| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 15.0, | |
| "eval_steps": 500, | |
| "global_step": 1215, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_B": { | |
| "f1-score": 0.7905866302864938, | |
| "precision": 0.7768096514745308, | |
| "recall": 0.8048611111111111, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9005331227169513, | |
| "precision": 0.9637593111099372, | |
| "recall": 0.8450919534905267, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.8288892663043479, | |
| "precision": 0.7463108800367001, | |
| "recall": 0.9320156593144275, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.8705373134328358, | |
| "eval_loss": 0.32930415868759155, | |
| "eval_macro avg": { | |
| "f1-score": 0.8400030064359308, | |
| "precision": 0.8289599475403894, | |
| "recall": 0.8606562413053552, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7349, | |
| "eval_samples_per_second": 17.107, | |
| "eval_steps_per_second": 2.323, | |
| "eval_weighted avg": { | |
| "f1-score": 0.8734092702599646, | |
| "precision": 0.8877430445874, | |
| "recall": 0.8705373134328358, | |
| "support": 33500.0 | |
| }, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_B": { | |
| "f1-score": 0.8656218739579861, | |
| "precision": 0.8325849903784477, | |
| "recall": 0.9013888888888889, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9547168948727912, | |
| "precision": 0.9423736462093862, | |
| "recall": 0.9673877796822161, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9013528191962081, | |
| "precision": 0.9332379102341274, | |
| "recall": 0.8715745249689678, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9345970149253732, | |
| "eval_loss": 0.2162298709154129, | |
| "eval_macro avg": { | |
| "f1-score": 0.9072305293423284, | |
| "precision": 0.902732182273987, | |
| "recall": 0.9134503978466909, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.753, | |
| "eval_samples_per_second": 17.042, | |
| "eval_steps_per_second": 2.314, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9342040950316517, | |
| "precision": 0.9347982961417612, | |
| "recall": 0.9345970149253732, | |
| "support": 33500.0 | |
| }, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_B": { | |
| "f1-score": 0.8603587759409076, | |
| "precision": 0.8717034925160371, | |
| "recall": 0.8493055555555555, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9434195218082959, | |
| "precision": 0.9677057963955188, | |
| "recall": 0.9203224162690509, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.8879310344827586, | |
| "precision": 0.8459410391631365, | |
| "recall": 0.9343072663038289, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9216417910447762, | |
| "eval_loss": 0.24082203209400177, | |
| "eval_macro avg": { | |
| "f1-score": 0.8972364440773206, | |
| "precision": 0.8951167760248975, | |
| "recall": 0.9013117460428117, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7554, | |
| "eval_samples_per_second": 17.033, | |
| "eval_steps_per_second": 2.313, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9225019575751797, | |
| "precision": 0.9255121957960801, | |
| "recall": 0.9216417910447762, | |
| "support": 33500.0 | |
| }, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_B": { | |
| "f1-score": 0.8804020100502512, | |
| "precision": 0.8504854368932039, | |
| "recall": 0.9125, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9589184992350025, | |
| "precision": 0.945595388218339, | |
| "recall": 0.9726224116366332, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9091178797468354, | |
| "precision": 0.9427750999897446, | |
| "recall": 0.877780960565263, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9403880597014925, | |
| "eval_loss": 0.22083505988121033, | |
| "eval_macro avg": { | |
| "f1-score": 0.9161461296773631, | |
| "precision": 0.9129519750337626, | |
| "recall": 0.9209677907339654, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7449, | |
| "eval_samples_per_second": 17.071, | |
| "eval_steps_per_second": 2.318, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9399744505088647, | |
| "precision": 0.9406253819936745, | |
| "recall": 0.9403880597014925, | |
| "support": 33500.0 | |
| }, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_B": { | |
| "f1-score": 0.8879781420765027, | |
| "precision": 0.8736559139784946, | |
| "recall": 0.9027777777777778, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9613201259451634, | |
| "precision": 0.9539317642765919, | |
| "recall": 0.9688238291564367, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.916006030835076, | |
| "precision": 0.9334853291038858, | |
| "recall": 0.899169292466342, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9442089552238806, | |
| "eval_loss": 0.2257838249206543, | |
| "eval_macro avg": { | |
| "f1-score": 0.921768099618914, | |
| "precision": 0.9203576691196576, | |
| "recall": 0.9235902998001855, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7845, | |
| "eval_samples_per_second": 16.93, | |
| "eval_steps_per_second": 2.299, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9440011117702735, | |
| "precision": 0.9440889959185916, | |
| "recall": 0.9442089552238806, | |
| "support": 33500.0 | |
| }, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_B": { | |
| "f1-score": 0.8873579056148813, | |
| "precision": 0.8803827751196173, | |
