 25, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007287455294593102, "step": 2000, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.3649139404296875, "val_fraction": 0.000594558599695586, "val_loss": 4.5906056213378905, "within_bucket_gap": 1.225691680908203}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.000393734765742286, "frequency": 26, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.000393734765742286, "step": 2000, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.665259679158528, "val_fraction": 0.0004280821917808219, "val_loss": 4.745495266384548, "within_bucket_gap": -0.9197644127739801}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007625310687180309, "frequency": 27, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007625310687180309, "step": 2000, "support_mode": "paired", "train_fraction": 0.0012128995433789954, "train_loss": 3.2675325730267692, "val_fraction": 0.000594558599695586, "val_loss": 4.550048828125, "within_bucket_gap": 1.2825162550982308}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004901564375269179, "frequency": 28, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004901564375269179, "step": 2000, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.084716796875, "val_fraction": 0.00045186453576864533, "val_loss": 5.1694587908293075, "within_bucket_gap": 1.0847419939543075}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0010981295034110108, "frequency": 29, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0010981295034110108, "step": 2000, "support_mode": "paired", "train_fraction": 0.0007372526636225266, "train_loss": 1.8535506956038936, "val_fraction": 0.00040429984779299846, "val_loss": 4.569677128511317, "within_bucket_gap": 2.716126432907423}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00030534428306668623, "frequency": 30, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00030534428306668623, "step": 2000, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.29514726003011, "val_fraction": 0.00033295281582952814, "val_loss": 5.212227003914969, "within_bucket_gap": 0.9170797438848588}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005809964393825472, "frequency": 31, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005809964393825472, "step": 2000, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 6.499295949935913, "val_fraction": 0.0002616057838660578, "val_loss": 4.2784106514670635, "within_bucket_gap": -2.2208852984688496}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.001413487219254055, "frequency": 32, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.001413487219254055, "step": 2000, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 5.728458563486735, "val_fraction": 0.0005232115677321156, "val_loss": 3.026898990977894, "within_bucket_gap": -2.7015595725088413}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001861801128106815, "frequency": 33, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001861801128106815, "step": 2000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.884066899617513, "val_fraction": 0.0002616057838660578, "val_loss": 4.172384955666282, "within_bucket_gap": -0.7116819439512305}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003444245653710508, "frequency": 34, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003444245653710508, "step": 2000, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 3.8647725847032337, "val_fraction": 0.0002616057838660578, "val_loss": 5.181351141496138, "within_bucket_gap": 1.316578556792904}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003290244660239604, "frequency": 35, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003290244660239604, "step": 2000, "support_mode": "paired", "train_fraction": 0.00028538812785388126, "train_loss": 3.808862050374349, "val_fraction": 0.00014269406392694063, "val_loss": 6.114665508270264, "within_bucket_gap": 2.3058034578959146}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.201796033732005e-05, "frequency": 36, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.201796033732005e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.541251579920451, "val_fraction": 0.0002616057838660578, "val_loss": 6.2659596529873935, "within_bucket_gap": -0.2752919269330576}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001424398282389076, "frequency": 37, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001424398282389076, "step": 2000, "support_mode": "paired", "train_fraction": 0.0003091704718417047, "train_loss": 3.801619162926307, "val_fraction": 0.00028538812785388126, "val_loss": 4.3007283210754395, "within_bucket_gap": 0.49910915814913226}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0007306132097041044, "frequency": 38, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0007306132097041044, "step": 2000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 6.88895320892334, "val_fraction": 0.0003567351598173516, "val_loss": 4.840898259480794, "within_bucket_gap": -2.0480549494425455}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0004448449963969305, "frequency": 39, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0004448449963969305, "step": 2000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.245007356007894, "val_fraction": 0.0002616057838660578, "val_loss": 5.9454475749622695, "within_bucket_gap": 1.700440218954376}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001523264749376411, "frequency": 40, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001523264749376411, "step": 2000, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 2.950785983692516, "val_fraction": 9.512937595129375e-05, "val_loss": 4.5520418882369995, "within_bucket_gap": 1.6012559045444834}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003289128130066522, "frequency": 41, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003289128130066522, "step": 2000, "support_mode": "paired", "train_fraction": 0.000594558599695586, "train_loss": 4.421300659179687, "val_fraction": 0.00021404109589041095, "val_loss": 5.957981321546766, "within_bucket_gap": 1.536680662367079}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00018002258848575137, "frequency": 42, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00018002258848575137, "step": 2000, "support_mode": "paired", "train_fraction": 0.005493721461187214, "train_loss": 2.4806344994218836, "val_fraction": 0.00014269406392694063, "val_loss": 3.7422327995300293, "within_bucket_gap": 1.2615983001081457}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00034272219725706496, "frequency": 43, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00034272219725706496, "step": 2000, "support_mode": "paired", "train_fraction": 0.00033295281582952814, "train_loss": 3.0215580804007396, "val_fraction": 0.00023782343987823439, "val_loss": 4.4626363754272464, "within_bucket_gap": 1.4410782950265069}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0007158498615706891, "frequency": 44, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0007158498615706891, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 0.8822290897369385, "val_fraction": 0.0002616057838660578, "val_loss": 3.6185977242209693, "within_bucket_gap": 2.736368634484031}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017836847957289806, "frequency": 45, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017836847957289806, "step": 2000, "support_mode": "paired", "train_fraction": 0.00021404109589041095, "train_loss": 4.213950475056966, "val_fraction": 0.00016647640791476407, "val_loss": 5.285384450639997, "within_bucket_gap": 1.071433975583031}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00031223126560948574, "frequency": 46, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00031223126560948574, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.375862121582031, "val_fraction": 0.000380517503805175, "val_loss": 3.5553183555603027, "within_bucket_gap": -0.8205437660217285}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00014517220576759732, "frequency": 47, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00014517220576759732, "step": 2000, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 4.823420660836356, "val_fraction": 7.134703196347032e-05, "val_loss": 6.858154296875, "within_bucket_gap": 2.034733636038644}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010757314401311598, "frequency": 48, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010757314401311598, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.377111554145813, "val_fraction": 4.756468797564688e-05, "val_loss": 2.1154937744140625, "within_bucket_gap": -2.2616177797317505}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011846941023051951, "frequency": 49, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011846941023051951, "step": 2000, "support_mode": "paired", "train_fraction": 0.0007848173515981735, "train_loss": 1.3126858075459797, "val_fraction": 4.756468797564688e-05, "val_loss": 3.803386688232422, "within_bucket_gap": 2.4907008806864424}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.896465319592173e-05, "frequency": 50, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.896465319592173e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.0002616057838660578, "train_loss": 4.713129737160423, "val_fraction": 0.00016647640791476407, "val_loss": 5.1874591282435825, "within_bucket_gap": 0.47432939108315963}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00048561795661438545, "frequency": 51, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00048561795661438545, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.268976211547852, "val_fraction": 7.134703196347032e-05, "val_loss": 1.462554931640625, "within_bucket_gap": -6.806421279907227}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -5.612043448393566e-05, "frequency": 52, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -5.612043448393566e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.4234370440244675, "val_fraction": 4.756468797564688e-05, "val_loss": 1.243561029434204, "within_bucket_gap": -1.1798760145902634}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011813647844177609, "frequency": 53, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011813647844177609, "step": 2000, "support_mode": "paired", "train_fraction": 0.00016647640791476407, "train_loss": 5.814495435782841, "val_fraction": 0.00014269406392694063, "val_loss": 4.986594994862874, "within_bucket_gap": -0.8279004409199668}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.6600376950460663e-05, "frequency": 54, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.6600376950460663e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.0001902587519025875, "train_loss": 2.127001702785492, "val_fraction": 2.378234398782344e-05, "val_loss": 2.825014352798462, "within_bucket_gap": 0.69801265001297}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.631699705777093e-05, "frequency": 55, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.631699705777093e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.203835296630859, "val_fraction": 0.00016647640791476407, "val_loss": 4.685341426304409, "within_bucket_gap": -0.5184938703264503}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.45469873680916e-06, "frequency": 56, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.45469873680916e-06, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 5.009920756022136, "val_fraction": 0.00014269406392694063, "val_loss": 5.041139284769694, "within_bucket_gap": 0.031218528747558594}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00041765722061990786, "frequency": 57, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00041765722061990786, "step": 2000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 5.105709743499756, "val_fraction": 0.00011891171993911719, "val_loss": 1.5933795809745788, "within_bucket_gap": -3.5123301625251773}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.872714298319417e-05, "frequency": 58, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.872714298319417e-05, "step": 2000, "support_mode": "paired", "train_fraction": 0.00076103500761035, "train_loss": 3.8056257367134094, "val_fraction": 9.512937595129375e-05, "val_loss": 4.317845463752747, "within_bucket_gap": 0.5122197270393372}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005192519413644683, "frequency": 59, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005192519413644683, "step": 2000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 3.674530029296875, "val_fraction": 0.00023782343987823439, "val_loss": 5.857880592346191, "within_bucket_gap": 2.1833505630493164}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006524306249236362, "frequency": 60, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006524306249236362, "step": 2000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.017030143737793, "val_fraction": 0.00021404109589041095, "val_loss": 2.9688742640945645, "within_bucket_gap": -3.048155879643229}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001049923570188757, "frequency": 61, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001049923570188757, "step": 2000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 6.582008361816406, "val_fraction": 0.00011891171993911719, "val_loss": 5.699064636230469, "within_bucket_gap": -0.8829437255859371}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.177904089473336e-05, "frequency": 62, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.177904089473336e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.673768520355225, "val_fraction": 7.134703196347032e-05, "val_loss": 5.679823557535808, "within_bucket_gap": 1.006055037180583}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00026739645420461667, "frequency": 63, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00026739645420461667, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.157257497310638, "val_fraction": 0.00016647640791476407, "val_loss": 4.551045196396964, "within_bucket_gap": -1.6062123009136746}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.696419076469572e-05, "frequency": 64, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.696419076469572e-05, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.298414468765259, "val_fraction": 0.0001902587519025875, "val_loss": 6.5978182554244995, "within_bucket_gap": 0.2994037866592407}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.619525760729751e-05, "frequency": 65, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.619525760729751e-05, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.031204342842102, "val_fraction": 0.0002616057838660578, "val_loss": 4.131336905739524, "within_bucket_gap": 0.10013256289742234}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.817950921878785e-05, "frequency": 66, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.817950921878785e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.666358709335327, "val_fraction": 4.756468797564688e-05, "val_loss": 4.022712707519531, "within_bucket_gap": -1.643646001815796}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00023240258114885884, "frequency": 67, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00023240258114885884, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.368028283119202, "val_fraction": 7.134703196347032e-05, "val_loss": 3.110673705736796, "within_bucket_gap": -3.2573545773824057}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -9.056413191820026e-06, "frequency": 68, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -9.056413191820026e-06, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.121572494506836, "val_fraction": 4.756468797564688e-05, "val_loss": 6.931170463562012, "within_bucket_gap": -0.19040203094482422}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.000335435027219752, "frequency": 69, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.000335435027219752, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.375744581222534, "val_fraction": 9.512937595129375e-05, "val_loss": 6.901837587356567, "within_bucket_gap": 3.526093006134033}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012987917870328487, "frequency": 70, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012987917870328487, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.437105178833008, "val_fraction": 9.512937595129375e-05, "val_loss": 3.8023951053619385, "within_bucket_gap": 1.3652899265289307}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00010414010585715234, "frequency": 71, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00010414010585715234, "step": 2000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 4.4087272644042965, "val_fraction": 7.134703196347032e-05, "val_loss": 2.949099540710449, "within_bucket_gap": -1.4596277236938473}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0005932717698894134, "frequency": 72, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0005932717698894134, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.8663266897201538, "val_fraction": 0.0001902587519025875, "val_loss": 3.984563112258911, "within_bucket_gap": 3.1182364225387573}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.104253342525418e-06, "frequency": 73, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.104253342525418e-06, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.4570651054382324, "val_fraction": 2.378234398782344e-05, "val_loss": 2.755784749984741, "within_bucket_gap": 0.2987196445465088}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00011994084130683445, "frequency": 74, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00011994084130683445, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 2.3426594734191895, "val_fraction": 7.134703196347032e-05, "val_loss": 4.023750305175781, "within_bucket_gap": 1.6810908317565918}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012871749483649953, "frequency": 76, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012871749483649953, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.2512495517730713, "val_fraction": 0.00011891171993911719, "val_loss": 3.3337121963500977, "within_bucket_gap": 1.0824626445770265}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002931983662341102, "frequency": 77, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002931983662341102, "step": 2000, "support_mode": "paired", "train_fraction": 0.00014269406392694063, "train_loss": 4.519212245941162, "val_fraction": 7.134703196347032e-05, "val_loss": 8.628680547078451, "within_bucket_gap": 4.109468301137289}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00020668376662415455, "frequency": 78, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00020668376662415455, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.016775727272034, "val_fraction": 9.512937595129375e-05, "val_loss": 6.1894354820251465, "within_bucket_gap": 2.172659754753113}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 79, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 6.079134941101074, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.346422635619861e-05, "frequency": 80, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.346422635619861e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.733731508255005, "val_fraction": 4.756468797564688e-05, "val_loss": 4.819939613342285, "within_bucket_gap": -0.9137918949127197}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0001563356347280006, "frequency": 81, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0001563356347280006, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.359658718109131, "val_fraction": 4.756468797564688e-05, "val_loss": 2.0728583335876465, "within_bucket_gap": -3.2868003845214844}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 82, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.430271625518799, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010690675932159888, "frequency": 83, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010690675932159888, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 2.6555882692337036, "val_fraction": 0.0001902587519025875, "val_loss": 3.2174901962280273, "within_bucket_gap": 0.5619019269943237}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0005971403789665419, "frequency": 84, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0005971403789665419, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 13.632430076599121, "val_fraction": 7.134703196347032e-05, "val_loss": 5.262910525004069, "within_bucket_gap": -8.369519551595051}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00018051173998669492, "frequency": 85, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00018051173998669492, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.9577720165252686, "val_fraction": 4.756468797564688e-05, "val_loss": 2.1626931950449944, "within_bucket_gap": -3.795078821480274}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.725319563586481e-05, "frequency": 86, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.725319563586481e-05, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.0318857828776045, "val_fraction": 9.512937595129375e-05, "val_loss": 3.7454001903533936, "within_bucket_gap": -0.2864855925242109}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.776527529619237e-05, "frequency": 87, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.776527529619237e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.34413480758667, "val_fraction": 9.512937595129375e-05, "val_loss": 3.8420262336730957, "within_bucket_gap": -0.5021085739135742}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.2220929453724954e-05, "frequency": 88, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.2220929453724954e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.07364022731781, "val_fraction": 2.378234398782344e-05, "val_loss": 3.269425868988037, "within_bucket_gap": 2.195785641670227}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 5.53670164929133e-05, "frequency": 89, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 5.53670164929133e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.0487654209136963, "val_fraction": 2.378234398782344e-05, "val_loss": 3.376837730407715, "within_bucket_gap": 2.3280723094940186}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002303792930145302, "frequency": 90, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002303792930145302, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.2298838297526045, "val_fraction": 4.756468797564688e-05, "val_loss": 9.073378086090088, "within_bucket_gap": 4.843494256337483}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00012504024774153663, "frequency": 91, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00012504024774153663, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.017629623413086, "val_fraction": 2.378234398782344e-05, "val_loss": 9.275321960449219, "within_bucket_gap": 5.257692337036133}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.1752084069056054e-05, "frequency": 93, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.1752084069056054e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.225140571594238, "val_fraction": 4.756468797564688e-05, "val_loss": 6.102936387062073, "within_bucket_gap": 0.8777958154678345}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00013135525410714214, "frequency": 94, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00013135525410714214, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.9839716553688049, "val_fraction": 2.378234398782344e-05, "val_loss": 6.507197380065918, "within_bucket_gap": 5.523225724697113}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00017058398670983277, "frequency": 95, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00017058398670983277, "step": 2000, "support_mode": "paired", "train_fraction": 0.00011891171993911719, "train_loss": 3.9374463081359865, "val_fraction": 2.378234398782344e-05, "val_loss": 11.110161781311035, "within_bucket_gap": 7.172715473175049}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.121568601878991e-05, "frequency": 97, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.121568601878991e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.207558572292328, "val_fraction": 4.756468797564688e-05, "val_loss": 4.915037155151367, "within_bucket_gap": 1.7074785828590393}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.0006336254638986864, "frequency": 100, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.0006336254638986864, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 12.636398315429688, "val_fraction": 7.134703196347032e-05, "val_loss": 3.7555038134256997, "within_bucket_gap": -8.880894502003988}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 101, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.0100295146306357, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 102, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.76811408996582, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.2146578350386481e-05, "frequency": 103, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.2146578350386481e-05, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 4.512790203094482, "val_fraction": 2.378234398782344e-05, "val_loss": 5.023529529571533, "within_bucket_gap": 0.5107393264770508}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 107, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.8184987902641296, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 108, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.276239395141602, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.081764513226949e-05, "frequency": 109, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.081764513226949e-05, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.093340277671814, "val_fraction": 2.378234398782344e-05, "val_loss": 3.2179999351501465, "within_bucket_gap": -0.8753403425216675}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 110, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.1753640174865723, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 111, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 4.282305836677551, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 112, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.311384201049805, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002792414291612974, "frequency": 114, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002792414291612974, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.9101925678551197, "val_fraction": 4.756468797564688e-05, "val_loss": 6.780964374542236, "within_bucket_gap": 5.870771806687117}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -4.599252474725337e-05, "frequency": 116, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -4.599252474725337e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.515669345855713, "val_fraction": 2.378234398782344e-05, "val_loss": 5.581775665283203, "within_bucket_gap": -1.9338936805725098}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.855263928481614e-05, "frequency": 117, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.855263928481614e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.6901092529296875, "val_fraction": 7.134703196347032e-05, "val_loss": 1.7911030451456706, "within_bucket_gap": 1.1009937922159831}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 120, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.099799156188965, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00019765028104900496, "frequency": 123, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00019765028104900496, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.0023480281233788, "val_fraction": 0.00014269406392694063, "val_loss": 1.617214858531952, "within_bucket_gap": -1.3851331695914268}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.00010718198853177747, "frequency": 124, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.00010718198853177747, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 3.6349096298217773, "val_fraction": 4.756468797564688e-05, "val_loss": 5.888303756713867, "within_bucket_gap": 2.25339412689209}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.171906799667683e-05, "frequency": 126, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.171906799667683e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.7138686180114746, "val_fraction": 2.378234398782344e-05, "val_loss": 6.149991989135742, "within_bucket_gap": 3.4361233711242676}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00013971168837046514, "frequency": 128, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00013971168837046514, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.123255729675293, "val_fraction": 2.378234398782344e-05, "val_loss": 2.2486586570739746, "within_bucket_gap": -5.874597072601318}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.511415067147265e-05, "frequency": 129, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.511415067147265e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.460736274719238, "val_fraction": 2.378234398782344e-05, "val_loss": 9.19865608215332, "within_bucket_gap": 2.737919807434082}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 133, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 9.19885802268982, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 137, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.899607241153717, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.056040570797623e-06, "frequency": 140, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.056040570797623e-06, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.762866497039795, "val_fraction": 4.756468797564688e-05, "val_loss": 3.890188694000244, "within_bucket_gap": 0.12732219696044922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 141, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.445751667022705, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 145, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.561453342437744, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 148, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.1248931884765625, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 149, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 2.7774130503336587, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 9.414206345422445e-05, "frequency": 156, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 9.414206345422445e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.3866654634475708, "val_fraction": 2.378234398782344e-05, "val_loss": 5.345150947570801, "within_bucket_gap": 3.95848548412323}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 158, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.603579998016357, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 163, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.4566179513931274, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 165, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.049157895147800446, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 166, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.810966491699219, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -6.968506485722743e-05, "frequency": 168, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -6.968506485722743e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.189199924468994, "val_fraction": 4.756468797564688e-05, "val_loss": 4.7241411209106445, "within_bucket_gap": -1.4650588035583496}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 173, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.900301456451416, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 174, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.2996039390563965, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 176, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.11176006495952606, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.396724168931331e-05, "frequency": 181, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.396724168931331e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.991925001144409, "val_fraction": 2.378234398782344e-05, "val_loss": 3.404630422592163, "within_bucket_gap": -0.5872945785522461}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.325754929169491e-05, "frequency": 183, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.325754929169491e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.720720291137695, "val_fraction": 2.378234398782344e-05, "val_loss": 8.539613723754883, "within_bucket_gap": 1.8188934326171875}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.517320025457095e-05, "frequency": 187, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.517320025457095e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.699150085449219, "val_fraction": 2.378234398782344e-05, "val_loss": 3.1177873611450195, "within_bucket_gap": -3.581362724304199}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 191, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.355153560638428, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 198, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 8.151145935058594, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 210, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.095751762390137, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0003373303551289226, "frequency": 211, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0003373303551289226, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.0535688400268555, "val_fraction": 4.756468797564688e-05, "val_loss": 11.145602226257324, "within_bucket_gap": 7.092033386230469}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 212, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.011635535396635532, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 215, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.180259704589844, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 217, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.1518070697784424, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 219, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.9181444644927979, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 226, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.790492057800293, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.486686558211775e-05, "frequency": 239, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.486686558211775e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 3.5219608545303345, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4763588905334473, "within_bucket_gap": -1.0456019639968872}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 241, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.07706069946289, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 247, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.14621242880821228, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 1.3641667193655191e-05, "frequency": 248, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 1.3641667193655191e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.1563355922698975, "val_fraction": 2.378234398782344e-05, "val_loss": 1.729940414428711, "within_bucket_gap": 0.5736048221588135}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 3.973988864338379e-06, "frequency": 251, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 3.973988864338379e-06, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.2509706020355225, "val_fraction": 2.378234398782344e-05, "val_loss": 2.4180688858032227, "within_bucket_gap": 0.1670982837677002}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 257, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 1.1819087266921997, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 258, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.19588851928711, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 261, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 4.095026731491089, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.806628704071045, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 278, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.804737091064453, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 284, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.215412139892578, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 288, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.9505682587623596, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -1.478680555497493e-05, "frequency": 291, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -1.478680555497493e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.650083303451538, "val_fraction": 2.378234398782344e-05, "val_loss": 2.028327703475952, "within_bucket_gap": -0.6217555999755859}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 292, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.547149658203125, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -8.431406722856257e-05, "frequency": 296, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -8.431406722856257e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.53886604309082, "val_fraction": 2.378234398782344e-05, "val_loss": 1.9936281442642212, "within_bucket_gap": -3.545237898826599}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014042494490644402, "frequency": 298, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014042494490644402, "step": 2000, "support_mode": "paired", "train_fraction": 7.134703196347032e-05, "train_loss": 6.666504542032878, "val_fraction": 2.378234398782344e-05, "val_loss": 0.76191645860672, "within_bucket_gap": -5.904588083426158}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 4.632968336479848e-06, "frequency": 302, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 4.632968336479848e-06, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.650097846984863, "val_fraction": 2.378234398782344e-05, "val_loss": 5.844904899597168, "within_bucket_gap": 0.1948070526123047}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 315, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.921658515930176, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 7.603839153205604e-06, "frequency": 320, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 7.603839153205604e-06, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 6.085413217544556, "val_fraction": 2.378234398782344e-05, "val_loss": 6.405139446258545, "within_bucket_gap": 0.31972622871398926}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 332, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 11.788642883300781, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 368, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.06534682214260101, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 376, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.05325980484485626, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 378, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 5.03030252456665, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 387, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 9.086109161376953, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 394, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.7240195274353027, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 407, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 10.921772956848145, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 411, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 0.009906318038702011, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 422, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.714578628540039, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 424, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.848679304122925, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 6.195808699867678e-05, "frequency": 431, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 6.195808699867678e-05, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 2.4009742736816406, "val_fraction": 2.378234398782344e-05, "val_loss": 5.006187915802002, "within_bucket_gap": 2.6052136421203613}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 433, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.231311798095703, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 467, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 4.665462493896484, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 8.049863284309161e-06, "frequency": 470, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 8.049863284309161e-06, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.3557537198066711, "val_fraction": 2.378234398782344e-05, "val_loss": 1.6942343711853027, "within_bucket_gap": 0.3384806513786316}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0001648378656267031, "frequency": 502, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0001648378656267031, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.38123053312301636, "val_fraction": 2.378234398782344e-05, "val_loss": 7.312333106994629, "within_bucket_gap": 6.9311025738716125}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 535, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.6258225440979, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 596, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 7.475223064422607, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 597, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.02473817765712738, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 0.0002130185759979296, "frequency": 674, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 0.0002130185759979296, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.7069620490074158, "val_fraction": 2.378234398782344e-05, "val_loss": 10.66396713256836, "within_bucket_gap": 8.957005083560944}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00014369137006807543, "frequency": 879, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00014369137006807543, "step": 2000, "support_mode": "paired", "train_fraction": 9.512937595129375e-05, "train_loss": 5.27180016040802, "val_fraction": 4.756468797564688e-05, "val_loss": 2.2508327960968018, "within_bucket_gap": -3.0209673643112183}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 944, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.3379838466644287, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1164, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.395466327667236, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1177, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 7.2734599113464355, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1179, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.11194416135549545, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1200, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.04047468304634094, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1265, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.0584486648440361, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1273, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 6.833982467651367, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 1551, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.0006889115320518613, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -2.8959284090015985e-06, "frequency": 2193, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -2.8959284090015985e-06, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.5979158878326416, "val_fraction": 2.378234398782344e-05, "val_loss": 3.4761478900909424, "within_bucket_gap": -0.12176799774169922}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2335, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 3.7545647621154785, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": null, "frequency": 2789, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": null, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 8.228927612304688, "val_fraction": null, "val_loss": null, "within_bucket_gap": null}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": 2.658592829058341e-09, "frequency": 8680, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": 2.658592829058341e-09, "step": 2000, "support_mode": "paired", "train_fraction": 2.378234398782344e-05, "train_loss": 0.2653353810310364, "val_fraction": 2.378234398782344e-05, "val_loss": 0.2654471695423126, "within_bucket_gap": 0.00011178851127624512}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -7.207302271730098e-05, "frequency": 13630, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -7.207302271730098e-05, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 1.7415355443954468, "val_fraction": 4.756468797564688e-05, "val_loss": 0.226272314786911, "within_bucket_gap": -1.5152632296085358}
{"branch": "exact_context", "bucket": null, "composition_component": null, "conditional_loss_component": null, "epoch": 1, "exact_global_gap_contribution": -0.00011558529652961312, "frequency": 19285, "frequency_definition": "exact_train_epoch_context_count", "frequency_key_type": "exact_context", "frequency_source_split": "train", "measurement": "allgram_fixed_frequency_decomposition", "probe_set": "train", "reconstruction_error": null, "schema_version": 2, "shaoyang_val_fraction_gap_contribution": -0.00011558529652961312, "step": 2000, "support_mode": "paired", "train_fraction": 4.756468797564688e-05, "train_loss": 5.329117774963379, "val_fraction": 9.512937595129375e-05, "val_loss": 4.114085137844086, "within_bucket_gap": -1.2150326371192932}