| { |
| "best_metric": 0.2911, |
| "best_model_checkpoint": "logs_T5/google-t5/t5-base/checkpoint-17640", |
| "epoch": 10.0, |
| "eval_steps": 500, |
| "global_step": 19600, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.03, |
| "grad_norm": 1.1786425113677979, |
| "learning_rate": 9.974489795918368e-05, |
| "loss": 2.143, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 0.9633323550224304, |
| "learning_rate": 9.948979591836736e-05, |
| "loss": 1.901, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 1.6018619537353516, |
| "learning_rate": 9.923469387755102e-05, |
| "loss": 1.9064, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 1.0619288682937622, |
| "learning_rate": 9.897959183673469e-05, |
| "loss": 1.8983, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.13, |
| "grad_norm": 1.1127651929855347, |
| "learning_rate": 9.872448979591837e-05, |
| "loss": 1.8773, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.15, |
| "grad_norm": 0.9425650238990784, |
| "learning_rate": 9.846938775510204e-05, |
| "loss": 1.8902, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.18, |
| "grad_norm": 1.0420337915420532, |
| "learning_rate": 9.821428571428572e-05, |
| "loss": 1.8951, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.2, |
| "grad_norm": 1.08526611328125, |
| "learning_rate": 9.79591836734694e-05, |
| "loss": 1.8587, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.23, |
| "grad_norm": 0.9797475337982178, |
| "learning_rate": 9.770408163265307e-05, |
| "loss": 1.8934, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.26, |
| "grad_norm": 1.2622222900390625, |
| "learning_rate": 9.744897959183674e-05, |
| "loss": 1.8724, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.28, |
| "grad_norm": 1.0583791732788086, |
| "learning_rate": 9.719387755102042e-05, |
| "loss": 1.8596, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.31, |
| "grad_norm": 1.0346555709838867, |
| "learning_rate": 9.693877551020408e-05, |
| "loss": 1.8985, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.33, |
| "grad_norm": 0.9425589442253113, |
| "learning_rate": 9.668367346938776e-05, |
| "loss": 1.8776, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.36, |
| "grad_norm": 1.1392202377319336, |
| "learning_rate": 9.642857142857143e-05, |
| "loss": 1.8504, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.38, |
| "grad_norm": 1.0559316873550415, |
| "learning_rate": 9.617346938775511e-05, |
| "loss": 1.845, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.41, |
| "grad_norm": 1.1446446180343628, |
| "learning_rate": 9.591836734693878e-05, |
| "loss": 1.8547, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.43, |
| "grad_norm": 1.0492786169052124, |
| "learning_rate": 9.566326530612246e-05, |
| "loss": 1.8786, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.46, |
| "grad_norm": 1.4957969188690186, |
| "learning_rate": 9.540816326530613e-05, |
| "loss": 1.8474, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.48, |
| "grad_norm": 1.0090183019638062, |
| "learning_rate": 9.515306122448981e-05, |
| "loss": 1.8548, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.51, |
| "grad_norm": 1.0634762048721313, |
| "learning_rate": 9.489795918367348e-05, |
| "loss": 1.8362, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.54, |
| "grad_norm": 0.9056524038314819, |
| "learning_rate": 9.464285714285715e-05, |
| "loss": 1.84, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.56, |
| "grad_norm": 0.8953872919082642, |
| "learning_rate": 9.438775510204082e-05, |
| "loss": 1.8572, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.59, |
| "grad_norm": 0.8523529171943665, |
| "learning_rate": 9.41326530612245e-05, |
| "loss": 1.8388, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.61, |
| "grad_norm": 0.9457581043243408, |
| "learning_rate": 9.387755102040817e-05, |
| "loss": 1.8485, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 2.3593177795410156, |
| "learning_rate": 9.362244897959183e-05, |
| "loss": 1.8342, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.66, |
| "grad_norm": 0.9455676674842834, |
| "learning_rate": 9.336734693877551e-05, |
| "loss": 1.8403, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.69, |
| "grad_norm": 0.9168422818183899, |
| "learning_rate": 9.311224489795918e-05, |
| "loss": 1.8373, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.71, |
| "grad_norm": 1.0216107368469238, |
| "learning_rate": 9.285714285714286e-05, |
| "loss": 1.8507, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.74, |
| "grad_norm": 1.0461299419403076, |
| "learning_rate": 9.260204081632653e-05, |
| "loss": 1.8382, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.77, |
| "grad_norm": 1.0399153232574463, |
| "learning_rate": 9.234693877551021e-05, |
| "loss": 1.8455, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.79, |
| "grad_norm": 0.998569667339325, |
| "learning_rate": 9.209183673469388e-05, |
| "loss": 1.8335, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.82, |
| "grad_norm": 0.9193490743637085, |
| "learning_rate": 9.183673469387756e-05, |
| "loss": 1.8678, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.84, |
| "grad_norm": 0.8332929611206055, |
| "learning_rate": 9.158163265306124e-05, |
| "loss": 1.8187, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.87, |
| "grad_norm": 0.793361485004425, |
| "learning_rate": 9.13265306122449e-05, |
| "loss": 1.8305, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.89, |
| "grad_norm": 1.3230149745941162, |
| "learning_rate": 9.107142857142857e-05, |
| "loss": 1.8641, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.92, |
| "grad_norm": 0.9166779518127441, |
| "learning_rate": 9.081632653061225e-05, |
| "loss": 1.8556, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.94, |
| "grad_norm": 0.8884773850440979, |
| "learning_rate": 9.056122448979592e-05, |
| "loss": 1.8376, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.97, |
| "grad_norm": 0.9267615675926208, |
| "learning_rate": 9.030612244897958e-05, |
| "loss": 1.8305, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.99, |
| "grad_norm": 0.9181408882141113, |
| "learning_rate": 9.005102040816327e-05, |
| "loss": 1.839, |
| "step": 1950 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_bert-score-f1": 0.35383370643719175, |
| "eval_bert-score-precision": 0.3846865674866766, |
| "eval_bert-score-recall": 0.32294123599954094, |
| "eval_gen_len": 47.14247078634485, |
| "eval_loss": 1.6373140811920166, |
| "eval_rouge1": 0.3751, |
| "eval_rouge2": 0.1562, |
| "eval_rougeL": 0.2812, |
| "eval_rougeLsum": 0.2813, |
| "eval_runtime": 1961.5884, |
| "eval_samples_per_second": 9.99, |
| "eval_steps_per_second": 0.25, |
| "step": 1960 |
| }, |
| { |
| "epoch": 1.02, |
| "grad_norm": 0.9284695386886597, |
| "learning_rate": 8.979591836734695e-05, |
| "loss": 1.7616, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.05, |
| "grad_norm": 1.0442299842834473, |
| "learning_rate": 8.954081632653062e-05, |
| "loss": 1.7602, |
| "step": 2050 |
| }, |
| { |
| "epoch": 1.07, |
| "grad_norm": 0.9256122708320618, |
| "learning_rate": 8.92857142857143e-05, |
| "loss": 1.7731, |
| "step": 2100 |
| }, |
| { |
| "epoch": 1.1, |
| "grad_norm": 1.0128663778305054, |
| "learning_rate": 8.903061224489796e-05, |
| "loss": 1.7481, |
| "step": 2150 |
| }, |
| { |
| "epoch": 1.12, |
| "grad_norm": 0.9690511226654053, |
| "learning_rate": 8.877551020408164e-05, |
| "loss": 1.7554, |
| "step": 2200 |
| }, |
| { |
| "epoch": 1.15, |
| "grad_norm": 1.0306673049926758, |
| "learning_rate": 8.852040816326531e-05, |
| "loss": 1.7462, |
| "step": 2250 |
| }, |
| { |
| "epoch": 1.17, |
| "grad_norm": 0.9653202891349792, |
| "learning_rate": 8.826530612244899e-05, |
| "loss": 1.742, |
| "step": 2300 |
| }, |
| { |
| "epoch": 1.2, |
| "grad_norm": 1.2552508115768433, |
| "learning_rate": 8.801020408163265e-05, |
| "loss": 1.7444, |
| "step": 2350 |
| }, |
| { |
| "epoch": 1.22, |
| "grad_norm": 1.1486296653747559, |
| "learning_rate": 8.775510204081632e-05, |
