| { | |
| "best_metric": 0.23823951184749603, | |
| "best_model_checkpoint": "./distilbert-marian-training1/checkpoint-48000", | |
| "epoch": 10.0, | |
| "global_step": 160000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.212500000000001e-07, | |
| "loss": 0.1004, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.2462500000000001e-06, | |
| "loss": 0.1263, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.8712500000000003e-06, | |
| "loss": 0.1256, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.4962500000000005e-06, | |
| "loss": 0.1014, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.12125e-06, | |
| "loss": 0.1124, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.7462500000000003e-06, | |
| "loss": 0.1479, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.37125e-06, | |
| "loss": 0.1206, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.9950000000000005e-06, | |
| "loss": 0.1157, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5.620000000000001e-06, | |
| "loss": 0.1104, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.245000000000001e-06, | |
| "loss": 0.1329, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.870000000000001e-06, | |
| "loss": 0.1493, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.493750000000001e-06, | |
| "loss": 0.1359, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.11875e-06, | |
| "loss": 0.1326, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.74375e-06, | |
| "loss": 0.1488, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.3675e-06, | |
| "loss": 0.1719, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.992500000000001e-06, | |
| "loss": 0.1606, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.9675e-06, | |
| "loss": 0.1633, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.934605263157895e-06, | |
| "loss": 0.1358, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.90171052631579e-06, | |
| "loss": 0.1801, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 9.868815789473684e-06, | |
| "loss": 0.1531, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.835921052631579e-06, | |
| "loss": 0.185, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.803026315789475e-06, | |
| "loss": 0.1619, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 9.77013157894737e-06, | |
| "loss": 0.136, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.737302631578948e-06, | |
| "loss": 0.1712, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 9.704473684210527e-06, | |
| "loss": 0.1482, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.671578947368422e-06, | |
| "loss": 0.1849, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 9.638684210526316e-06, | |
| "loss": 0.1395, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.605789473684211e-06, | |
| "loss": 0.152, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.572894736842107e-06, | |
| "loss": 0.158, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.54e-06, | |
| "loss": 0.1458, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.507105263157895e-06, | |
| "loss": 0.1783, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.474210526315791e-06, | |
| "loss": 0.1681, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_BLEU": 72.24934811777383, | |
| "eval_BLEU-Bigram-Precision": 74.84708063021316, | |
| "eval_BLEU-Trigram-Precision": 68.90927624872579, | |
| "eval_BLEU-Unigram-Precision": 81.34860050890586, | |
| "eval_ROUGE-2": 75.2339725526462, | |
| "eval_ROUGE-L": 89.68424579231484, | |
| "eval_Sacre-Bigram-Precision": 73.79434037465126, | |
| "eval_Sacre-Trigram-Precision": 68.54019434628975, | |
| "eval_Sacre-Unigram-Precision": 80.34613990576295, | |
| "eval_SacreBLEU": 71.54791264346096, | |
| "eval_loss": 0.24808737635612488, | |
| "eval_runtime": 1428.7921, | |
| "eval_samples_per_second": 0.7, | |
| "eval_steps_per_second": 0.7, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.441315789473686e-06, | |
| "loss": 0.1191, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.408486842105265e-06, | |
| "loss": 0.1805, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.375592105263158e-06, | |
| "loss": 0.1555, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 9.342697368421053e-06, | |
| "loss": 0.1624, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 9.309802631578949e-06, | |
| "loss": 0.1512, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.276907894736843e-06, | |
| "loss": 0.159, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.244013157894738e-06, | |
| "loss": 0.133, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 9.211118421052633e-06, | |
| "loss": 0.1277, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.178223684210527e-06, | |
| "loss": 0.1622, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.145394736842107e-06, | |
| "loss": 0.1263, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.112565789473685e-06, | |
| "loss": 0.1463, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.079671052631581e-06, | |
| "loss": 0.1262, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 9.046776315789474e-06, | |
| "loss": 0.1825, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.013881578947368e-06, | |
| "loss": 0.1582, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 8.981052631578948e-06, | |
| "loss": 0.1307, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 8.948157894736843e-06, | |
| "loss": 0.1333, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 8.915263157894739e-06, | |
| "loss": 0.1462, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.882368421052632e-06, | |
| "loss": 0.1527, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.849473684210526e-06, | |
| "loss": 0.1388, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 8.81657894736842e-06, | |
| "loss": 0.1202, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.783684210526317e-06, | |
| "loss": 0.1144, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.750789473684212e-06, | |
| "loss": 0.119, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.717960526315791e-06, | |
| "loss": 0.1077, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.685065789473684e-06, | |
| "loss": 0.1226, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.652171052631579e-06, | |
| "loss": 0.1432, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.619276315789475e-06, | |
| "loss": 0.1415, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 8.58638157894737e-06, | |
| "loss": 0.1462, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 8.553552631578949e-06, | |
| "loss": 0.1162, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 8.520657894736842e-06, | |
| "loss": 0.1478, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.487763157894737e-06, | |
| "loss": 0.1861, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 8.454868421052633e-06, | |
| "loss": 0.1166, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.421973684210527e-06, | |
| "loss": 0.1209, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_BLEU": 77.2930486436819, | |
| "eval_BLEU-Bigram-Precision": 79.7232309353224, | |
| "eval_BLEU-Trigram-Precision": 74.12314040612445, | |
| "eval_BLEU-Unigram-Precision": 85.70024130842792, | |
| "eval_ROUGE-2": 76.44781505694336, | |
| "eval_ROUGE-L": 90.63044702330345, | |
| "eval_Sacre-Bigram-Precision": 78.87249177195031, | |
| "eval_Sacre-Trigram-Precision": 74.09645692617609, | |
| "eval_Sacre-Unigram-Precision": 84.96016892216143, | |
| "eval_SacreBLEU": 76.88465263390154, | |
| "eval_loss": 0.24807412922382355, | |
| "eval_runtime": 1016.4629, | |
| "eval_samples_per_second": 0.984, | |
| "eval_steps_per_second": 0.984, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.389078947368422e-06, | |
| "loss": 0.1208, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.356184210526316e-06, | |
| "loss": 0.1029, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 8.323355263157894e-06, | |
| "loss": 0.1046, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 8.290526315789474e-06, | |
| "loss": 0.1182, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 8.257631578947369e-06, | |
| "loss": 0.1045, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 8.224736842105265e-06, | |
| "loss": 0.0835, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.191842105263158e-06, | |
| "loss": 0.1486, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 8.158947368421052e-06, | |
| "loss": 0.1506, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.126052631578949e-06, | |
| "loss": 0.1248, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.093157894736843e-06, | |
| "loss": 0.0867, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 8.060263157894738e-06, | |
| "loss": 0.0927, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 8.027434210526317e-06, | |
| "loss": 0.1175, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 7.99453947368421e-06, | |
| "loss": 0.1003, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 7.961644736842106e-06, | |
| "loss": 0.1343, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.928750000000001e-06, | |
| "loss": 0.1368, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.89592105263158e-06, | |
| "loss": 0.1264, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.863026315789475e-06, | |
| "loss": 0.1269, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.830131578947368e-06, | |
| "loss": 0.1214, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 7.797236842105264e-06, | |
| "loss": 0.1248, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 7.764342105263159e-06, | |
| "loss": 0.1191, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 7.731513157894738e-06, | |
| "loss": 0.1247, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.698618421052633e-06, | |
| "loss": 0.0996, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 7.665723684210526e-06, | |
| "loss": 0.135, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 7.632828947368422e-06, | |
| "loss": 0.1153, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 7.599934210526317e-06, | |
| "loss": 0.1054, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.567039473684211e-06, | |
| "loss": 0.1013, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.534210526315791e-06, | |
| "loss": 0.1396, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.501315789473685e-06, | |
| "loss": 0.1169, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 7.468421052631579e-06, | |
| "loss": 0.1182, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 7.435526315789474e-06, | |
| "loss": 0.108, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 7.402631578947369e-06, | |
| "loss": 0.1106, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.369802631578949e-06, | |
| "loss": 0.0983, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_BLEU": 76.73720363059847, | |
| "eval_BLEU-Bigram-Precision": 79.20062087698875, | |
| "eval_BLEU-Trigram-Precision": 73.47770154373929, | |
| "eval_BLEU-Unigram-Precision": 85.37318712415988, | |
| "eval_ROUGE-2": 76.72216094294652, | |
| "eval_ROUGE-L": 90.70575387918443, | |
| "eval_Sacre-Bigram-Precision": 78.34241408214585, | |
| "eval_Sacre-Trigram-Precision": 73.58710882765062, | |
| "eval_Sacre-Unigram-Precision": 84.60735963581183, | |
| "eval_SacreBLEU": 76.40271655688608, | |
| "eval_loss": 0.23823951184749603, | |
| "eval_runtime": 1594.3998, | |
| "eval_samples_per_second": 0.627, | |
| "eval_steps_per_second": 0.627, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.3369078947368425e-06, | |
| "loss": 0.1156, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.304078947368421e-06, | |
| "loss": 0.0855, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 7.271184210526317e-06, | |
| "loss": 0.0923, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 7.238289473684211e-06, | |
| "loss": 0.1026, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 7.205394736842106e-06, | |
| "loss": 0.1265, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 7.172500000000001e-06, | |
| "loss": 0.0995, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 7.139605263157895e-06, | |
| "loss": 0.0903, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 7.1067105263157895e-06, | |
| "loss": 0.0929, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 7.073815789473685e-06, | |
| "loss": 0.0773, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 7.0409210526315795e-06, | |
| "loss": 0.0851, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 7.008026315789475e-06, | |
| "loss": 0.1169, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 6.975131578947369e-06, | |
| "loss": 0.1006, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 6.942236842105263e-06, | |
| "loss": 0.0852, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 6.909407894736843e-06, | |
| "loss": 0.0989, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 6.876513157894737e-06, | |
| "loss": 0.0805, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 6.843684210526317e-06, | |
| "loss": 0.1044, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 6.810789473684211e-06, | |
| "loss": 0.0902, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.777894736842105e-06, | |
| "loss": 0.0765, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 6.745000000000001e-06, | |
| "loss": 0.0951, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 6.7121710526315794e-06, | |
| "loss": 0.1134, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 6.679342105263158e-06, | |
| "loss": 0.1051, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 6.646447368421053e-06, | |
| "loss": 0.0955, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 6.613552631578948e-06, | |
| "loss": 0.0964, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 6.580657894736843e-06, | |
| "loss": 0.1046, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 6.547763157894737e-06, | |
| "loss": 0.0741, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 6.514868421052633e-06, | |
| "loss": 0.0695, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 6.481973684210526e-06, | |
| "loss": 0.0885, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 6.449078947368421e-06, | |
| "loss": 0.0889, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 6.416184210526316e-06, | |
| "loss": 0.0942, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 6.383289473684211e-06, | |
| "loss": 0.103, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.350460526315791e-06, | |
| "loss": 0.0768, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 6.317565789473685e-06, | |
| "loss": 0.0773, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_BLEU": 75.32461616904307, | |
| "eval_BLEU-Bigram-Precision": 77.5221070647523, | |
| "eval_BLEU-Trigram-Precision": 72.33640939597315, | |
| "eval_BLEU-Unigram-Precision": 83.58947642615264, | |
| "eval_ROUGE-2": 75.57825134251199, | |
| "eval_ROUGE-L": 88.70065022361639, | |
| "eval_Sacre-Bigram-Precision": 76.6359918200409, | |
| "eval_Sacre-Trigram-Precision": 72.25821116035914, | |
| "eval_Sacre-Unigram-Precision": 82.69016697588125, | |
| "eval_SacreBLEU": 74.94210102448932, | |
| "eval_loss": 0.24377790093421936, | |
| "eval_runtime": 1720.9311, | |
| "eval_samples_per_second": 0.581, | |
| "eval_steps_per_second": 0.581, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 6.284671052631579e-06, | |
| "loss": 0.0805, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 6.251776315789474e-06, | |
| "loss": 0.0701, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 6.218881578947369e-06, | |
| "loss": 0.07, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 6.185986842105263e-06, | |
| "loss": 0.0785, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 6.153092105263159e-06, | |
| "loss": 0.0832, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 6.120197368421053e-06, | |
| "loss": 0.094, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 6.087368421052632e-06, | |
| "loss": 0.074, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 6.054539473684211e-06, | |
| "loss": 0.0915, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 6.021644736842106e-06, | |
| "loss": 0.0904, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 5.988750000000001e-06, | |
| "loss": 0.0586, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 5.955855263157895e-06, | |
| "loss": 0.0754, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 5.92296052631579e-06, | |
| "loss": 0.074, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 5.890131578947369e-06, | |
| "loss": 0.0723, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 5.857236842105264e-06, | |
| "loss": 0.0814, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 5.824342105263159e-06, | |
| "loss": 0.0623, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 5.7914473684210525e-06, | |
| "loss": 0.0859, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 5.758552631578948e-06, | |