| "recall": 0.8944444444444445, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9622060984570168, | |
| "precision": 0.9539267015706806, | |
| "recall": 0.9706304720433594, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9182769530299342, | |
| "precision": 0.9365567911040509, | |
| "recall": 0.9006970304592762, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9454925373134329, | |
| "eval_loss": 0.2242102175951004, | |
| "eval_macro avg": { | |
| "f1-score": 0.9226136523672773, | |
| "precision": 0.923622089264783, | |
| "recall": 0.9219239823156933, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7715, | |
| "eval_samples_per_second": 16.976, | |
| "eval_steps_per_second": 2.305, | |
| "eval_weighted avg": { | |
| "f1-score": 0.945255312255509, | |
| "precision": 0.9453351097376493, | |
| "recall": 0.9454925373134329, | |
| "support": 33500.0 | |
| }, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "grad_norm": 1.8705081939697266, | |
| "learning_rate": 1.7530864197530865e-05, | |
| "loss": 0.1601, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_B": { | |
| "f1-score": 0.8876749743939911, | |
| "precision": 0.8730691739422431, | |
| "recall": 0.9027777777777778, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9611415113943341, | |
| "precision": 0.9581108451482232, | |
| "recall": 0.9641914114976606, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9164739884393064, | |
| "precision": 0.9247594050743657, | |
| "recall": 0.9083357204239473, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.944089552238806, | |
| "eval_loss": 0.22517083585262299, | |
| "eval_macro avg": { | |
| "f1-score": 0.9217634914092105, | |
| "precision": 0.9186464747216107, | |
| "recall": 0.9251016365664619, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7926, | |
| "eval_samples_per_second": 16.901, | |
| "eval_steps_per_second": 2.295, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9440192791200506, | |
| "precision": 0.9440287663891153, | |
| "recall": 0.944089552238806, | |
| "support": 33500.0 | |
| }, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_B": { | |
| "f1-score": 0.894540522210919, | |
| "precision": 0.8740888005301524, | |
| "recall": 0.9159722222222222, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9638018385639943, | |
| "precision": 0.9540255968049378, | |
| "recall": 0.9737805160513272, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9203132648066569, | |
| "precision": 0.9441598875163202, | |
| "recall": 0.8976415544734078, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9474925373134329, | |
| "eval_loss": 0.24869142472743988, | |
| "eval_macro avg": { | |
| "f1-score": 0.9262185418605234, | |
| "precision": 0.9240914282838034, | |
| "recall": 0.9291314309156524, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7743, | |
| "eval_samples_per_second": 16.966, | |
| "eval_steps_per_second": 2.304, | |
| "eval_weighted avg": { | |
| "f1-score": 0.947228939205516, | |
| "precision": 0.9475052218791054, | |
| "recall": 0.9474925373134329, | |
| "support": 33500.0 | |
| }, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_B": { | |
| "f1-score": 0.8851236868858015, | |
| "precision": 0.8643282594308405, | |
| "recall": 0.9069444444444444, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9617001723147618, | |
| "precision": 0.9540067462849849, | |
| "recall": 0.9695186918052532, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9165854609237966, | |
| "precision": 0.9358272808675754, | |
| "recall": 0.8981189725961998, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9445074626865672, | |
| "eval_loss": 0.3045330345630646, | |
| "eval_macro avg": { | |
| "f1-score": 0.9211364400414532, | |
| "precision": 0.9180540955278002, | |
| "recall": 0.9248607029486324, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7599, | |
| "eval_samples_per_second": 17.017, | |
| "eval_steps_per_second": 2.311, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9443044555560967, | |
| "precision": 0.9444685205421044, | |
| "recall": 0.9445074626865672, | |
| "support": 33500.0 | |
| }, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_B": { | |
| "f1-score": 0.8815068493150685, | |
| "precision": 0.8695945945945946, | |
| "recall": 0.89375, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.956297959947595, | |
| "precision": 0.9660159757999717, | |
| "recall": 0.9467735211006625, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9101049868766404, | |
| "precision": 0.8937678357728068, | |
| "recall": 0.9270505108373914, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9383283582089552, | |
| "eval_loss": 0.32627996802330017, | |
| "eval_macro avg": { | |
| "f1-score": 0.9159699320464346, | |
| "precision": 0.909792802055791, | |
| "recall": 0.9225246773126846, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7534, | |
| "eval_samples_per_second": 17.04, | |
| "eval_steps_per_second": 2.314, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9386418940884026, | |
| "precision": 0.9392845859659347, | |
| "recall": 0.9383283582089552, | |
| "support": 33500.0 | |
| }, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_B": { | |
| "f1-score": 0.8901136755080951, | |
| "precision": 0.8831168831168831, | |
| "recall": 0.8972222222222223, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9620073001078947, | |