| "loss": 1.7483, |
| "step": 2400 |
| }, |
| { |
| "epoch": 1.25, |
| "grad_norm": 0.958289384841919, |
| "learning_rate": 8.75e-05, |
| "loss": 1.7648, |
| "step": 2450 |
| }, |
| { |
| "epoch": 1.28, |
| "grad_norm": 1.4296537637710571, |
| "learning_rate": 8.724489795918367e-05, |
| "loss": 1.7658, |
| "step": 2500 |
| }, |
| { |
| "epoch": 1.3, |
| "grad_norm": 0.9161971211433411, |
| "learning_rate": 8.698979591836735e-05, |
| "loss": 1.7525, |
| "step": 2550 |
| }, |
| { |
| "epoch": 1.33, |
| "grad_norm": 1.0688705444335938, |
| "learning_rate": 8.673469387755102e-05, |
| "loss": 1.7425, |
| "step": 2600 |
| }, |
| { |
| "epoch": 1.35, |
| "grad_norm": 1.053911566734314, |
| "learning_rate": 8.64795918367347e-05, |
| "loss": 1.7458, |
| "step": 2650 |
| }, |
| { |
| "epoch": 1.38, |
| "grad_norm": 1.1161508560180664, |
| "learning_rate": 8.622448979591838e-05, |
| "loss": 1.7359, |
| "step": 2700 |
| }, |
| { |
| "epoch": 1.4, |
| "grad_norm": 0.9859876036643982, |
| "learning_rate": 8.596938775510205e-05, |
| "loss": 1.7788, |
| "step": 2750 |
| }, |
| { |
| "epoch": 1.43, |
| "grad_norm": 1.025814175605774, |
| "learning_rate": 8.571428571428571e-05, |
| "loss": 1.751, |
| "step": 2800 |
| }, |
| { |
| "epoch": 1.45, |
| "grad_norm": 1.0681238174438477, |
| "learning_rate": 8.545918367346939e-05, |
| "loss": 1.7733, |
| "step": 2850 |
| }, |
| { |
| "epoch": 1.48, |
| "grad_norm": 0.9048791527748108, |
| "learning_rate": 8.520408163265306e-05, |
| "loss": 1.7579, |
| "step": 2900 |
| }, |
| { |
| "epoch": 1.51, |
| "grad_norm": 0.9508095979690552, |
| "learning_rate": 8.494897959183674e-05, |
| "loss": 1.7594, |
| "step": 2950 |
| }, |
| { |
| "epoch": 1.53, |
| "grad_norm": 0.98777836561203, |
| "learning_rate": 8.469387755102041e-05, |
| "loss": 1.7589, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1.56, |
| "grad_norm": 0.970637321472168, |
| "learning_rate": 8.443877551020409e-05, |
| "loss": 1.803, |
| "step": 3050 |
| }, |
| { |
| "epoch": 1.58, |
| "grad_norm": 0.8410449624061584, |
| "learning_rate": 8.418367346938776e-05, |
| "loss": 1.7643, |
| "step": 3100 |
| }, |
| { |
| "epoch": 1.61, |
| "grad_norm": 1.006286859512329, |
| "learning_rate": 8.392857142857144e-05, |
| "loss": 1.8038, |
| "step": 3150 |
| }, |
| { |
| "epoch": 1.63, |
| "grad_norm": 0.943855345249176, |
| "learning_rate": 8.367346938775511e-05, |
| "loss": 1.7648, |
| "step": 3200 |
| }, |
| { |
| "epoch": 1.66, |
| "grad_norm": 0.9942700266838074, |
| "learning_rate": 8.341836734693878e-05, |
| "loss": 1.7311, |
| "step": 3250 |
| }, |
| { |
| "epoch": 1.68, |
| "grad_norm": 1.001306414604187, |
| "learning_rate": 8.316326530612245e-05, |
| "loss": 1.7475, |
| "step": 3300 |
| }, |
| { |
| "epoch": 1.71, |
| "grad_norm": 1.1053903102874756, |
| "learning_rate": 8.290816326530613e-05, |
| "loss": 1.7569, |
| "step": 3350 |
| }, |
| { |
| "epoch": 1.73, |
| "grad_norm": 0.8580662608146667, |
| "learning_rate": 8.26530612244898e-05, |
| "loss": 1.7452, |
| "step": 3400 |
| }, |
| { |
| "epoch": 1.76, |
| "grad_norm": 1.0214366912841797, |
| "learning_rate": 8.239795918367348e-05, |
| "loss": 1.7852, |
| "step": 3450 |
| }, |
| { |
| "epoch": 1.79, |
| "grad_norm": 0.9641438126564026, |
| "learning_rate": 8.214285714285714e-05, |
| "loss": 1.7648, |
| "step": 3500 |
| }, |
| { |
| "epoch": 1.81, |
| "grad_norm": 0.934022068977356, |
| "learning_rate": 8.188775510204081e-05, |
| "loss": 1.7573, |
| "step": 3550 |
| }, |
| { |
| "epoch": 1.84, |
| "grad_norm": 0.9156440496444702, |
| "learning_rate": 8.163265306122449e-05, |
| "loss": 1.7707, |
| "step": 3600 |
| }, |
| { |
| "epoch": 1.86, |
| "grad_norm": 0.9485116004943848, |
| "learning_rate": 8.137755102040817e-05, |
| "loss": 1.7513, |
| "step": 3650 |
| }, |
| { |
| "epoch": 1.89, |
| "grad_norm": 1.0946673154830933, |
| "learning_rate": 8.112244897959184e-05, |
| "loss": 1.7277, |
| "step": 3700 |
| }, |
| { |
| "epoch": 1.91, |
| "grad_norm": 0.9547972083091736, |
| "learning_rate": 8.086734693877552e-05, |
| "loss": 1.7371, |
| "step": 3750 |
| }, |
| { |
| "epoch": 1.94, |
| "grad_norm": 0.9103617072105408, |
| "learning_rate": 8.061224489795919e-05, |
| "loss": 1.7575, |
| "step": 3800 |
| }, |
| { |
| "epoch": 1.96, |
| "grad_norm": 1.0051575899124146, |
| "learning_rate": 8.035714285714287e-05, |
| "loss": 1.7626, |
| "step": 3850 |
| }, |
| { |
| "epoch": 1.99, |
| "grad_norm": 1.0175752639770508, |
| "learning_rate": 8.010204081632653e-05, |
| "loss": 1.7599, |
| "step": 3900 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_bert-score-f1": 0.35579173118946883, |
| "eval_bert-score-precision": 0.3807178700166959, |
| "eval_bert-score-recall": 0.330699935046396, |
| "eval_gen_len": 48.06179517273052, |
| "eval_loss": 1.6298102140426636, |
| "eval_rouge1": 0.3794, |
| "eval_rouge2": 0.1597, |
| "eval_rougeL": 0.2844, |
| "eval_rougeLsum": 0.2846, |
| "eval_runtime": 1644.5373, |
| "eval_samples_per_second": 11.916, |
| "eval_steps_per_second": 0.298, |
| "step": 3920 |
| }, |
| { |
| "epoch": 2.02, |
| "grad_norm": 0.99552983045578, |
| "learning_rate": 7.98469387755102e-05, |
| "loss": 1.7165, |
| "step": 3950 |
| }, |
| { |
| "epoch": 2.04, |
| "grad_norm": 1.0231791734695435, |
| "learning_rate": 7.959183673469388e-05, |
| "loss": 1.6766, |
| "step": 4000 |
| }, |
| { |
| "epoch": 2.07, |
| "grad_norm": 0.9459535479545593, |
| "learning_rate": 7.933673469387755e-05, |
| "loss": 1.6492, |
| "step": 4050 |
| }, |
| { |
| "epoch": 2.09, |
| "grad_norm": 0.9600710868835449, |
| "learning_rate": 7.908163265306123e-05, |
| "loss": 1.6922, |
| "step": 4100 |
| }, |
| { |
| "epoch": 2.12, |
| "grad_norm": 1.0735735893249512, |
| "learning_rate": 7.882653061224489e-05, |
| "loss": 1.6794, |
| "step": 4150 |
| }, |
| { |
| "epoch": 2.14, |
| "grad_norm": 1.0312355756759644, |
| "learning_rate": 7.857142857142858e-05, |
| "loss": 1.6807, |
| "step": 4200 |
| }, |
| { |
| "epoch": 2.17, |
| "grad_norm": 1.1497960090637207, |
| "learning_rate": 7.831632653061226e-05, |
| "loss": 1.6969, |
| "step": 4250 |
| }, |
| { |
| "epoch": 2.19, |
| "grad_norm": 0.9301525950431824, |
| "learning_rate": 7.806122448979593e-05, |
| "loss": 1.6831, |
| "step": 4300 |
| }, |
| { |
| "epoch": 2.22, |
| "grad_norm": 0.9076269268989563, |
| "learning_rate": 7.780612244897959e-05, |
| "loss": 1.7112, |
| "step": 4350 |
| }, |
| { |
| "epoch": 2.24, |
| "grad_norm": 0.9961591958999634, |
| "learning_rate": 7.755102040816327e-05, |
| "loss": 1.7087, |
| "step": 4400 |
| }, |
| { |
| "epoch": 2.27, |
| "grad_norm": 0.9585267305374146, |
| "learning_rate": 7.729591836734694e-05, |
| "loss": 1.6773, |
| "step": 4450 |
| }, |
| { |
| "epoch": 2.3, |
| "grad_norm": 0.9868924021720886, |
| "learning_rate": 7.704081632653062e-05, |
| "loss": 1.7028, |
| "step": 4500 |
| }, |
| { |
| "epoch": 2.32, |
| "grad_norm": 1.0738240480422974, |
| "learning_rate": 7.67857142857143e-05, |
| "loss": 1.7072, |
| "step": 4550 |
| }, |
| { |
| "epoch": 2.35, |
| "grad_norm": 0.8686535954475403, |
| "learning_rate": 7.653061224489796e-05, |
| "loss": 1.6531, |
| "step": 4600 |
| }, |
| { |
| "epoch": 2.37, |
| "grad_norm": 1.0066442489624023, |
| "learning_rate": 7.627551020408163e-05, |
| "loss": 1.7061, |
| "step": 4650 |
| }, |
| { |
| "epoch": 2.4, |
| "grad_norm": 0.9233015775680542, |
| "learning_rate": 7.60204081632653e-05, |
| "loss": 1.6787, |
| "step": 4700 |
| }, |
| { |
| "epoch": 2.42, |
| "grad_norm": 1.0817601680755615, |
| "learning_rate": 7.576530612244898e-05, |
| "loss": 1.7002, |
| "step": 4750 |
| }, |
| { |
| "epoch": 2.45, |
| "grad_norm": 1.0920677185058594, |
| "learning_rate": 7.551020408163266e-05, |
| "loss": 1.7126, |
| "step": 4800 |
| }, |
| { |
| "epoch": 2.47, |
| "grad_norm": 1.0090982913970947, |
| "learning_rate": 7.525510204081633e-05, |
| "loss": 1.693, |
| "step": 4850 |
| }, |
| { |
| "epoch": 2.5, |
| "grad_norm": 0.9864045977592468, |
| "learning_rate": 7.500000000000001e-05, |
| "loss": 1.6798, |
| "step": 4900 |
| }, |
| { |
| "epoch": 2.53, |
| "grad_norm": 0.9712895154953003, |
| "learning_rate": 7.474489795918368e-05, |
| "loss": 1.7083, |
| "step": 4950 |
| }, |
| { |
| "epoch": 2.55, |
| "grad_norm": 1.0552610158920288, |
| "learning_rate": 7.448979591836736e-05, |
| "loss": 1.7034, |
| "step": 5000 |
| }, |
| { |
| "epoch": 2.58, |
| "grad_norm": 0.9989820122718811, |
| "learning_rate": 7.423469387755102e-05, |
| "loss": 1.6867, |
| "step": 5050 |
| }, |
| { |
| "epoch": 2.6, |
| "grad_norm": 0.9927219748497009, |
| "learning_rate": 7.39795918367347e-05, |
| "loss": 1.705, |
| "step": 5100 |
| }, |
| { |
| "epoch": 2.63, |
| "grad_norm": 0.9960281252861023, |
| "learning_rate": 7.372448979591837e-05, |
| "loss": 1.7184, |
| "step": 5150 |
| }, |
| { |
| "epoch": 2.65, |
| "grad_norm": 1.005506157875061, |
| "learning_rate": 7.346938775510205e-05, |
| "loss": 1.6787, |
| "step": 5200 |
| }, |
| { |
| "epoch": 2.68, |
| "grad_norm": 1.0306086540222168, |
| "learning_rate": 7.321428571428571e-05, |
| "loss": 1.7111, |
| "step": 5250 |
| }, |
| { |
| "epoch": 2.7, |
| "grad_norm": 0.9831746816635132, |
| "learning_rate": 7.29591836734694e-05, |
| "loss": 1.6914, |
| "step": 5300 |
| }, |
| { |
| "epoch": 2.73, |
| "grad_norm": 0.903492271900177, |
| "learning_rate": 7.270408163265307e-05, |
| "loss": 1.6923, |
| "step": 5350 |
| }, |
| { |
| "epoch": 2.76, |
| "grad_norm": 1.0123188495635986, |
| "learning_rate": 7.244897959183675e-05, |
| "loss": 1.7069, |
| "step": 5400 |
| }, |
| { |
| "epoch": 2.78, |
| "grad_norm": 0.9369603395462036, |
| "learning_rate": 7.219387755102042e-05, |
| "loss": 1.6903, |
| "step": 5450 |
| }, |
| { |
| "epoch": 2.81, |
| "grad_norm": 0.9350219368934631, |
| "learning_rate": 7.193877551020408e-05, |
| "loss": 1.7194, |
| "step": 5500 |
| }, |
| { |
| "epoch": 2.83, |
| "grad_norm": 0.9026333093643188, |
| "learning_rate": 7.168367346938776e-05, |
| "loss": 1.7041, |
| "step": 5550 |
| }, |
| { |
| "epoch": 2.86, |
| "grad_norm": 1.100752592086792, |
| "learning_rate": 7.142857142857143e-05, |
| "loss": 1.7159, |
| "step": 5600 |
| }, |
| { |
| "epoch": 2.88, |
| "grad_norm": 0.9471935629844666, |
| "learning_rate": 7.117346938775511e-05, |
| "loss": 1.6925, |
| "step": 5650 |
| }, |
| { |
| "epoch": 2.91, |
| "grad_norm": 0.9510202407836914, |
| "learning_rate": 7.091836734693877e-05, |
| "loss": 1.7098, |
| "step": 5700 |
| }, |
| { |
| "epoch": 2.93, |
| "grad_norm": 0.9223019480705261, |
| "learning_rate": 7.066326530612245e-05, |
| "loss": 1.6925, |
| "step": 5750 |
| }, |
| { |
| "epoch": 2.96, |
| "grad_norm": 0.9820858836174011, |
| "learning_rate": 7.040816326530612e-05, |
| "loss": 1.6947, |
| "step": 5800 |
| }, |
| { |
| "epoch": 2.98, |
| "grad_norm": 0.9341661930084229, |
| "learning_rate": 7.01530612244898e-05, |
| "loss": 1.6943, |
| "step": 5850 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_bert-score-f1": 0.36010894499395646, |
| "eval_bert-score-precision": 0.3884696647793401, |
| "eval_bert-score-recall": 0.3316552115749222, |
| "eval_gen_len": 47.269377966015206, |
| "eval_loss": 1.6264128684997559, |
| "eval_rouge1": 0.3814, |
| "eval_rouge2": 0.1616, |
| "eval_rougeL": 0.2862, |
| "eval_rougeLsum": 0.2863, |
| "eval_runtime": 1896.2915, |
| "eval_samples_per_second": 10.334, |
| "eval_steps_per_second": 0.258, |
| "step": 5880 |
| }, |
| { |
| "epoch": 3.01, |
| "grad_norm": 1.0649360418319702, |
| "learning_rate": 6.989795918367347e-05, |
| "loss": 1.6813, |
| "step": 5900 |
| }, |
| { |
| "epoch": 3.04, |
| "grad_norm": 0.9178709983825684, |
| "learning_rate": 6.964285714285715e-05, |
| "loss": 1.6533, |
| "step": 5950 |
| }, |
| { |
| "epoch": 3.06, |
| "grad_norm": 1.0868624448776245, |
| "learning_rate": 6.938775510204082e-05, |
| "loss": 1.6367, |
| "step": 6000 |
| }, |
| { |
| "epoch": 3.09, |
| "grad_norm": 1.0165528059005737, |
| "learning_rate": 6.91326530612245e-05, |
| "loss": 1.6236, |
| "step": 6050 |
| }, |
| { |
| "epoch": 3.11, |
| "grad_norm": 1.1030255556106567, |
| "learning_rate": 6.887755102040817e-05, |
| "loss": 1.6123, |
| "step": 6100 |
| }, |
| { |
| "epoch": 3.14, |
| "grad_norm": 1.0415186882019043, |
| "learning_rate": 6.862244897959184e-05, |
| "loss": 1.6487, |
| "step": 6150 |
| }, |
| { |
| "epoch": 3.16, |
| "grad_norm": 1.0924690961837769, |
| "learning_rate": 6.836734693877551e-05, |
| "loss": 1.6591, |
| "step": 6200 |
| }, |
| { |
| "epoch": 3.19, |
| "grad_norm": 0.9578896760940552, |
| "learning_rate": 6.811224489795919e-05, |
| "loss": 1.6191, |
| "step": 6250 |
| }, |
| { |
| "epoch": 3.21, |
| "grad_norm": 0.990861177444458, |
| "learning_rate": 6.785714285714286e-05, |
| "loss": 1.6508, |
| "step": 6300 |
| }, |
| { |
| "epoch": 3.24, |
| "grad_norm": 0.9668008089065552, |
| "learning_rate": 6.760204081632652e-05, |
| "loss": 1.6449, |
| "step": 6350 |
| }, |
| { |
| "epoch": 3.27, |
| "grad_norm": 0.9364696145057678, |
| "learning_rate": 6.73469387755102e-05, |
| "loss": 1.6544, |
| "step": 6400 |
| }, |
| { |
| "epoch": 3.29, |
| "grad_norm": 1.1126731634140015, |
| "learning_rate": 6.709183673469389e-05, |
| "loss": 1.6263, |
| "step": 6450 |
| }, |
| { |
| "epoch": 3.32, |
| "grad_norm": 1.3058950901031494, |
| "learning_rate": 6.683673469387756e-05, |
| "loss": 1.6623, |
| "step": 6500 |
| }, |
| { |
| "epoch": 3.34, |
| "grad_norm": 1.0124415159225464, |
| "learning_rate": 6.658163265306124e-05, |
| "loss": 1.6561, |
| "step": 6550 |
| }, |
| { |
| "epoch": 3.37, |
| "grad_norm": 0.998085618019104, |
| "learning_rate": 6.63265306122449e-05, |
| "loss": 1.6352, |
| "step": 6600 |
| }, |
| { |
| "epoch": 3.39, |
| "grad_norm": 0.9867570996284485, |
| "learning_rate": 6.607142857142857e-05, |
| "loss": 1.6399, |
| "step": 6650 |
| }, |
| { |
| "epoch": 3.42, |
| "grad_norm": 1.0731745958328247, |
| "learning_rate": 6.581632653061225e-05, |
| "loss": 1.6277, |
| "step": 6700 |
| }, |
| { |
| "epoch": 3.44, |
| "grad_norm": 0.9528840780258179, |
| "learning_rate": 6.556122448979592e-05, |
| "loss": 1.6435, |
| "step": 6750 |
| }, |
| { |
| "epoch": 3.47, |
| "grad_norm": 0.9235923886299133, |
| "learning_rate": 6.530612244897959e-05, |
| "loss": 1.6615, |
| "step": 6800 |
| }, |
| { |
| "epoch": 3.49, |
| "grad_norm": 0.9793564081192017, |
| "learning_rate": 6.505102040816326e-05, |
| "loss": 1.6573, |
| "step": 6850 |
| }, |
| { |
| "epoch": 3.52, |
| "grad_norm": 1.0197653770446777, |
| "learning_rate": 6.479591836734694e-05, |
| "loss": 1.6369, |
| "step": 6900 |
| }, |
| { |
| "epoch": 3.55, |
| "grad_norm": 1.0002648830413818, |
| "learning_rate": 6.454081632653061e-05, |
| "loss": 1.6422, |
| "step": 6950 |
| }, |
| { |
| "epoch": 3.57, |
| "grad_norm": 0.9143182039260864, |
| "learning_rate": 6.428571428571429e-05, |
| "loss": 1.6289, |
| "step": 7000 |
| }, |
| { |
| "epoch": 3.6, |
| "grad_norm": 1.0049608945846558, |
| "learning_rate": 6.403061224489796e-05, |
| "loss": 1.6504, |
| "step": 7050 |
| }, |
| { |
| "epoch": 3.62, |
| "grad_norm": 0.9720739722251892, |
| "learning_rate": 6.377551020408164e-05, |
| "loss": 1.626, |
| "step": 7100 |
| }, |
| { |
| "epoch": 3.65, |
| "grad_norm": 0.9640092849731445, |
| "learning_rate": 6.352040816326531e-05, |
| "loss": 1.6639, |
| "step": 7150 |
| }, |
| { |
| "epoch": 3.67, |
| "grad_norm": 1.0053060054779053, |
| "learning_rate": 6.326530612244899e-05, |
| "loss": 1.6393, |
| "step": 7200 |
| }, |
| { |
| "epoch": 3.7, |
| "grad_norm": 0.8709841370582581, |
| "learning_rate": 6.301020408163265e-05, |
| "loss": 1.6602, |
| "step": 7250 |
| }, |
| { |
| "epoch": 3.72, |
| "grad_norm": 0.980208158493042, |
| "learning_rate": 6.275510204081633e-05, |
| "loss": 1.6572, |
| "step": 7300 |
| }, |
| { |
| "epoch": 3.75, |
| "grad_norm": 0.9883060455322266, |
| "learning_rate": 6.25e-05, |
| "loss": 1.6554, |
| "step": 7350 |
| }, |
| { |
| "epoch": 3.78, |
| "grad_norm": 1.1915491819381714, |
| "learning_rate": 6.224489795918368e-05, |
| "loss": 1.6539, |
| "step": 7400 |
| }, |
| { |
| "epoch": 3.8, |
| "grad_norm": 0.9820300340652466, |
| "learning_rate": 6.198979591836735e-05, |
| "loss": 1.6274, |
| "step": 7450 |
| }, |
| { |
| "epoch": 3.83, |
| "grad_norm": 0.9642151594161987, |
| "learning_rate": 6.173469387755101e-05, |
| "loss": 1.635, |
| "step": 7500 |
| }, |
| { |
| "epoch": 3.85, |
| "grad_norm": 1.0585176944732666, |
| "learning_rate": 6.14795918367347e-05, |
| "loss": 1.6283, |
| "step": 7550 |
| }, |
| { |
| "epoch": 3.88, |
| "grad_norm": 1.110522747039795, |
| "learning_rate": 6.122448979591838e-05, |
| "loss": 1.6573, |
| "step": 7600 |
| }, |
| { |
| "epoch": 3.9, |
| "grad_norm": 1.215560793876648, |
| "learning_rate": 6.0969387755102046e-05, |
| "loss": 1.6598, |
| "step": 7650 |
| }, |
| { |
| "epoch": 3.93, |
| "grad_norm": 1.0057196617126465, |
| "learning_rate": 6.0714285714285715e-05, |
| "loss": 1.6595, |
| "step": 7700 |
| }, |
| { |
| "epoch": 3.95, |
| "grad_norm": 0.9869462847709656, |
| "learning_rate": 6.045918367346939e-05, |
| "loss": 1.6336, |
| "step": 7750 |
| }, |
| { |
| "epoch": 3.98, |
| "grad_norm": 1.0604314804077148, |
| "learning_rate": 6.0204081632653065e-05, |
| "loss": 1.6601, |
| "step": 7800 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_bert-score-f1": 0.36134206724120493, |
| "eval_bert-score-precision": 0.3865282032135588, |
| "eval_bert-score-recall": 0.33597884116813037, |
| "eval_gen_len": 47.81777823136194, |
| "eval_loss": 1.6249974966049194, |
| "eval_rouge1": 0.3835, |
| "eval_rouge2": 0.162, |
| "eval_rougeL": 0.2878, |
| "eval_rougeLsum": 0.2879, |
| "eval_runtime": 3018.7419, |
| "eval_samples_per_second": 6.492, |
| "eval_steps_per_second": 0.162, |
| "step": 7840 |
| }, |
| { |
| "epoch": 4.01, |
| "grad_norm": 0.9901429414749146, |
| "learning_rate": 5.994897959183674e-05, |
| "loss": 1.6288, |
| "step": 7850 |
| }, |
| { |
| "epoch": 4.03, |
| "grad_norm": 1.0381147861480713, |
| "learning_rate": 5.9693877551020416e-05, |
| "loss": 1.5792, |
| "step": 7900 |
| }, |
| { |
| "epoch": 4.06, |
| "grad_norm": 1.4534790515899658, |
| "learning_rate": 5.9438775510204084e-05, |
| "loss": 1.5935, |
| "step": 7950 |
| }, |
| { |
| "epoch": 4.08, |
| "grad_norm": 1.0487029552459717, |
| "learning_rate": 5.918367346938776e-05, |
| "loss": 1.6284, |
| "step": 8000 |
| }, |
| { |
| "epoch": 4.11, |
| "grad_norm": 1.150761604309082, |
| "learning_rate": 5.8928571428571435e-05, |
| "loss": 1.5871, |
| "step": 8050 |
| }, |
| { |
| "epoch": 4.13, |
| "grad_norm": 1.0310755968093872, |
| "learning_rate": 5.867346938775511e-05, |
| "loss": 1.5819, |
| "step": 8100 |
| }, |
| { |
| "epoch": 4.16, |
| "grad_norm": 1.0441168546676636, |
| "learning_rate": 5.841836734693877e-05, |
| "loss": 1.6167, |
| "step": 8150 |
| }, |
| { |
| "epoch": 4.18, |
| "grad_norm": 1.0517609119415283, |
| "learning_rate": 5.816326530612245e-05, |
| "loss": 1.6117, |
| "step": 8200 |
| }, |
| { |
| "epoch": 4.21, |
| "grad_norm": 1.164240837097168, |
| "learning_rate": 5.790816326530612e-05, |
| "loss": 1.6033, |
| "step": 8250 |
| }, |
| { |
| "epoch": 4.23, |
| "grad_norm": 1.07903254032135, |
| "learning_rate": 5.7653061224489805e-05, |
| "loss": 1.5992, |
| "step": 8300 |
| }, |
| { |
| "epoch": 4.26, |
| "grad_norm": 0.9475494027137756, |
| "learning_rate": 5.739795918367348e-05, |
| "loss": 1.579, |
| "step": 8350 |
| }, |
| { |
| "epoch": 4.29, |
| "grad_norm": 0.9718815684318542, |
| "learning_rate": 5.714285714285714e-05, |
| "loss": 1.6266, |
| "step": 8400 |
| }, |
| { |
| "epoch": 4.31, |
| "grad_norm": 0.9643973708152771, |
| "learning_rate": 5.688775510204082e-05, |
| "loss": 1.5926, |
| "step": 8450 |
| }, |
| { |
| "epoch": 4.34, |
| "grad_norm": 1.127752423286438, |
| "learning_rate": 5.663265306122449e-05, |
| "loss": 1.5743, |
| "step": 8500 |
| }, |
| { |
| "epoch": 4.36, |
| "grad_norm": 1.1558603048324585, |
| "learning_rate": 5.637755102040817e-05, |
| "loss": 1.6025, |
| "step": 8550 |
| }, |
| { |
| "epoch": 4.39, |
| "grad_norm": 0.9887247681617737, |
| "learning_rate": 5.6122448979591836e-05, |
| "loss": 1.609, |
| "step": 8600 |
| }, |
| { |
| "epoch": 4.41, |
| "grad_norm": 1.2572548389434814, |
| "learning_rate": 5.586734693877551e-05, |
| "loss": 1.6042, |
| "step": 8650 |
| }, |
| { |
| "epoch": 4.44, |
| "grad_norm": 1.0265618562698364, |
| "learning_rate": 5.561224489795919e-05, |
| "loss": 1.6211, |
| "step": 8700 |
| }, |
| { |
| "epoch": 4.46, |
| "grad_norm": 1.0404796600341797, |
| "learning_rate": 5.535714285714286e-05, |
| "loss": 1.6131, |
| "step": 8750 |
| }, |
| { |
| "epoch": 4.49, |
| "grad_norm": 1.1078702211380005, |
| "learning_rate": 5.510204081632653e-05, |
| "loss": 1.6167, |
| "step": 8800 |
| }, |
| { |
| "epoch": 4.52, |
| "grad_norm": 1.1032508611679077, |
| "learning_rate": 5.4846938775510206e-05, |
| "loss": 1.6117, |
| "step": 8850 |
| }, |
| { |
| "epoch": 4.54, |
| "grad_norm": 0.918499231338501, |
| "learning_rate": 5.459183673469388e-05, |
| "loss": 1.5912, |
| "step": 8900 |
| }, |
| { |
| "epoch": 4.57, |
| "grad_norm": 1.0469586849212646, |
| "learning_rate": 5.4336734693877556e-05, |
| "loss": 1.6056, |
| "step": 8950 |
| }, |
| { |
| "epoch": 4.59, |
| "grad_norm": 0.9040560126304626, |
| "learning_rate": 5.408163265306123e-05, |
| "loss": 1.6161, |
| "step": 9000 |
| }, |
| { |
| "epoch": 4.62, |
| "grad_norm": 0.8977251052856445, |
| "learning_rate": 5.382653061224489e-05, |
| "loss": 1.6162, |
| "step": 9050 |
| }, |
| { |
| "epoch": 4.64, |
| "grad_norm": 1.089157223701477, |
| "learning_rate": 5.3571428571428575e-05, |
| "loss": 1.6313, |
| "step": 9100 |
| }, |
| { |
| "epoch": 4.67, |
| "grad_norm": 1.0531835556030273, |
| "learning_rate": 5.331632653061225e-05, |
| "loss": 1.6137, |
| "step": 9150 |
| }, |
| { |
| "epoch": 4.69, |
| "grad_norm": 0.9584686756134033, |
| "learning_rate": 5.3061224489795926e-05, |
| "loss": 1.6048, |
| "step": 9200 |
| }, |
| { |
| "epoch": 4.72, |
| "grad_norm": 0.9945049285888672, |
| "learning_rate": 5.280612244897959e-05, |
| "loss": 1.6566, |
| "step": 9250 |
| }, |
| { |
| "epoch": 4.74, |
| "grad_norm": 0.988216757774353, |
| "learning_rate": 5.255102040816326e-05, |
| "loss": 1.5985, |
| "step": 9300 |
| }, |
| { |
| "epoch": 4.77, |
| "grad_norm": 0.9579148888587952, |
| "learning_rate": 5.229591836734694e-05, |
| "loss": 1.605, |
| "step": 9350 |
| }, |
| { |
| "epoch": 4.8, |
| "grad_norm": 1.0835684537887573, |
| "learning_rate": 5.2040816326530614e-05, |
| "loss": 1.5772, |
| "step": 9400 |
| }, |
| { |
| "epoch": 4.82, |
| "grad_norm": 1.0432782173156738, |
| "learning_rate": 5.1785714285714296e-05, |
| "loss": 1.6155, |
| "step": 9450 |
| }, |
| { |
| "epoch": 4.85, |
| "grad_norm": 1.102050542831421, |
| "learning_rate": 5.153061224489796e-05, |
| "loss": 1.5985, |
| "step": 9500 |
| }, |
| { |
| "epoch": 4.87, |
| "grad_norm": 1.0847904682159424, |
| "learning_rate": 5.127551020408163e-05, |
| "loss": 1.5924, |
| "step": 9550 |
| }, |
| { |
| "epoch": 4.9, |
| "grad_norm": 0.9382128119468689, |
| "learning_rate": 5.102040816326531e-05, |
| "loss": 1.6218, |
| "step": 9600 |
| }, |
| { |
| "epoch": 4.92, |
| "grad_norm": 1.0820624828338623, |
| "learning_rate": 5.076530612244898e-05, |
| "loss": 1.6115, |
| "step": 9650 |
| }, |
| { |
| "epoch": 4.95, |
| "grad_norm": 1.0046682357788086, |
| "learning_rate": 5.051020408163265e-05, |
| "loss": 1.5861, |
| "step": 9700 |
| }, |
| { |
| "epoch": 4.97, |
| "grad_norm": 1.0074535608291626, |
| "learning_rate": 5.025510204081633e-05, |
| "loss": 1.5962, |
| "step": 9750 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 1.1204075813293457, |
| "learning_rate": 5e-05, |
| "loss": 1.6076, |
| "step": 9800 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_bert-score-f1": 0.36255005643080956, |
| "eval_bert-score-precision": 0.3871743720172369, |
| "eval_bert-score-recall": 0.33773628318397175, |
| "eval_gen_len": 47.81287952237587, |
| "eval_loss": 1.6293747425079346, |
| "eval_rouge1": 0.384, |
| "eval_rouge2": 0.1631, |
| "eval_rougeL": 0.2883, |
| "eval_rougeLsum": 0.2884, |
| "eval_runtime": 2826.4897, |
| "eval_samples_per_second": 6.933, |
| "eval_steps_per_second": 0.173, |
| "step": 9800 |
| }, |
| { |
| "epoch": 5.03, |
| "grad_norm": 0.9807945489883423, |
| "learning_rate": 4.974489795918368e-05, |
| "loss": 1.5527, |
| "step": 9850 |
| }, |
| { |
| "epoch": 5.05, |
| "grad_norm": 0.9892862439155579, |
| "learning_rate": 4.9489795918367346e-05, |
| "loss": 1.5641, |
| "step": 9900 |
| }, |
| { |
| "epoch": 5.08, |
| "grad_norm": 1.0735799074172974, |
| "learning_rate": 4.923469387755102e-05, |
| "loss": 1.5635, |
| "step": 9950 |
| }, |
| { |
| "epoch": 5.1, |
| "grad_norm": 1.0304920673370361, |
| "learning_rate": 4.89795918367347e-05, |
| "loss": 1.5696, |
| "step": 10000 |
| }, |
| { |
| "epoch": 5.13, |
| "grad_norm": 1.0004465579986572, |
| "learning_rate": 4.872448979591837e-05, |
| "loss": 1.5715, |
| "step": 10050 |
| }, |
| { |
| "epoch": 5.15, |
| "grad_norm": 0.9926764965057373, |
| "learning_rate": 4.846938775510204e-05, |
| "loss": 1.5843, |
| "step": 10100 |
| }, |
| { |
| "epoch": 5.18, |
| "grad_norm": 1.0199840068817139, |
| "learning_rate": 4.8214285714285716e-05, |
| "loss": 1.5528, |
| "step": 10150 |
| }, |
| { |
| "epoch": 5.2, |
| "grad_norm": 1.0013351440429688, |
| "learning_rate": 4.795918367346939e-05, |
| "loss": 1.5418, |
| "step": 10200 |
| }, |
| { |
| "epoch": 5.23, |
| "grad_norm": 1.006237506866455, |
| "learning_rate": 4.7704081632653066e-05, |
| "loss": 1.5607, |
| "step": 10250 |
| }, |
| { |
| "epoch": 5.26, |
| "grad_norm": 1.1133328676223755, |
| "learning_rate": 4.744897959183674e-05, |
| "loss": 1.5704, |
| "step": 10300 |
| }, |
| { |
| "epoch": 5.28, |
| "grad_norm": 0.9636327624320984, |
| "learning_rate": 4.719387755102041e-05, |
| "loss": 1.5708, |
| "step": 10350 |
| }, |
| { |
| "epoch": 5.31, |
| "grad_norm": 1.0857808589935303, |
| "learning_rate": 4.6938775510204086e-05, |
| "loss": 1.5905, |
| "step": 10400 |
| }, |
| { |
| "epoch": 5.33, |
| "grad_norm": 0.9688696265220642, |
| "learning_rate": 4.6683673469387754e-05, |
| "loss": 1.5827, |
| "step": 10450 |
| }, |
| { |
| "epoch": 5.36, |
| "grad_norm": 0.9749208092689514, |
| "learning_rate": 4.642857142857143e-05, |
| "loss": 1.5496, |
| "step": 10500 |
| }, |
| { |
| "epoch": 5.38, |
| "grad_norm": 0.9618894457817078, |
| "learning_rate": 4.6173469387755105e-05, |
| "loss": 1.5654, |
| "step": 10550 |
| }, |
| { |
| "epoch": 5.41, |
| "grad_norm": 1.057608723640442, |
| "learning_rate": 4.591836734693878e-05, |
| "loss": 1.5476, |
| "step": 10600 |
| }, |
| { |
| "epoch": 5.43, |
| "grad_norm": 0.9891800880432129, |
| "learning_rate": 4.566326530612245e-05, |
| "loss": 1.5424, |
| "step": 10650 |
| }, |
| { |
| "epoch": 5.46, |
| "grad_norm": 1.1025562286376953, |
| "learning_rate": 4.5408163265306124e-05, |
| "loss": 1.5891, |
| "step": 10700 |
| }, |
| { |
| "epoch": 5.48, |
| "grad_norm": 1.0106899738311768, |
| "learning_rate": 4.515306122448979e-05, |
| "loss": 1.5626, |
| "step": 10750 |
| }, |
| { |
| "epoch": 5.51, |
| "grad_norm": 1.0138410329818726, |
| "learning_rate": 4.4897959183673474e-05, |
| "loss": 1.5939, |
| "step": 10800 |
| }, |
| { |
| "epoch": 5.54, |
| "grad_norm": 1.1968185901641846, |
| "learning_rate": 4.464285714285715e-05, |
| "loss": 1.6027, |
| "step": 10850 |
| }, |
| { |
| "epoch": 5.56, |
| "grad_norm": 1.106550693511963, |
| "learning_rate": 4.438775510204082e-05, |
| "loss": 1.5711, |
| "step": 10900 |
| }, |
| { |
| "epoch": 5.59, |
| "grad_norm": 0.992158830165863, |
| "learning_rate": 4.4132653061224493e-05, |
| "loss": 1.571, |
| "step": 10950 |
| }, |
| { |
| "epoch": 5.61, |
| "grad_norm": 0.9270959496498108, |
| "learning_rate": 4.387755102040816e-05, |
| "loss": 1.5805, |
| "step": 11000 |
| }, |
| { |
| "epoch": 5.64, |
| "grad_norm": 0.9585280418395996, |
| "learning_rate": 4.362244897959184e-05, |
| "loss": 1.5634, |
| "step": 11050 |
| }, |
| { |
| "epoch": 5.66, |
| "grad_norm": 1.068963646888733, |
| "learning_rate": 4.336734693877551e-05, |
| "loss": 1.5866, |
| "step": 11100 |
| }, |
| { |
| "epoch": 5.69, |
| "grad_norm": 1.0922127962112427, |
| "learning_rate": 4.311224489795919e-05, |
| "loss": 1.57, |
| "step": 11150 |
| }, |
| { |
| "epoch": 5.71, |
| "grad_norm": 1.019897222518921, |
| "learning_rate": 4.2857142857142856e-05, |
| "loss": 1.5698, |
| "step": 11200 |
| }, |
| { |
| "epoch": 5.74, |
| "grad_norm": 0.9871524572372437, |
| "learning_rate": 4.260204081632653e-05, |
| "loss": 1.5739, |
| "step": 11250 |
| }, |
| { |
| "epoch": 5.77, |
| "grad_norm": 1.0337634086608887, |
| "learning_rate": 4.234693877551021e-05, |
| "loss": 1.5717, |
| "step": 11300 |
| }, |
| { |
| "epoch": 5.79, |
| "grad_norm": 1.0199995040893555, |
| "learning_rate": 4.209183673469388e-05, |
| "loss": 1.5941, |
| "step": 11350 |
| }, |
| { |
| "epoch": 5.82, |
| "grad_norm": 1.0667041540145874, |
| "learning_rate": 4.183673469387756e-05, |
| "loss": 1.5694, |
| "step": 11400 |
| }, |
| { |
| "epoch": 5.84, |
| "grad_norm": 1.03786301612854, |
| "learning_rate": 4.1581632653061226e-05, |
| "loss": 1.573, |
| "step": 11450 |
| }, |
| { |
| "epoch": 5.87, |
| "grad_norm": 1.1216983795166016, |
| "learning_rate": 4.13265306122449e-05, |
| "loss": 1.5773, |
| "step": 11500 |
| }, |
| { |
| "epoch": 5.89, |
| "grad_norm": 1.034957766532898, |
| "learning_rate": 4.107142857142857e-05, |
| "loss": 1.5731, |
| "step": 11550 |
| }, |
| { |
| "epoch": 5.92, |
| "grad_norm": 1.0856486558914185, |
| "learning_rate": 4.0816326530612245e-05, |
| "loss": 1.5673, |
| "step": 11600 |
| }, |
| { |
| "epoch": 5.94, |
| "grad_norm": 1.1110382080078125, |
| "learning_rate": 4.056122448979592e-05, |
| "loss": 1.5593, |
| "step": 11650 |
| }, |
| { |
| "epoch": 5.97, |
| "grad_norm": 0.9986453652381897, |
| "learning_rate": 4.0306122448979596e-05, |
| "loss": 1.5712, |
| "step": 11700 |
| }, |
| { |
| "epoch": 5.99, |
| "grad_norm": 1.0996954441070557, |
| "learning_rate": 4.0051020408163264e-05, |
| "loss": 1.5619, |
| "step": 11750 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_bert-score-f1": 0.3648191453890135, |
| "eval_bert-score-precision": 0.3915913428781525, |
| "eval_bert-score-recall": 0.33787645117386983, |
| "eval_gen_len": 47.19681583915906, |
| "eval_loss": 1.6308201551437378, |
| "eval_rouge1": 0.3849, |
| "eval_rouge2": 0.1637, |
| "eval_rougeL": 0.2894, |
| "eval_rougeLsum": 0.2896, |
| "eval_runtime": 2300.6436, |
| "eval_samples_per_second": 8.518, |
| "eval_steps_per_second": 0.213, |
| "step": 11760 |
| }, |
| { |
| "epoch": 6.02, |
| "grad_norm": 1.1412689685821533, |
| "learning_rate": 3.979591836734694e-05, |
| "loss": 1.5218, |
| "step": 11800 |
| }, |
| { |
| "epoch": 6.05, |
| "grad_norm": 1.2123606204986572, |
| "learning_rate": 3.9540816326530615e-05, |
| "loss": 1.562, |
| "step": 11850 |
| }, |
| { |
| "epoch": 6.07, |
| "grad_norm": 1.1469295024871826, |
| "learning_rate": 3.928571428571429e-05, |
| "loss": 1.5137, |
| "step": 11900 |
| }, |
| { |
| "epoch": 6.1, |
| "grad_norm": 0.9149675965309143, |
| "learning_rate": 3.9030612244897965e-05, |
| "loss": 1.5262, |
| "step": 11950 |
| }, |
| { |
| "epoch": 6.12, |
| "grad_norm": 1.1064791679382324, |
| "learning_rate": 3.8775510204081634e-05, |
| "loss": 1.5294, |
| "step": 12000 |
| }, |
| { |
| "epoch": 6.15, |
| "grad_norm": 1.293892741203308, |
| "learning_rate": 3.852040816326531e-05, |
| "loss": 1.5476, |
| "step": 12050 |
| }, |
| { |
| "epoch": 6.17, |
| "grad_norm": 0.9802124500274658, |
| "learning_rate": 3.826530612244898e-05, |
| "loss": 1.5462, |
| "step": 12100 |
| }, |
| { |
| "epoch": 6.2, |
| "grad_norm": 1.1105841398239136, |
| "learning_rate": 3.801020408163265e-05, |
| "loss": 1.5462, |
| "step": 12150 |
| }, |
| { |
| "epoch": 6.22, |
| "grad_norm": 1.0522410869598389, |
| "learning_rate": 3.775510204081633e-05, |
| "loss": 1.5343, |
| "step": 12200 |
| }, |
| { |
| "epoch": 6.25, |
| "grad_norm": 1.1229223012924194, |
| "learning_rate": 3.7500000000000003e-05, |
| "loss": 1.5449, |
| "step": 12250 |
| }, |
| { |
| "epoch": 6.28, |
| "grad_norm": 1.1412584781646729, |
| "learning_rate": 3.724489795918368e-05, |
| "loss": 1.5243, |
| "step": 12300 |
| }, |
| { |
| "epoch": 6.3, |
| "grad_norm": 1.090889811515808, |
| "learning_rate": 3.698979591836735e-05, |
| "loss": 1.5372, |
| "step": 12350 |
| }, |
| { |
| "epoch": 6.33, |
| "grad_norm": 0.9657461643218994, |
| "learning_rate": 3.673469387755102e-05, |
| "loss": 1.5438, |
| "step": 12400 |
| }, |
| { |
| "epoch": 6.35, |
| "grad_norm": 1.086765170097351, |
| "learning_rate": 3.64795918367347e-05, |
| "loss": 1.5347, |
| "step": 12450 |
| }, |
| { |
| "epoch": 6.38, |
| "grad_norm": 1.4515738487243652, |
| "learning_rate": 3.622448979591837e-05, |
| "loss": 1.553, |
| "step": 12500 |
| }, |
| { |
| "epoch": 6.4, |
| "grad_norm": 0.9871132969856262, |
| "learning_rate": 3.596938775510204e-05, |
| "loss": 1.544, |
| "step": 12550 |
| }, |
| { |
| "epoch": 6.43, |
| "grad_norm": 1.0419111251831055, |
| "learning_rate": 3.571428571428572e-05, |
| "loss": 1.5473, |
| "step": 12600 |
| }, |
| { |
| "epoch": 6.45, |
| "grad_norm": 1.0475398302078247, |
| "learning_rate": 3.5459183673469385e-05, |
| "loss": 1.5196, |
| "step": 12650 |
| }, |
| { |
| "epoch": 6.48, |
| "grad_norm": 1.2095977067947388, |
| "learning_rate": 3.520408163265306e-05, |
| "loss": 1.5292, |
| "step": 12700 |
| }, |
| { |
| "epoch": 6.51, |
| "grad_norm": 0.9942277073860168, |
| "learning_rate": 3.4948979591836736e-05, |
| "loss": 1.5325, |
| "step": 12750 |
| }, |
| { |
| "epoch": 6.53, |
| "grad_norm": 1.0773732662200928, |
| "learning_rate": 3.469387755102041e-05, |
| "loss": 1.531, |
| "step": 12800 |
| }, |
| { |
| "epoch": 6.56, |
| "grad_norm": 1.0953007936477661, |
| "learning_rate": 3.443877551020409e-05, |
| "loss": 1.5352, |
| "step": 12850 |
| }, |
| { |
| "epoch": 6.58, |
| "grad_norm": 1.0514061450958252, |
| "learning_rate": 3.4183673469387755e-05, |
| "loss": 1.5504, |
| "step": 12900 |
| }, |
| { |
| "epoch": 6.61, |
| "grad_norm": 1.0488322973251343, |
| "learning_rate": 3.392857142857143e-05, |
| "loss": 1.5336, |
| "step": 12950 |
| }, |
| { |
| "epoch": 6.63, |
| "grad_norm": 1.1011238098144531, |
| "learning_rate": 3.36734693877551e-05, |
| "loss": 1.5728, |
| "step": 13000 |
| }, |
| { |
| "epoch": 6.66, |
| "grad_norm": 1.5754514932632446, |
| "learning_rate": 3.341836734693878e-05, |
| "loss": 1.5464, |
| "step": 13050 |
| }, |
| { |
| "epoch": 6.68, |
| "grad_norm": 1.2174322605133057, |
| "learning_rate": 3.316326530612245e-05, |
| "loss": 1.5269, |
| "step": 13100 |
| }, |
| { |
| "epoch": 6.71, |
| "grad_norm": 0.9907726645469666, |
| "learning_rate": 3.2908163265306125e-05, |
| "loss": 1.5687, |
| "step": 13150 |
| }, |
| { |
| "epoch": 6.73, |
| "grad_norm": 1.0002177953720093, |
| "learning_rate": 3.265306122448979e-05, |
| "loss": 1.5469, |
| "step": 13200 |
| }, |
| { |
| "epoch": 6.76, |
| "grad_norm": 0.966558039188385, |
| "learning_rate": 3.239795918367347e-05, |
| "loss": 1.5481, |
| "step": 13250 |
| }, |
| { |
| "epoch": 6.79, |
| "grad_norm": 1.07396399974823, |
| "learning_rate": 3.2142857142857144e-05, |
| "loss": 1.5619, |
| "step": 13300 |
| }, |
| { |
| "epoch": 6.81, |
| "grad_norm": 1.0261961221694946, |
| "learning_rate": 3.188775510204082e-05, |
| "loss": 1.5315, |
| "step": 13350 |
| }, |
| { |
| "epoch": 6.84, |
| "grad_norm": 0.9166232347488403, |
| "learning_rate": 3.1632653061224494e-05, |
| "loss": 1.5256, |
| "step": 13400 |
| }, |
| { |
| "epoch": 6.86, |
| "grad_norm": 1.0450319051742554, |
| "learning_rate": 3.137755102040816e-05, |
| "loss": 1.5527, |
| "step": 13450 |
| }, |
| { |
| "epoch": 6.89, |
| "grad_norm": 1.129510521888733, |
| "learning_rate": 3.112244897959184e-05, |
| "loss": 1.5481, |
| "step": 13500 |
| }, |
| { |
| "epoch": 6.91, |
| "grad_norm": 0.9677942991256714, |
| "learning_rate": 3.086734693877551e-05, |
| "loss": 1.511, |
| "step": 13550 |
| }, |
| { |
| "epoch": 6.94, |
| "grad_norm": 0.9299940466880798, |
| "learning_rate": 3.061224489795919e-05, |
| "loss": 1.5684, |
| "step": 13600 |
| }, |
| { |
| "epoch": 6.96, |
| "grad_norm": 1.1531319618225098, |
| "learning_rate": 3.0357142857142857e-05, |
| "loss": 1.5662, |
| "step": 13650 |
| }, |
| { |
| "epoch": 6.99, |
| "grad_norm": 1.0405688285827637, |
| "learning_rate": 3.0102040816326533e-05, |
| "loss": 1.5528, |
| "step": 13700 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_bert-score-f1": 0.36617497401220955, |
| "eval_bert-score-precision": 0.39058827233191745, |
| "eval_bert-score-recall": 0.34153906320910943, |
| "eval_gen_len": 47.82650405674338, |
| "eval_loss": 1.6326719522476196, |
| "eval_rouge1": 0.3871, |
| "eval_rouge2": 0.1645, |
| "eval_rougeL": 0.2903, |
| "eval_rougeLsum": 0.2905, |
| "eval_runtime": 1859.9749, |
| "eval_samples_per_second": 10.536, |
| "eval_steps_per_second": 0.263, |
| "step": 13720 |
| }, |
| { |
| "epoch": 7.02, |
| "grad_norm": 1.022812843322754, |
| "learning_rate": 2.9846938775510208e-05, |
| "loss": 1.5406, |
| "step": 13750 |
| }, |
| { |
| "epoch": 7.04, |
| "grad_norm": 1.0894478559494019, |
| "learning_rate": 2.959183673469388e-05, |
| "loss": 1.4711, |
| "step": 13800 |
| }, |
| { |
| "epoch": 7.07, |
| "grad_norm": 1.1041587591171265, |
| "learning_rate": 2.9336734693877555e-05, |
| "loss": 1.5304, |
| "step": 13850 |
| }, |
| { |
| "epoch": 7.09, |
| "grad_norm": 1.1092400550842285, |
| "learning_rate": 2.9081632653061224e-05, |
| "loss": 1.5113, |
| "step": 13900 |
| }, |
| { |
| "epoch": 7.12, |
| "grad_norm": 1.0444506406784058, |
| "learning_rate": 2.8826530612244902e-05, |
| "loss": 1.5124, |
| "step": 13950 |
| }, |
| { |
| "epoch": 7.14, |
| "grad_norm": 1.0849798917770386, |
| "learning_rate": 2.857142857142857e-05, |
| "loss": 1.5037, |
| "step": 14000 |
| }, |
| { |
| "epoch": 7.17, |
| "grad_norm": 1.0416877269744873, |
| "learning_rate": 2.8316326530612246e-05, |
| "loss": 1.5271, |
| "step": 14050 |
| }, |
| { |
| "epoch": 7.19, |
| "grad_norm": 1.0705244541168213, |
| "learning_rate": 2.8061224489795918e-05, |
| "loss": 1.5058, |
| "step": 14100 |
| }, |
| { |
| "epoch": 7.22, |
| "grad_norm": 1.0970416069030762, |
| "learning_rate": 2.7806122448979593e-05, |
| "loss": 1.5411, |
| "step": 14150 |
| }, |
| { |
| "epoch": 7.24, |
| "grad_norm": 1.0542025566101074, |
| "learning_rate": 2.7551020408163265e-05, |
| "loss": 1.5495, |
| "step": 14200 |
| }, |
| { |
| "epoch": 7.27, |
| "grad_norm": 1.071016788482666, |
| "learning_rate": 2.729591836734694e-05, |
| "loss": 1.4893, |
| "step": 14250 |
| }, |
| { |
| "epoch": 7.3, |
| "grad_norm": 1.0210118293762207, |
| "learning_rate": 2.7040816326530616e-05, |
| "loss": 1.5394, |
| "step": 14300 |
| }, |
| { |
| "epoch": 7.32, |
| "grad_norm": 1.005900263786316, |
| "learning_rate": 2.6785714285714288e-05, |
| "loss": 1.5445, |
| "step": 14350 |
| }, |
| { |
| "epoch": 7.35, |
| "grad_norm": 3.122129201889038, |
| "learning_rate": 2.6530612244897963e-05, |
| "loss": 1.5022, |
| "step": 14400 |
| }, |
| { |
| "epoch": 7.37, |
| "grad_norm": 1.248120903968811, |
| "learning_rate": 2.627551020408163e-05, |
| "loss": 1.547, |
| "step": 14450 |
| }, |
| { |
| "epoch": 7.4, |
| "grad_norm": 1.4089281558990479, |
| "learning_rate": 2.6020408163265307e-05, |
| "loss": 1.5169, |
| "step": 14500 |
| }, |
| { |
| "epoch": 7.42, |
| "grad_norm": 1.1626540422439575, |
| "learning_rate": 2.576530612244898e-05, |
| "loss": 1.5259, |
| "step": 14550 |
| }, |
| { |
| "epoch": 7.45, |
| "grad_norm": 1.0578802824020386, |
| "learning_rate": 2.5510204081632654e-05, |
| "loss": 1.541, |
| "step": 14600 |
| }, |
| { |
| "epoch": 7.47, |
| "grad_norm": 1.1344016790390015, |
| "learning_rate": 2.5255102040816326e-05, |
| "loss": 1.5065, |
| "step": 14650 |
| }, |
| { |
| "epoch": 7.5, |
| "grad_norm": 1.2003172636032104, |
| "learning_rate": 2.5e-05, |
| "loss": 1.5055, |
| "step": 14700 |
| }, |
| { |
| "epoch": 7.53, |
| "grad_norm": 1.0988688468933105, |
| "learning_rate": 2.4744897959183673e-05, |
| "loss": 1.525, |
| "step": 14750 |
| }, |
| { |
| "epoch": 7.55, |
| "grad_norm": 0.9494357705116272, |
| "learning_rate": 2.448979591836735e-05, |
| "loss": 1.5059, |
| "step": 14800 |
| }, |
| { |
| "epoch": 7.58, |
| "grad_norm": 1.1251728534698486, |
| "learning_rate": 2.423469387755102e-05, |
| "loss": 1.5289, |
| "step": 14850 |
| }, |
| { |
| "epoch": 7.6, |
| "grad_norm": 0.9671918749809265, |
| "learning_rate": 2.3979591836734696e-05, |
| "loss": 1.5257, |
| "step": 14900 |
| }, |
| { |
| "epoch": 7.63, |
| "grad_norm": 0.9528857469558716, |
| "learning_rate": 2.372448979591837e-05, |
| "loss": 1.5113, |
| "step": 14950 |
| }, |
| { |
| "epoch": 7.65, |
| "grad_norm": 1.1447895765304565, |
| "learning_rate": 2.3469387755102043e-05, |
| "loss": 1.5001, |
| "step": 15000 |
| }, |
| { |
| "epoch": 7.68, |
| "grad_norm": 0.9951090216636658, |
| "learning_rate": 2.3214285714285715e-05, |
| "loss": 1.5116, |
| "step": 15050 |
| }, |
| { |
| "epoch": 7.7, |
| "grad_norm": 1.0298829078674316, |
| "learning_rate": 2.295918367346939e-05, |
| "loss": 1.5212, |
| "step": 15100 |
| }, |
| { |
| "epoch": 7.73, |
| "grad_norm": 1.0963783264160156, |
| "learning_rate": 2.2704081632653062e-05, |
| "loss": 1.4996, |
| "step": 15150 |
| }, |
| { |
| "epoch": 7.76, |
| "grad_norm": 1.2518473863601685, |
| "learning_rate": 2.2448979591836737e-05, |
| "loss": 1.5218, |
| "step": 15200 |
| }, |
| { |
| "epoch": 7.78, |
| "grad_norm": 1.125091791152954, |
| "learning_rate": 2.219387755102041e-05, |
| "loss": 1.5216, |
| "step": 15250 |
| }, |
| { |
| "epoch": 7.81, |
| "grad_norm": 0.9920032024383545, |
| "learning_rate": 2.193877551020408e-05, |
| "loss": 1.4963, |
| "step": 15300 |
| }, |
| { |
| "epoch": 7.83, |
| "grad_norm": 1.0573756694793701, |
| "learning_rate": 2.1683673469387756e-05, |
| "loss": 1.5017, |
| "step": 15350 |
| }, |
| { |
| "epoch": 7.86, |
| "grad_norm": 1.1425116062164307, |
| "learning_rate": 2.1428571428571428e-05, |
| "loss": 1.5403, |
| "step": 15400 |
| }, |
| { |
| "epoch": 7.88, |
| "grad_norm": 1.0933986902236938, |
| "learning_rate": 2.1173469387755103e-05, |
| "loss": 1.5295, |
| "step": 15450 |
| }, |
| { |
| "epoch": 7.91, |
| "grad_norm": 1.0652328729629517, |
| "learning_rate": 2.091836734693878e-05, |
| "loss": 1.5405, |
| "step": 15500 |
| }, |
| { |
| "epoch": 7.93, |
| "grad_norm": 0.996286153793335, |
| "learning_rate": 2.066326530612245e-05, |
| "loss": 1.5202, |
| "step": 15550 |
| }, |
| { |
| "epoch": 7.96, |
| "grad_norm": 1.198167085647583, |
| "learning_rate": 2.0408163265306123e-05, |
| "loss": 1.527, |
| "step": 15600 |
| }, |
| { |
| "epoch": 7.98, |
| "grad_norm": 1.0896735191345215, |
| "learning_rate": 2.0153061224489798e-05, |
| "loss": 1.546, |
| "step": 15650 |
| }, |
| { |
| "epoch": 8.0, |
| "eval_bert-score-f1": 0.36673917509967924, |
| "eval_bert-score-precision": 0.3916437830545447, |
| "eval_bert-score-recall": 0.3416356591297163, |
| "eval_gen_len": 47.75766698984538, |
| "eval_loss": 1.6359426975250244, |
| "eval_rouge1": 0.3871, |
| "eval_rouge2": 0.165, |
| "eval_rougeL": 0.2909, |
| "eval_rougeLsum": 0.291, |
| "eval_runtime": 1692.2224, |
| "eval_samples_per_second": 11.581, |
| "eval_steps_per_second": 0.29, |
| "step": 15680 |
| }, |
| { |
| "epoch": 8.01, |
| "grad_norm": 1.1501215696334839, |
| "learning_rate": 1.989795918367347e-05, |
| "loss": 1.4992, |
| "step": 15700 |
| }, |
| { |
| "epoch": 8.04, |
| "grad_norm": 1.2085614204406738, |
| "learning_rate": 1.9642857142857145e-05, |
| "loss": 1.4951, |
| "step": 15750 |
| }, |
| { |
| "epoch": 8.06, |
| "grad_norm": 1.107185959815979, |
| "learning_rate": 1.9387755102040817e-05, |
| "loss": 1.4915, |
| "step": 15800 |
| }, |
| { |
| "epoch": 8.09, |
| "grad_norm": 1.0195456743240356, |
| "learning_rate": 1.913265306122449e-05, |
| "loss": 1.5137, |
| "step": 15850 |
| }, |
| { |
| "epoch": 8.11, |
| "grad_norm": 1.058617115020752, |
| "learning_rate": 1.8877551020408164e-05, |
| "loss": 1.4909, |
| "step": 15900 |
| }, |
| { |
| "epoch": 8.14, |
| "grad_norm": 0.969308078289032, |
| "learning_rate": 1.862244897959184e-05, |
| "loss": 1.4956, |
| "step": 15950 |
| }, |
| { |
| "epoch": 8.16, |
| "grad_norm": 1.1257976293563843, |
| "learning_rate": 1.836734693877551e-05, |
| "loss": 1.5048, |
| "step": 16000 |
| }, |
| { |
| "epoch": 8.19, |
| "grad_norm": 1.9717227220535278, |
| "learning_rate": 1.8112244897959187e-05, |
| "loss": 1.4974, |
| "step": 16050 |
| }, |
| { |
| "epoch": 8.21, |
| "grad_norm": 1.035287618637085, |
| "learning_rate": 1.785714285714286e-05, |
| "loss": 1.4886, |
| "step": 16100 |
| }, |
| { |
| "epoch": 8.24, |
| "grad_norm": 1.2189005613327026, |
| "learning_rate": 1.760204081632653e-05, |
| "loss": 1.4969, |
| "step": 16150 |
| }, |
| { |
| "epoch": 8.27, |
| "grad_norm": 1.28898024559021, |
| "learning_rate": 1.7346938775510206e-05, |
| "loss": 1.5073, |
| "step": 16200 |
| }, |
| { |
| "epoch": 8.29, |
| "grad_norm": 1.028761386871338, |
| "learning_rate": 1.7091836734693878e-05, |
| "loss": 1.4939, |
| "step": 16250 |
| }, |
| { |
| "epoch": 8.32, |
| "grad_norm": 1.1025140285491943, |
| "learning_rate": 1.683673469387755e-05, |
| "loss": 1.5073, |
| "step": 16300 |
| }, |
| { |
| "epoch": 8.34, |
| "grad_norm": 1.0877070426940918, |
| "learning_rate": 1.6581632653061225e-05, |
| "loss": 1.4869, |
| "step": 16350 |
| }, |
| { |
| "epoch": 8.37, |
| "grad_norm": 1.0531331300735474, |
| "learning_rate": 1.6326530612244897e-05, |
| "loss": 1.5178, |
| "step": 16400 |
| }, |
| { |
| "epoch": 8.39, |
| "grad_norm": 1.103705644607544, |
| "learning_rate": 1.6071428571428572e-05, |
| "loss": 1.4661, |
| "step": 16450 |
| }, |
| { |
| "epoch": 8.42, |
| "grad_norm": 1.182047963142395, |
| "learning_rate": 1.5816326530612247e-05, |
| "loss": 1.4913, |
| "step": 16500 |
| }, |
| { |
| "epoch": 8.44, |
| "grad_norm": 1.153593897819519, |
| "learning_rate": 1.556122448979592e-05, |
| "loss": 1.5186, |
| "step": 16550 |
| }, |
| { |
| "epoch": 8.47, |
| "grad_norm": 0.9880523085594177, |
| "learning_rate": 1.5306122448979594e-05, |
| "loss": 1.493, |
| "step": 16600 |
| }, |
| { |
| "epoch": 8.49, |
| "grad_norm": 1.0156357288360596, |
| "learning_rate": 1.5051020408163266e-05, |
| "loss": 1.4971, |
| "step": 16650 |
| }, |
| { |
| "epoch": 8.52, |
| "grad_norm": 1.075388789176941, |
| "learning_rate": 1.479591836734694e-05, |
| "loss": 1.4923, |
| "step": 16700 |
| }, |
| { |
| "epoch": 8.55, |
| "grad_norm": 1.1611677408218384, |
| "learning_rate": 1.4540816326530612e-05, |
| "loss": 1.5351, |
| "step": 16750 |
| }, |
| { |
| "epoch": 8.57, |
| "grad_norm": 1.0871764421463013, |
| "learning_rate": 1.4285714285714285e-05, |
| "loss": 1.5222, |
| "step": 16800 |
| }, |
| { |
| "epoch": 8.6, |
| "grad_norm": 1.5180578231811523, |
| "learning_rate": 1.4030612244897959e-05, |
| "loss": 1.5226, |
| "step": 16850 |
| }, |
| { |
| "epoch": 8.62, |
| "grad_norm": 1.066369652748108, |
| "learning_rate": 1.3775510204081633e-05, |
| "loss": 1.5006, |
| "step": 16900 |
| }, |
| { |
| "epoch": 8.65, |
| "grad_norm": 1.079680323600769, |
| "learning_rate": 1.3520408163265308e-05, |
| "loss": 1.5242, |
| "step": 16950 |
| }, |
| { |
| "epoch": 8.67, |
| "grad_norm": 1.0913327932357788, |
| "learning_rate": 1.3265306122448982e-05, |
| "loss": 1.4997, |
| "step": 17000 |
| }, |
| { |
| "epoch": 8.7, |
| "grad_norm": 1.0590647459030151, |
| "learning_rate": 1.3010204081632653e-05, |
| "loss": 1.4928, |
| "step": 17050 |
| }, |
| { |
| "epoch": 8.72, |
| "grad_norm": 1.055536150932312, |
| "learning_rate": 1.2755102040816327e-05, |
| "loss": 1.5015, |
| "step": 17100 |
| }, |
| { |
| "epoch": 8.75, |
| "grad_norm": 1.1097902059555054, |
| "learning_rate": 1.25e-05, |
| "loss": 1.5209, |
| "step": 17150 |
| }, |
| { |
| "epoch": 8.78, |
| "grad_norm": 1.232444167137146, |
| "learning_rate": 1.2244897959183674e-05, |
| "loss": 1.4893, |
| "step": 17200 |
| }, |
| { |
| "epoch": 8.8, |
| "grad_norm": 0.9735280275344849, |
| "learning_rate": 1.1989795918367348e-05, |
| "loss": 1.5097, |
| "step": 17250 |
| }, |
| { |
| "epoch": 8.83, |
| "grad_norm": 1.06132972240448, |
| "learning_rate": 1.1734693877551021e-05, |
| "loss": 1.5091, |
| "step": 17300 |
| }, |
| { |
| "epoch": 8.85, |
| "grad_norm": 1.140011191368103, |
| "learning_rate": 1.1479591836734695e-05, |
| "loss": 1.496, |
| "step": 17350 |
| }, |
| { |
| "epoch": 8.88, |
| "grad_norm": 2.5968542098999023, |
| "learning_rate": 1.1224489795918369e-05, |
| "loss": 1.5102, |
| "step": 17400 |
| }, |
| { |
| "epoch": 8.9, |
| "grad_norm": 1.043047547340393, |
| "learning_rate": 1.096938775510204e-05, |
| "loss": 1.4859, |
| "step": 17450 |
| }, |
| { |
| "epoch": 8.93, |
| "grad_norm": 1.0941978693008423, |
| "learning_rate": 1.0714285714285714e-05, |
| "loss": 1.4938, |
| "step": 17500 |
| }, |
| { |
| "epoch": 8.95, |
| "grad_norm": 0.9536131620407104, |
| "learning_rate": 1.045918367346939e-05, |
| "loss": 1.4984, |
| "step": 17550 |
| }, |
| { |
| "epoch": 8.98, |
| "grad_norm": 1.0615918636322021, |
| "learning_rate": 1.0204081632653061e-05, |
| "loss": 1.5129, |
| "step": 17600 |
| }, |
| { |
| "epoch": 9.0, |
| "eval_bert-score-f1": 0.3671480010073071, |
| "eval_bert-score-precision": 0.39056838285564194, |
| "eval_bert-score-recall": 0.3434888773986903, |
| "eval_gen_len": 48.14736949533092, |
| "eval_loss": 1.6392825841903687, |
| "eval_rouge1": 0.3879, |
| "eval_rouge2": 0.1656, |
| "eval_rougeL": 0.2911, |
| "eval_rougeLsum": 0.2912, |
| "eval_runtime": 1616.6267, |
| "eval_samples_per_second": 12.122, |
| "eval_steps_per_second": 0.303, |
| "step": 17640 |
| }, |
| { |
| "epoch": 9.01, |
| "grad_norm": 1.149622917175293, |
| "learning_rate": 9.948979591836735e-06, |
| "loss": 1.5035, |
| "step": 17650 |
| }, |
| { |
| "epoch": 9.03, |
| "grad_norm": 1.4033907651901245, |
| "learning_rate": 9.693877551020408e-06, |
| "loss": 1.4836, |
| "step": 17700 |
| }, |
| { |
| "epoch": 9.06, |
| "grad_norm": 2.5814456939697266, |
| "learning_rate": 9.438775510204082e-06, |
| "loss": 1.47, |
| "step": 17750 |
| }, |
| { |
| "epoch": 9.08, |
| "grad_norm": 1.0983906984329224, |
| "learning_rate": 9.183673469387756e-06, |
| "loss": 1.5105, |
| "step": 17800 |
| }, |
| { |
| "epoch": 9.11, |
| "grad_norm": 1.0510094165802002, |
| "learning_rate": 8.92857142857143e-06, |
| "loss": 1.4632, |
| "step": 17850 |
| }, |
| { |
| "epoch": 9.13, |
| "grad_norm": 1.2236474752426147, |
| "learning_rate": 8.673469387755103e-06, |
| "loss": 1.4844, |
| "step": 17900 |
| }, |
| { |
| "epoch": 9.16, |
| "grad_norm": 1.0287030935287476, |
| "learning_rate": 8.418367346938775e-06, |
| "loss": 1.4837, |
| "step": 17950 |
| }, |
| { |
| "epoch": 9.18, |
| "grad_norm": 1.120025396347046, |
| "learning_rate": 8.163265306122448e-06, |
| "loss": 1.4715, |
| "step": 18000 |
| }, |
| { |
| "epoch": 9.21, |
| "grad_norm": 1.0506755113601685, |
| "learning_rate": 7.908163265306124e-06, |
| "loss": 1.4948, |
| "step": 18050 |
| }, |
| { |
| "epoch": 9.23, |
| "grad_norm": 1.130733847618103, |
| "learning_rate": 7.653061224489797e-06, |
| "loss": 1.5022, |
| "step": 18100 |
| }, |
| { |
| "epoch": 9.26, |
| "grad_norm": 1.2154968976974487, |
| "learning_rate": 7.39795918367347e-06, |
| "loss": 1.4784, |
| "step": 18150 |
| }, |
| { |
| "epoch": 9.29, |
| "grad_norm": 1.0027872323989868, |
| "learning_rate": 7.142857142857143e-06, |
| "loss": 1.5011, |
| "step": 18200 |
| }, |
| { |
| "epoch": 9.31, |
| "grad_norm": 1.062143325805664, |
| "learning_rate": 6.887755102040816e-06, |
| "loss": 1.4815, |
| "step": 18250 |
| }, |
| { |
| "epoch": 9.34, |
| "grad_norm": 1.133151888847351, |
| "learning_rate": 6.632653061224491e-06, |
| "loss": 1.4992, |
| "step": 18300 |
| }, |
| { |
| "epoch": 9.36, |
| "grad_norm": 1.0543354749679565, |
| "learning_rate": 6.3775510204081635e-06, |
| "loss": 1.4862, |
| "step": 18350 |
| }, |
| { |
| "epoch": 9.39, |
| "grad_norm": 1.2052801847457886, |
| "learning_rate": 6.122448979591837e-06, |
| "loss": 1.4969, |
| "step": 18400 |
| }, |
| { |
| "epoch": 9.41, |
| "grad_norm": 1.0532727241516113, |
| "learning_rate": 5.867346938775511e-06, |
| "loss": 1.5116, |
| "step": 18450 |
| }, |
| { |
| "epoch": 9.44, |
| "grad_norm": 1.125794768333435, |
| "learning_rate": 5.612244897959184e-06, |
| "loss": 1.4888, |
| "step": 18500 |
| }, |
| { |
| "epoch": 9.46, |
| "grad_norm": 1.0929290056228638, |
| "learning_rate": 5.357142857142857e-06, |
| "loss": 1.4982, |
| "step": 18550 |
| }, |
| { |
| "epoch": 9.49, |
| "grad_norm": 1.112062931060791, |
| "learning_rate": 5.102040816326531e-06, |
| "loss": 1.4838, |
| "step": 18600 |
| }, |
| { |
| "epoch": 9.52, |
| "grad_norm": 1.066090703010559, |
| "learning_rate": 4.846938775510204e-06, |
| "loss": 1.4872, |
| "step": 18650 |
| }, |
| { |
| "epoch": 9.54, |
| "grad_norm": 1.0639134645462036, |
| "learning_rate": 4.591836734693878e-06, |
| "loss": 1.4915, |
| "step": 18700 |
| }, |
| { |
| "epoch": 9.57, |
| "grad_norm": 2.9859976768493652, |
| "learning_rate": 4.336734693877551e-06, |
| "loss": 1.4783, |
| "step": 18750 |
| }, |
| { |
| "epoch": 9.59, |
| "grad_norm": 1.1283775568008423, |
| "learning_rate": 4.081632653061224e-06, |
| "loss": 1.4755, |
| "step": 18800 |
| }, |
| { |
| "epoch": 9.62, |
| "grad_norm": 0.9808042049407959, |
| "learning_rate": 3.826530612244899e-06, |
| "loss": 1.4688, |
| "step": 18850 |
| }, |
| { |
| "epoch": 9.64, |
| "grad_norm": 0.9905554056167603, |
| "learning_rate": 3.5714285714285714e-06, |
| "loss": 1.4827, |
| "step": 18900 |
| }, |
| { |
| "epoch": 9.67, |
| "grad_norm": 1.1627849340438843, |
| "learning_rate": 3.3163265306122454e-06, |
| "loss": 1.4733, |
| "step": 18950 |
| }, |
| { |
| "epoch": 9.69, |
| "grad_norm": 1.0419009923934937, |
| "learning_rate": 3.0612244897959185e-06, |
| "loss": 1.4899, |
| "step": 19000 |
| }, |
| { |
| "epoch": 9.72, |
| "grad_norm": 1.1076046228408813, |
| "learning_rate": 2.806122448979592e-06, |
| "loss": 1.5123, |
| "step": 19050 |
| }, |
| { |
| "epoch": 9.74, |
| "grad_norm": 1.0322104692459106, |
| "learning_rate": 2.5510204081632653e-06, |
| "loss": 1.4881, |
| "step": 19100 |
| }, |
| { |
| "epoch": 9.77, |
| "grad_norm": 1.100888729095459, |
| "learning_rate": 2.295918367346939e-06, |
| "loss": 1.5016, |
| "step": 19150 |
| }, |
| { |
| "epoch": 9.8, |
| "grad_norm": 1.0705766677856445, |
| "learning_rate": 2.040816326530612e-06, |
| "loss": 1.5047, |
| "step": 19200 |
| }, |
| { |
| "epoch": 9.82, |
| "grad_norm": 1.507298469543457, |
| "learning_rate": 1.7857142857142857e-06, |
| "loss": 1.4783, |
| "step": 19250 |
| }, |
| { |
| "epoch": 9.85, |
| "grad_norm": 1.0733141899108887, |
| "learning_rate": 1.5306122448979593e-06, |
| "loss": 1.4867, |
| "step": 19300 |
| }, |
| { |
| "epoch": 9.87, |
| "grad_norm": 1.0101690292358398, |
| "learning_rate": 1.2755102040816327e-06, |
| "loss": 1.5068, |
| "step": 19350 |
| }, |
| { |
| "epoch": 9.9, |
| "grad_norm": 1.1356680393218994, |
| "learning_rate": 1.020408163265306e-06, |
| "loss": 1.4895, |
| "step": 19400 |
| }, |
| { |
| "epoch": 9.92, |
| "grad_norm": 1.2502034902572632, |
| "learning_rate": 7.653061224489796e-07, |
| "loss": 1.4965, |
| "step": 19450 |
| }, |
| { |
| "epoch": 9.95, |
| "grad_norm": 1.0862705707550049, |
| "learning_rate": 5.10204081632653e-07, |
| "loss": 1.478, |
| "step": 19500 |
| }, |
| { |
| "epoch": 9.97, |
| "grad_norm": 0.9468031525611877, |
| "learning_rate": 2.551020408163265e-07, |
| "loss": 1.4919, |
| "step": 19550 |
| }, |
| { |
| "epoch": 10.0, |
| "grad_norm": 1.5404436588287354, |
| "learning_rate": 0.0, |
| "loss": 1.4758, |
| "step": 19600 |
| }, |
| { |
| "epoch": 10.0, |
| "eval_bert-score-f1": 0.3667195147856337, |
| "eval_bert-score-precision": 0.3912901967504612, |
| "eval_bert-score-recall": 0.34192645261905996, |
| "eval_gen_len": 47.8131346634689, |
| "eval_loss": 1.639068841934204, |
| "eval_rouge1": 0.3872, |
| "eval_rouge2": 0.1651, |
| "eval_rougeL": 0.2909, |
| "eval_rougeLsum": 0.291, |
| "eval_runtime": 1546.1204, |
| "eval_samples_per_second": 12.675, |
| "eval_steps_per_second": 0.317, |
| "step": 19600 |
| } |
| ], |
| "logging_steps": 50, |
| "max_steps": 19600, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 10, |
| "save_steps": 500, |
| "total_flos": 4.773438216732672e+17, |
| "train_batch_size": 40, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
