| { |
| "best_metric": 95.33333333333333, |
| "best_model_checkpoint": "../results/phrase_retrieval/PR-pass/qa/mpsDistillbert/finetuned/checkpoint-4500", |
| "epoch": 2.0, |
| "eval_steps": 100, |
| "global_step": 5068, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.03946329913180742, |
| "grad_norm": 67.25347137451172, |
| "learning_rate": 2.940805051302289e-05, |
| "loss": 2.8412, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.03946329913180742, |
| "eval_exact_match": 69.43333333333334, |
| "eval_f1": 73.48511445732589, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.07892659826361484, |
| "grad_norm": 28.299945831298828, |
| "learning_rate": 2.8816101026045778e-05, |
| "loss": 0.7987, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.07892659826361484, |
| "eval_exact_match": 85.1, |
| "eval_f1": 88.92439170564563, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.11838989739542226, |
| "grad_norm": 35.28098678588867, |
| "learning_rate": 2.8224151539068667e-05, |
| "loss": 0.6134, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.11838989739542226, |
| "eval_exact_match": 89.56666666666666, |
| "eval_f1": 92.56194565862985, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.15785319652722968, |
| "grad_norm": 28.96463966369629, |
| "learning_rate": 2.7632202052091556e-05, |
| "loss": 0.4986, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.15785319652722968, |
| "eval_exact_match": 89.2, |
| "eval_f1": 91.95230270592397, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.19731649565903708, |
| "grad_norm": 59.035316467285156, |
| "learning_rate": 2.7040252565114444e-05, |
| "loss": 0.5004, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.19731649565903708, |
| "eval_exact_match": 89.96666666666667, |
| "eval_f1": 92.31457766489807, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.23677979479084452, |
| "grad_norm": 21.016122817993164, |
| "learning_rate": 2.6448303078137333e-05, |
| "loss": 0.4413, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.23677979479084452, |
| "eval_exact_match": 91.33333333333333, |
| "eval_f1": 93.73205210366977, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.27624309392265195, |
| "grad_norm": 50.69818878173828, |
| "learning_rate": 2.585635359116022e-05, |
| "loss": 0.3997, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.27624309392265195, |
| "eval_exact_match": 92.1, |
| "eval_f1": 94.01741907702439, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.31570639305445936, |
| "grad_norm": 103.70169067382812, |
| "learning_rate": 2.5264404104183114e-05, |
| "loss": 0.3915, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.31570639305445936, |
| "eval_exact_match": 91.03333333333333, |
| "eval_f1": 93.42441436925648, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.35516969218626676, |
| "grad_norm": 10.776900291442871, |
| "learning_rate": 2.4672454617206e-05, |
| "loss": 0.3936, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.35516969218626676, |
| "eval_exact_match": 92.6, |
| "eval_f1": 94.21299117549118, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.39463299131807417, |
| "grad_norm": 70.64266204833984, |
| "learning_rate": 2.4080505130228888e-05, |
| "loss": 0.4522, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.39463299131807417, |
| "eval_exact_match": 92.0, |
| "eval_f1": 93.84013468013471, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.4340962904498816, |
| "grad_norm": 122.04886627197266, |
| "learning_rate": 2.3488555643251776e-05, |
| "loss": 0.5036, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.4340962904498816, |
| "eval_exact_match": 92.3, |
| "eval_f1": 94.030187928014, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.47355958958168903, |
| "grad_norm": 44.26261901855469, |
| "learning_rate": 2.2896606156274665e-05, |
| "loss": 0.4178, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.47355958958168903, |
| "eval_exact_match": 93.2, |
| "eval_f1": 95.1092140822141, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.5130228887134964, |
| "grad_norm": 50.030704498291016, |
| "learning_rate": 2.2304656669297554e-05, |
| "loss": 0.4134, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.5130228887134964, |
| "eval_exact_match": 92.93333333333334, |
| "eval_f1": 94.6116678691679, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.5524861878453039, |
| "grad_norm": 6.0724310874938965, |
| "learning_rate": 2.1712707182320442e-05, |
| "loss": 0.2996, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.5524861878453039, |
| "eval_exact_match": 92.43333333333334, |
| "eval_f1": 94.43055409989623, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.5919494869771112, |
| "grad_norm": 138.5492706298828, |
| "learning_rate": 2.112075769534333e-05, |
| "loss": 0.3272, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.5919494869771112, |
| "eval_exact_match": 92.66666666666667, |
| "eval_f1": 94.70460317460318, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.6314127861089187, |
| "grad_norm": 10.509523391723633, |
| "learning_rate": 2.052880820836622e-05, |
| "loss": 0.435, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.6314127861089187, |
| "eval_exact_match": 93.63333333333334, |
| "eval_f1": 95.12072967772967, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.6708760852407262, |
| "grad_norm": 63.85009765625, |
| "learning_rate": 1.993685872138911e-05, |
| "loss": 0.3181, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.6708760852407262, |
| "eval_exact_match": 93.76666666666667, |
| "eval_f1": 95.14681913116127, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.7103393843725335, |
| "grad_norm": 75.05247497558594, |
| "learning_rate": 1.9344909234411997e-05, |
| "loss": 0.2922, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.7103393843725335, |
| "eval_exact_match": 94.23333333333333, |
| "eval_f1": 95.71164746364745, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.749802683504341, |
| "grad_norm": 63.5960693359375, |
| "learning_rate": 1.8752959747434886e-05, |
| "loss": 0.2787, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.749802683504341, |
| "eval_exact_match": 94.06666666666666, |
| "eval_f1": 95.59337810337813, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.7892659826361483, |
| "grad_norm": 153.86695861816406, |
| "learning_rate": 1.8161010260457775e-05, |
| "loss": 0.3093, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.7892659826361483, |
| "eval_exact_match": 93.43333333333334, |
| "eval_f1": 95.11906084656084, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.8287292817679558, |
| "grad_norm": 88.632080078125, |
| "learning_rate": 1.7569060773480663e-05, |
| "loss": 0.3384, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.8287292817679558, |
| "eval_exact_match": 94.06666666666666, |
| "eval_f1": 95.70005291005292, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.8681925808997633, |
| "grad_norm": 17.5891056060791, |
| "learning_rate": 1.6977111286503552e-05, |
| "loss": 0.351, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.8681925808997633, |
| "eval_exact_match": 94.66666666666667, |
| "eval_f1": 96.17507936507938, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.9076558800315706, |
| "grad_norm": 324.85894775390625, |
| "learning_rate": 1.638516179952644e-05, |
| "loss": 0.3736, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.9076558800315706, |
| "eval_exact_match": 94.33333333333333, |
| "eval_f1": 95.90269841269844, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.9471191791633781, |
| "grad_norm": 53.7849006652832, |
| "learning_rate": 1.5793212312549333e-05, |
| "loss": 0.2718, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.9471191791633781, |
| "eval_exact_match": 94.43333333333334, |
| "eval_f1": 95.9603174603175, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.9865824782951855, |
| "grad_norm": 9.57298469543457, |
| "learning_rate": 1.5201262825572218e-05, |
| "loss": 0.296, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.9865824782951855, |
| "eval_exact_match": 94.7, |
| "eval_f1": 95.994126984127, |
| "step": 2500 |
| }, |
| { |
| "epoch": 1.0260457774269929, |
| "grad_norm": 3.4919021129608154, |
| "learning_rate": 1.4609313338595107e-05, |
| "loss": 0.1504, |
| "step": 2600 |
| }, |
| { |
| "epoch": 1.0260457774269929, |
| "eval_exact_match": 94.7, |
| "eval_f1": 96.06396825396828, |
| "step": 2600 |
| }, |
| { |
| "epoch": 1.0655090765588002, |
| "grad_norm": 0.06277454644441605, |
| "learning_rate": 1.4017363851617995e-05, |
| "loss": 0.151, |
| "step": 2700 |
| }, |
| { |
| "epoch": 1.0655090765588002, |
| "eval_exact_match": 94.4, |
| "eval_f1": 95.89519140989731, |
| "step": 2700 |
| }, |
| { |
| "epoch": 1.1049723756906078, |
| "grad_norm": 0.00011412293679313734, |
| "learning_rate": 1.3425414364640886e-05, |
| "loss": 0.1005, |
| "step": 2800 |
| }, |
| { |
| "epoch": 1.1049723756906078, |
| "eval_exact_match": 94.9, |
| "eval_f1": 96.31682539682541, |
| "step": 2800 |
| }, |
| { |
| "epoch": 1.1444356748224151, |
| "grad_norm": 19.318458557128906, |
| "learning_rate": 1.2833464877663773e-05, |
| "loss": 0.1291, |
| "step": 2900 |
| }, |
| { |
| "epoch": 1.1444356748224151, |
| "eval_exact_match": 93.93333333333334, |
| "eval_f1": 95.529126984127, |
| "step": 2900 |
| }, |
| { |
| "epoch": 1.1838989739542225, |
| "grad_norm": 4.399935245513916, |
| "learning_rate": 1.2241515390686661e-05, |
| "loss": 0.1054, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1.1838989739542225, |
| "eval_exact_match": 93.86666666666666, |
| "eval_f1": 95.59808897243109, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1.22336227308603, |
| "grad_norm": 40.74488830566406, |
| "learning_rate": 1.164956590370955e-05, |
| "loss": 0.1713, |
| "step": 3100 |
| }, |
| { |
| "epoch": 1.22336227308603, |
| "eval_exact_match": 94.43333333333334, |
| "eval_f1": 96.00482804232804, |
| "step": 3100 |
| }, |
| { |
| "epoch": 1.2628255722178374, |
| "grad_norm": 0.04602056369185448, |
| "learning_rate": 1.1057616416732439e-05, |
| "loss": 0.0881, |
| "step": 3200 |
| }, |
| { |
| "epoch": 1.2628255722178374, |
| "eval_exact_match": 94.46666666666667, |
| "eval_f1": 96.07236772486773, |
| "step": 3200 |
| }, |
| { |
| "epoch": 1.3022888713496448, |
| "grad_norm": 78.22419738769531, |
| "learning_rate": 1.0465666929755327e-05, |
| "loss": 0.1207, |
| "step": 3300 |
| }, |
| { |
| "epoch": 1.3022888713496448, |
| "eval_exact_match": 94.7, |
| "eval_f1": 96.0661904761905, |
| "step": 3300 |
| }, |
| { |
| "epoch": 1.3417521704814521, |
| "grad_norm": 0.2205381691455841, |
| "learning_rate": 9.873717442778216e-06, |
| "loss": 0.1689, |
| "step": 3400 |
| }, |
| { |
| "epoch": 1.3417521704814521, |
| "eval_exact_match": 95.2, |
| "eval_f1": 96.48216304650515, |
| "step": 3400 |
| }, |
| { |
| "epoch": 1.3812154696132597, |
| "grad_norm": 46.83411407470703, |
| "learning_rate": 9.281767955801105e-06, |
| "loss": 0.0989, |
| "step": 3500 |
| }, |
| { |
| "epoch": 1.3812154696132597, |
| "eval_exact_match": 94.36666666666666, |
| "eval_f1": 95.95884920634923, |
| "step": 3500 |
| }, |
| { |
| "epoch": 1.420678768745067, |
| "grad_norm": 0.009372607804834843, |
| "learning_rate": 8.689818468823995e-06, |
| "loss": 0.0866, |
| "step": 3600 |
| }, |
| { |
| "epoch": 1.420678768745067, |
| "eval_exact_match": 94.56666666666666, |
| "eval_f1": 96.29847883597886, |
| "step": 3600 |
| }, |
| { |
| "epoch": 1.4601420678768746, |
| "grad_norm": 58.012062072753906, |
| "learning_rate": 8.097868981846882e-06, |
| "loss": 0.1983, |
| "step": 3700 |
| }, |
| { |
| "epoch": 1.4601420678768746, |
| "eval_exact_match": 94.6, |
| "eval_f1": 96.33773809523815, |
| "step": 3700 |
| }, |
| { |
| "epoch": 1.499605367008682, |
| "grad_norm": 0.0017438705544918776, |
| "learning_rate": 7.505919494869771e-06, |
| "loss": 0.1228, |
| "step": 3800 |
| }, |
| { |
| "epoch": 1.499605367008682, |
| "eval_exact_match": 95.1, |
| "eval_f1": 96.49228021978023, |
| "step": 3800 |
| }, |
| { |
| "epoch": 1.5390686661404893, |
| "grad_norm": 0.00036309007555246353, |