| "precision": 0.9535359970874671, | |
| "recall": 0.9706304720433594, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.917413322945072, | |
| "precision": 0.9361025539103647, | |
| "recall": 0.8994557433400172, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9452238805970149, | |
| "eval_loss": 0.3123624622821808, | |
| "eval_macro avg": { | |
| "f1-score": 0.9231780995203539, | |
| "precision": 0.9242518113715716, | |
| "recall": 0.922436145868533, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7539, | |
| "eval_samples_per_second": 17.039, | |
| "eval_steps_per_second": 2.314, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9449756719810304, | |
| "precision": 0.9450588635199616, | |
| "recall": 0.9452238805970149, | |
| "support": 33500.0 | |
| }, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_B": { | |
| "f1-score": 0.8800819952169457, | |
| "precision": 0.8661735036987223, | |
| "recall": 0.8944444444444445, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9595463311598861, | |
| "precision": 0.9512039692293687, | |
| "recall": 0.9680363181544448, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9115367743822197, | |
| "precision": 0.9310035842293907, | |
| "recall": 0.8928673732454884, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9413731343283582, | |
| "eval_loss": 0.3427491784095764, | |
| "eval_macro avg": { | |
| "f1-score": 0.9170550335863505, | |
| "precision": 0.9161270190524938, | |
| "recall": 0.9184493786147926, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7455, | |
| "eval_samples_per_second": 17.069, | |
| "eval_steps_per_second": 2.318, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9411214734915179, | |
| "precision": 0.9412337452750731, | |
| "recall": 0.9413731343283582, | |
| "support": 33500.0 | |
| }, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 12.35, | |
| "grad_norm": 0.47551417350769043, | |
| "learning_rate": 1.506172839506173e-05, | |
| "loss": 0.0222, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_B": { | |
| "f1-score": 0.8896434634974534, | |
| "precision": 0.8704318936877077, | |
| "recall": 0.9097222222222222, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9620136667970457, | |
| "precision": 0.9556591698665204, | |
| "recall": 0.9684532357437347, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9183177933177933, | |
| "precision": 0.9343808676746714, | |
| "recall": 0.9027976701995608, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9454029850746268, | |
| "eval_loss": 0.3816670775413513, | |
| "eval_macro avg": { | |
| "f1-score": 0.9233249745374308, | |
| "precision": 0.9201573104096331, | |
| "recall": 0.9269910427218392, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7247, | |
| "eval_samples_per_second": 17.144, | |
| "eval_steps_per_second": 2.328, | |
| "eval_weighted avg": { | |
| "f1-score": 0.945242324238848, | |
| "precision": 0.9453435001186778, | |
| "recall": 0.9454029850746268, | |
| "support": 33500.0 | |
| }, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_B": { | |
| "f1-score": 0.8865272540281112, | |
| "precision": 0.8754231550440081, | |
| "recall": 0.8979166666666667, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9599206715247671, | |
| "precision": 0.9557331129172981, | |
| "recall": 0.9641450873210728, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9141367826632559, | |
| "precision": 0.9242631270739801, | |
| "recall": 0.9042299245679366, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9425671641791045, | |
| "eval_loss": 0.3827010989189148, | |
| "eval_macro avg": { | |
| "f1-score": 0.9201949027387114, | |
| "precision": 0.9184731316784287, | |
| "recall": 0.9220972261852255, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7319, | |
| "eval_samples_per_second": 17.118, | |
| "eval_steps_per_second": 2.325, | |
| "eval_weighted avg": { | |
| "f1-score": 0.9424525763235196, | |
| "precision": 0.9424426203479068, | |
| "recall": 0.9425671641791045, | |
| "support": 33500.0 | |
| }, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_B": { | |
| "f1-score": 0.8871919472405415, | |
| "precision": 0.8868841082581541, | |
| "recall": 0.8875, | |
| "support": 1440.0 | |
| }, | |
| "eval_I": { | |
| "f1-score": 0.9616856471022177, | |
| "precision": 0.9571841585975862, | |
| "recall": 0.9662296752675221, | |
| "support": 21587.0 | |
| }, | |
| "eval_O": { | |
| "f1-score": 0.9181813798756087, | |
| "precision": 0.9273470977795092, | |
| "recall": 0.9091950730449728, | |
| "support": 10473.0 | |
| }, | |
| "eval_accuracy": 0.9450149253731344, | |
| "eval_loss": 0.39016807079315186, | |
| "eval_macro avg": { | |
| "f1-score": 0.9223529914061226, | |
| "precision": 0.9238051215450831, | |
| "recall": 0.920974916104165, | |
| "support": 33500.0 | |
| }, | |
| "eval_runtime": 4.7376, | |
| "eval_samples_per_second": 17.097, | |
| "eval_steps_per_second": 2.322, | |
| "eval_weighted avg": { | |
| "f1-score": 0.944882927148036, | |
| "precision": 0.9448344388830637, | |
| "recall": 0.9450149253731344, | |
| "support": 33500.0 | |
| }, | |
| "step": 1215 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 4050, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 50, | |
| "save_steps": 500, | |
| "total_flos": 2150101655037000.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |