longformer-spans / checkpoint-1458 /trainer_state.json
Theoreticallyhugo's picture
Training in progress, epoch 18, checkpoint
a16e833 verified
raw
history blame
21 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 18.0,
"eval_steps": 500,
"global_step": 1458,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.0,
"eval_B": {
"f1-score": 0.8087666523420713,
"precision": 0.8189730200174065,
"recall": 0.7988115449915111,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9374597034171502,
"precision": 0.9521418826739427,
"recall": 0.9232234509762421,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.8881362400419788,
"precision": 0.863303347862376,
"recall": 0.9144400785854617,
"support": 10180.0
},
"eval_accuracy": 0.9154245298608586,
"eval_loss": 0.2228638380765915,
"eval_macro avg": {
"f1-score": 0.8781208652670669,
"precision": 0.8781394168512416,
"recall": 0.8788250248510717,
"support": 30257.0
},
"eval_runtime": 4.7167,
"eval_samples_per_second": 17.173,
"eval_steps_per_second": 2.332,
"eval_weighted avg": {
"f1-score": 0.9158543469268937,
"precision": 0.9170673807540184,
"recall": 0.9154245298608586,
"support": 30257.0
},
"step": 81
},
{
"epoch": 2.0,
"eval_B": {
"f1-score": 0.8401515151515152,
"precision": 0.7585499316005472,
"recall": 0.9414261460101867,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9568235603470944,
"precision": 0.9510219016256338,
"recall": 0.9626964389650247,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9185648054827655,
"precision": 0.9430877483443708,
"recall": 0.8952848722986247,
"support": 10180.0
},
"eval_accuracy": 0.9391876260038999,
"eval_loss": 0.16692574322223663,
"eval_macro avg": {
"f1-score": 0.905179960327125,
"precision": 0.8842198605235172,
"recall": 0.9331358190912787,
"support": 30257.0
},
"eval_runtime": 4.7116,
"eval_samples_per_second": 17.192,
"eval_steps_per_second": 2.335,
"eval_weighted avg": {
"f1-score": 0.9394089523635119,
"precision": 0.9408589092241132,
"recall": 0.9391876260038999,
"support": 30257.0
},
"step": 162
},
{
"epoch": 3.0,
"eval_B": {
"f1-score": 0.8612244897959184,
"precision": 0.8294025157232704,
"recall": 0.8955857385398981,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.951106371634231,
"precision": 0.9584654236741712,
"recall": 0.9438594634636753,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9106743212999903,
"precision": 0.9021592442645074,
"recall": 0.9193516699410609,
"support": 10180.0
},
"eval_accuracy": 0.9337343424662061,
"eval_loss": 0.1742253601551056,
"eval_macro avg": {
"f1-score": 0.9076683942433799,
"precision": 0.8966757278873163,
"recall": 0.9195989573148781,
"support": 30257.0
},
"eval_runtime": 4.7094,
"eval_samples_per_second": 17.2,
"eval_steps_per_second": 2.336,
"eval_weighted avg": {
"f1-score": 0.9340035812317422,
"precision": 0.9344963252190521,
"recall": 0.9337343424662061,
"support": 30257.0
},
"step": 243
},
{
"epoch": 4.0,
"eval_B": {
"f1-score": 0.8717948717948719,
"precision": 0.8254931714719271,
"recall": 0.9235993208828522,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9561271586371415,
"precision": 0.9374586871408959,
"recall": 0.9755542621302714,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9117828500925355,
"precision": 0.956427955133736,
"recall": 0.8711198428290766,
"support": 10180.0
},
"eval_accuracy": 0.9383944211256899,
"eval_loss": 0.18792153894901276,
"eval_macro avg": {
"f1-score": 0.9132349601748496,
"precision": 0.9064599379155197,
"recall": 0.9234244752807333,
"support": 30257.0
},
"eval_runtime": 4.7342,
"eval_samples_per_second": 17.11,
"eval_steps_per_second": 2.324,
"eval_weighted avg": {
"f1-score": 0.9379241479327001,
"precision": 0.9394817486046585,
"recall": 0.9383944211256899,
"support": 30257.0
},
"step": 324
},
{
"epoch": 5.0,
"eval_B": {
"f1-score": 0.8742949234488315,
"precision": 0.8320552147239264,
"recall": 0.9210526315789473,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9583957528450158,
"precision": 0.9521148825065274,
"recall": 0.9647600402137679,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9198818996146724,
"precision": 0.9375701315923697,
"recall": 0.9028487229862475,
"support": 10180.0
},
"eval_accuracy": 0.9422282447037049,
"eval_loss": 0.19990424811840057,
"eval_macro avg": {
"f1-score": 0.9175241919695066,
"precision": 0.9072467429409411,
"recall": 0.9295537982596542,
"support": 30257.0
},
"eval_runtime": 4.7101,
"eval_samples_per_second": 17.197,
"eval_steps_per_second": 2.335,
"eval_weighted avg": {
"f1-score": 0.9421634164298524,
"precision": 0.9425469857238316,
"recall": 0.9422282447037049,
"support": 30257.0
},
"step": 405
},
{
"epoch": 6.0,
"eval_B": {
"f1-score": 0.8808247422680412,
"precision": 0.8564554931836408,
"recall": 0.9066213921901528,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.957406827435014,
"precision": 0.9446361435855178,
"recall": 0.9705275411397428,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9163000050574014,
"precision": 0.9443344105076619,
"recall": 0.8898821218074656,
"support": 10180.0
},
"eval_accuracy": 0.9409062365733549,
"eval_loss": 0.20615626871585846,
"eval_macro avg": {
"f1-score": 0.9181771915868189,
"precision": 0.9151420157589403,
"recall": 0.922343685045787,
"support": 30257.0
},
"eval_runtime": 4.6949,
"eval_samples_per_second": 17.253,
"eval_steps_per_second": 2.343,
"eval_weighted avg": {
"f1-score": 0.9405948120953971,
"precision": 0.941101475610967,
"recall": 0.9409062365733549,
"support": 30257.0
},
"step": 486
},
{
"epoch": 6.17,
"grad_norm": 2.4104158878326416,
"learning_rate": 1.7530864197530865e-05,
"loss": 0.1662,
"step": 500
},
{
"epoch": 7.0,
"eval_B": {
"f1-score": 0.8836820083682009,
"precision": 0.8712871287128713,
"recall": 0.8964346349745331,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9614174269469787,
"precision": 0.9609854091774159,
"recall": 0.9618498333245146,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9271751440248166,
"precision": 0.9295093296475466,
"recall": 0.924852652259332,
"support": 10180.0
},
"eval_accuracy": 0.9468552731599299,
"eval_loss": 0.22925782203674316,
"eval_macro avg": {
"f1-score": 0.9240915264466655,
"precision": 0.9205939558459445,
"recall": 0.9277123735194599,
"support": 30257.0
},
"eval_runtime": 4.7288,
"eval_samples_per_second": 17.129,
"eval_steps_per_second": 2.326,
"eval_weighted avg": {
"f1-score": 0.9468700903559945,
"precision": 0.9469030129054358,
"recall": 0.9468552731599299,
"support": 30257.0
},
"step": 567
},
{
"epoch": 8.0,
"eval_B": {
"f1-score": 0.8803630363036303,
"precision": 0.8563402889245586,
"recall": 0.9057724957555179,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9580651967507121,
"precision": 0.9550928116947994,
"recall": 0.9610561405365363,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9208882720333101,
"precision": 0.9294576746047628,
"recall": 0.912475442043222,
"support": 10180.0
},
"eval_accuracy": 0.9425587467362925,
"eval_loss": 0.22973769903182983,
"eval_macro avg": {
"f1-score": 0.9197721683625509,
"precision": 0.9136302584080402,
"recall": 0.9264346927784253,
"support": 30257.0
},
"eval_runtime": 4.7191,
"eval_samples_per_second": 17.164,
"eval_steps_per_second": 2.331,
"eval_weighted avg": {
"f1-score": 0.9425317916335554,
"precision": 0.942623096673485,
"recall": 0.9425587467362925,
"support": 30257.0
},
"step": 648
},
{
"epoch": 9.0,
"eval_B": {
"f1-score": 0.8885191347753745,
"precision": 0.8711256117455138,
"recall": 0.9066213921901528,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9620326487625066,
"precision": 0.9574445783763954,
"recall": 0.9666649029049156,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9275707898658718,
"precision": 0.9382914572864322,
"recall": 0.9170923379174852,
"support": 10180.0
},
"eval_accuracy": 0.9476484780381399,
"eval_loss": 0.24401217699050903,
"eval_macro avg": {
"f1-score": 0.9260408578012509,
"precision": 0.9222872158027805,
"recall": 0.9301262110041845,
"support": 30257.0
},
"eval_runtime": 4.701,
"eval_samples_per_second": 17.23,
"eval_steps_per_second": 2.34,
"eval_weighted avg": {
"f1-score": 0.94757580760031,
"precision": 0.9476398219436029,