| "learning_rate": 6.91397000789266e-06, |
| "loss": 0.0761, |
| "step": 3900 |
| }, |
| { |
| "epoch": 1.5390686661404893, |
| "eval_exact_match": 94.83333333333333, |
| "eval_f1": 96.32456349206349, |
| "step": 3900 |
| }, |
| { |
| "epoch": 1.5785319652722967, |
| "grad_norm": 0.058913152664899826, |
| "learning_rate": 6.322020520915548e-06, |
| "loss": 0.1332, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1.5785319652722967, |
| "eval_exact_match": 94.76666666666667, |
| "eval_f1": 96.26157647907651, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1.617995264404104, |
| "grad_norm": 8.819950016913936e-05, |
| "learning_rate": 5.730071033938438e-06, |
| "loss": 0.1538, |
| "step": 4100 |
| }, |
| { |
| "epoch": 1.617995264404104, |
| "eval_exact_match": 95.33333333333333, |
| "eval_f1": 96.71250000000002, |
| "step": 4100 |
| }, |
| { |
| "epoch": 1.6574585635359116, |
| "grad_norm": 0.2561735510826111, |
| "learning_rate": 5.1381215469613265e-06, |
| "loss": 0.0977, |
| "step": 4200 |
| }, |
| { |
| "epoch": 1.6574585635359116, |
| "eval_exact_match": 94.83333333333333, |
| "eval_f1": 96.35535714285714, |
| "step": 4200 |
| }, |
| { |
| "epoch": 1.6969218626677192, |
| "grad_norm": 0.0023131032940000296, |
| "learning_rate": 4.546172059984215e-06, |
| "loss": 0.1299, |
| "step": 4300 |
| }, |
| { |
| "epoch": 1.6969218626677192, |
| "eval_exact_match": 94.66666666666667, |
| "eval_f1": 96.20515873015874, |
| "step": 4300 |
| }, |
| { |
| "epoch": 1.7363851617995265, |
| "grad_norm": 0.04482404515147209, |
| "learning_rate": 3.954222573007103e-06, |
| "loss": 0.0895, |
| "step": 4400 |
| }, |
| { |
| "epoch": 1.7363851617995265, |
| "eval_exact_match": 95.26666666666667, |
| "eval_f1": 96.61154761904761, |
| "step": 4400 |
| }, |
| { |
| "epoch": 1.7758484609313339, |
| "grad_norm": 11.488630294799805, |
| "learning_rate": 3.362273086029992e-06, |
| "loss": 0.1245, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1.7758484609313339, |
| "eval_exact_match": 95.33333333333333, |
| "eval_f1": 96.5898015873016, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1.8153117600631412, |
| "grad_norm": 0.000506121723446995, |
| "learning_rate": 2.7703235990528812e-06, |
| "loss": 0.1046, |
| "step": 4600 |
| }, |
| { |
| "epoch": 1.8153117600631412, |
| "eval_exact_match": 95.3, |
| "eval_f1": 96.55313492063492, |
| "step": 4600 |
| }, |
| { |
| "epoch": 1.8547750591949486, |
| "grad_norm": 12.496238708496094, |
| "learning_rate": 2.1783741120757695e-06, |
| "loss": 0.0985, |
| "step": 4700 |
| }, |
| { |
| "epoch": 1.8547750591949486, |
| "eval_exact_match": 95.1, |
| "eval_f1": 96.48452380952381, |
| "step": 4700 |
| }, |
| { |
| "epoch": 1.8942383583267561, |
| "grad_norm": 97.22492218017578, |
| "learning_rate": 1.5864246250986582e-06, |
| "loss": 0.1193, |
| "step": 4800 |
| }, |
| { |
| "epoch": 1.8942383583267561, |
| "eval_exact_match": 95.26666666666667, |
| "eval_f1": 96.62202380952381, |
| "step": 4800 |
| }, |
| { |
| "epoch": 1.9337016574585635, |
| "grad_norm": 0.0009776244405657053, |
| "learning_rate": 9.944751381215469e-07, |
| "loss": 0.0811, |
| "step": 4900 |
| }, |
| { |
| "epoch": 1.9337016574585635, |
| "eval_exact_match": 95.1, |
| "eval_f1": 96.48424603174604, |
| "step": 4900 |
| }, |
| { |
| "epoch": 1.973164956590371, |
| "grad_norm": 0.05864416062831879, |
| "learning_rate": 4.025256511444357e-07, |
| "loss": 0.0894, |
| "step": 5000 |
| }, |
| { |
| "epoch": 1.973164956590371, |
| "eval_exact_match": 95.2, |
| "eval_f1": 96.51480158730159, |
| "step": 5000 |
| }, |
| { |
| "epoch": 2.0, |
| "step": 5068, |
| "total_flos": 3.764792404688486e+16, |
| "train_loss": 0.30838469322403306, |
| "train_runtime": 8673.587, |
| "train_samples_per_second": 4.674, |
| "train_steps_per_second": 0.584 |
| } |
| ], |
| "logging_steps": 100, |
| "max_steps": 5068, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 2, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 3.764792404688486e+16, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