"recall": 0.9476484780381399,
"support": 30257.0
},
"step": 729
},
{
"epoch": 10.0,
"eval_B": {
"f1-score": 0.875959595959596,
"precision": 0.8357748650732459,
"recall": 0.9202037351443124,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.955652945166171,
"precision": 0.9424749163879599,
"recall": 0.9692047198264458,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9127632580563309,
"precision": 0.9441469816272966,
"recall": 0.8833988212180747,
"support": 10180.0
},
"eval_accuracy": 0.9384274713289487,
"eval_loss": 0.29471203684806824,
"eval_macro avg": {
"f1-score": 0.9147919330606994,
"precision": 0.9074655876961675,
"recall": 0.9242690920629443,
"support": 30257.0
},
"eval_runtime": 4.6985,
"eval_samples_per_second": 17.24,
"eval_steps_per_second": 2.341,
"eval_weighted avg": {
"f1-score": 0.9381199518045186,
"precision": 0.9388833165494999,
"recall": 0.9384274713289487,
"support": 30257.0
},
"step": 810
},
{
"epoch": 11.0,
"eval_B": {
"f1-score": 0.875515251442704,
"precision": 0.8509615384615384,
"recall": 0.9015280135823429,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9575287538250502,
"precision": 0.9547582723972855,
"recall": 0.96031536060109,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9195242814667988,
"precision": 0.9278,
"recall": 0.9113948919449901,
"support": 10180.0
},
"eval_accuracy": 0.9415672406385299,
"eval_loss": 0.30941879749298096,
"eval_macro avg": {
"f1-score": 0.9175227622448511,
"precision": 0.9111732702862746,
"recall": 0.9244127553761411,
"support": 30257.0
},
"eval_runtime": 4.7017,
"eval_samples_per_second": 17.228,
"eval_steps_per_second": 2.34,
"eval_weighted avg": {
"f1-score": 0.9415490653426031,
"precision": 0.9416470001105195,
"recall": 0.9415672406385299,
"support": 30257.0
},
"step": 891
},
{
"epoch": 12.0,
"eval_B": {
"f1-score": 0.8818181818181817,
"precision": 0.8590982286634461,
"recall": 0.9057724957555179,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9542487087386278,
"precision": 0.9457436856875585,
"recall": 0.9629080903751521,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.910740239562973,
"precision": 0.9297042873222142,
"recall": 0.8925343811394892,
"support": 10180.0
},
"eval_accuracy": 0.9370063125888224,
"eval_loss": 0.28610122203826904,
"eval_macro avg": {
"f1-score": 0.9156023767065942,
"precision": 0.9115154005577396,
"recall": 0.9204049890900531,
"support": 30257.0
},
"eval_runtime": 4.6911,
"eval_samples_per_second": 17.267,
"eval_steps_per_second": 2.345,
"eval_weighted avg": {
"f1-score": 0.9367902899621314,
"precision": 0.9369738332985704,
"recall": 0.9370063125888224,
"support": 30257.0
},
"step": 972
},
{
"epoch": 12.35,
"grad_norm": 0.9797911643981934,
"learning_rate": 1.506172839506173e-05,
"loss": 0.0233,
"step": 1000
},
{
"epoch": 13.0,
"eval_B": {
"f1-score": 0.8875154511742892,
"precision": 0.8622898318654924,
"recall": 0.9142614601018676,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9602009576889865,
"precision": 0.9497877627083549,
"recall": 0.9708450182549341,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9229994967287367,
"precision": 0.9463364293085655,
"recall": 0.9007858546168959,
"support": 10180.0
},
"eval_accuracy": 0.9450705621839575,
"eval_loss": 0.3311145305633545,
"eval_macro avg": {
"f1-score": 0.9235719685306708,
"precision": 0.9194713412941375,
"recall": 0.9286307776578991,
"support": 30257.0
},
"eval_runtime": 4.7187,
"eval_samples_per_second": 17.166,
"eval_steps_per_second": 2.331,
"eval_weighted avg": {
"f1-score": 0.9448546114137558,
"precision": 0.9452199887538072,
"recall": 0.9450705621839575,
"support": 30257.0
},
"step": 1053
},
{
"epoch": 14.0,
"eval_B": {
"f1-score": 0.8758762886597938,
"precision": 0.8516439454691259,
"recall": 0.9015280135823429,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9588910891089109,
"precision": 0.9569456155143339,
"recall": 0.9608444891264089,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9223310576827941,
"precision": 0.9290412597169623,
"recall": 0.9157170923379175,
"support": 10180.0
},
"eval_accuracy": 0.9433519516145025,
"eval_loss": 0.32673054933547974,
"eval_macro avg": {
"f1-score": 0.9190328118171663,
"precision": 0.9125436069001407,
"recall": 0.9260298650155564,
"support": 30257.0
},
"eval_runtime": 4.7171,
"eval_samples_per_second": 17.171,
"eval_steps_per_second": 2.332,
"eval_weighted avg": {
"f1-score": 0.9433584006451857,
"precision": 0.943457440568685,
"recall": 0.9433519516145025,
"support": 30257.0
},
"step": 1134
},
{
"epoch": 15.0,
"eval_B": {
"f1-score": 0.8844750709363599,
"precision": 0.8463925523661753,
"recall": 0.9261460101867572,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9554282158321882,
"precision": 0.9367500508440105,
"recall": 0.974866395047357,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9105749486652978,
"precision": 0.9536559139784946,
"recall": 0.8712180746561886,
"support": 10180.0
},
"eval_accuracy": 0.9380969692963612,
"eval_loss": 0.4276476800441742,
"eval_macro avg": {
"f1-score": 0.9168260784779486,
"precision": 0.9122661723962269,
"recall": 0.924076826630101,
"support": 30257.0
},
"eval_runtime": 4.6973,
"eval_samples_per_second": 17.244,
"eval_steps_per_second": 2.342,
"eval_weighted avg": {
"f1-score": 0.9375748574540862,
"precision": 0.9389201454833387,
"recall": 0.9380969692963612,
"support": 30257.0
},
"step": 1215
},
{
"epoch": 16.0,
"eval_B": {
"f1-score": 0.8840336134453782,
"precision": 0.8752079866888519,
"recall": 0.8930390492359932,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9589266737513285,
"precision": 0.9629688917346994,
"recall": 0.9549182496428382,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.923489801893237,
"precision": 0.9174907892185379,
"recall": 0.9295677799607073,
"support": 10180.0
},
"eval_accuracy": 0.9439799054764186,
"eval_loss": 0.3447544574737549,
"eval_macro avg": {
"f1-score": 0.9221500296966479,
"precision": 0.9185558892140296,
"recall": 0.9258416929465129,
"support": 30257.0
},
"eval_runtime": 4.7359,
"eval_samples_per_second": 17.103,
"eval_steps_per_second": 2.323,
"eval_weighted avg": {
"f1-score": 0.9440880783665984,
"precision": 0.9442509279657028,
"recall": 0.9439799054764186,
"support": 30257.0
},
"step": 1296
},
{
"epoch": 17.0,
"eval_B": {
"f1-score": 0.8751040799333889,
"precision": 0.8586601307189542,
"recall": 0.8921901528013583,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.9534606521336053,
"precision": 0.954598493688342,
"recall": 0.9523255198687761,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9118326047448303,
"precision": 0.9118773946360154,
"recall": 0.9117878192534381,
"support": 10180.0
},
"eval_accuracy": 0.9363453085236474,
"eval_loss": 0.3998367488384247,
"eval_macro avg": {
"f1-score": 0.9134657789372748,
"precision": 0.9083786730144373,
"recall": 0.9187678306411908,
"support": 30257.0
},
"eval_runtime": 4.7114,
"eval_samples_per_second": 17.192,
"eval_steps_per_second": 2.335,
"eval_weighted avg": {
"f1-score": 0.9364041837306049,
"precision": 0.936489752572877,
"recall": 0.9363453085236474,
"support": 30257.0
},
"step": 1377
},
{
"epoch": 18.0,
"eval_B": {
"f1-score": 0.8807644370585792,
"precision": 0.8624898291293734,
"recall": 0.8998302207130731,
"support": 1178.0
},
"eval_I": {
"f1-score": 0.958089514470979,
"precision": 0.9545669415410474,
"recall": 0.961638181914387,
"support": 18899.0
},
"eval_O": {
"f1-score": 0.9206207546234321,
"precision": 0.9294223646010612,
"recall": 0.9119842829076621,
"support": 10180.0
},
"eval_accuracy": 0.9425256965330336,
"eval_loss": 0.3564019203186035,
"eval_macro avg": {
"f1-score": 0.9198249020509968,
"precision": 0.915493045090494,
"recall": 0.9244842285117074,
"support": 30257.0
},
"eval_runtime": 4.6986,
"eval_samples_per_second": 17.239,
"eval_steps_per_second": 2.341,
"eval_weighted avg": {
"f1-score": 0.9424726021386316,
"precision": 0.9425221706890129,
"recall": 0.9425256965330336,
"support": 30257.0
},
"step": 1458
}
],
"logging_steps": 500,
"max_steps": 4050,
"num_input_tokens_seen": 0,
"num_train_epochs": 50,
"save_steps": 500,
"total_flos": 2580121986044400.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}