| "loss": 0.0713, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 5.7256578947368425e-06, | |
| "loss": 0.0891, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 5.692763157894737e-06, | |
| "loss": 0.0661, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 5.6598684210526324e-06, | |
| "loss": 0.0893, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 5.62703947368421e-06, | |
| "loss": 0.0694, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 5.594144736842106e-06, | |
| "loss": 0.0836, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 5.56125e-06, | |
| "loss": 0.074, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 5.528355263157895e-06, | |
| "loss": 0.0798, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 5.4955263157894745e-06, | |
| "loss": 0.0899, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 5.4626973684210525e-06, | |
| "loss": 0.0717, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 5.429802631578948e-06, | |
| "loss": 0.0754, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 5.3969078947368424e-06, | |
| "loss": 0.0923, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 5.364013157894738e-06, | |
| "loss": 0.0679, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.331118421052632e-06, | |
| "loss": 0.0722, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 5.298223684210527e-06, | |
| "loss": 0.071, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 5.265328947368421e-06, | |
| "loss": 0.0836, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_BLEU": 80.51272222480867, | |
| "eval_BLEU-Bigram-Precision": 82.70449743434953, | |
| "eval_BLEU-Trigram-Precision": 77.53097443911152, | |
| "eval_BLEU-Unigram-Precision": 88.20733156595666, | |
| "eval_ROUGE-2": 77.48308477253742, | |
| "eval_ROUGE-L": 90.67140065243555, | |
| "eval_Sacre-Bigram-Precision": 82.0950723376482, | |
| "eval_Sacre-Trigram-Precision": 77.98611956654085, | |
| "eval_Sacre-Unigram-Precision": 87.60914352987345, | |
| "eval_SacreBLEU": 80.51595106600885, | |
| "eval_loss": 0.24554118514060974, | |
| "eval_runtime": 1564.2824, | |
| "eval_samples_per_second": 0.639, | |
| "eval_steps_per_second": 0.639, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 5.232434210526316e-06, | |
| "loss": 0.0516, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 5.199539473684211e-06, | |
| "loss": 0.0681, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 5.1667763157894745e-06, | |
| "loss": 0.066, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 5.133881578947368e-06, | |
| "loss": 0.0517, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 5.100986842105264e-06, | |
| "loss": 0.0446, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 5.068092105263158e-06, | |
| "loss": 0.0642, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 5.035263157894738e-06, | |
| "loss": 0.0708, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 5.002368421052632e-06, | |
| "loss": 0.0607, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 4.969473684210527e-06, | |
| "loss": 0.062, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 4.9365789473684215e-06, | |
| "loss": 0.0844, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 4.90375e-06, | |
| "loss": 0.0611, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 4.870855263157896e-06, | |
| "loss": 0.0702, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 4.837960526315789e-06, | |
| "loss": 0.048, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 4.805065789473685e-06, | |
| "loss": 0.0725, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 4.772171052631579e-06, | |
| "loss": 0.0492, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 4.739276315789474e-06, | |
| "loss": 0.0443, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 4.7063815789473685e-06, | |
| "loss": 0.074, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 4.673486842105264e-06, | |
| "loss": 0.0742, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 4.6405921052631585e-06, | |
| "loss": 0.0579, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 4.6078289473684215e-06, | |
| "loss": 0.0673, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 4.574934210526316e-06, | |
| "loss": 0.0685, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 4.5420394736842114e-06, | |
| "loss": 0.0708, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 4.509144736842105e-06, | |
| "loss": 0.076, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 4.476315789473685e-06, | |
| "loss": 0.0717, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 4.443421052631579e-06, | |
| "loss": 0.0781, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 4.410526315789474e-06, | |
| "loss": 0.0678, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 4.377631578947369e-06, | |
| "loss": 0.0584, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 4.344736842105263e-06, | |
| "loss": 0.0736, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 4.3118421052631584e-06, | |
| "loss": 0.0472, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 4.278947368421053e-06, | |
| "loss": 0.0676, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 4.246052631578948e-06, | |
| "loss": 0.0595, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 4.213223684210526e-06, | |
| "loss": 0.0567, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_BLEU": 81.21370204393791, | |
| "eval_BLEU-Bigram-Precision": 83.26283987915409, | |
| "eval_BLEU-Trigram-Precision": 78.32402234636872, | |
| "eval_BLEU-Unigram-Precision": 88.63677950594693, | |
| "eval_ROUGE-2": 77.3545431272267, | |
| "eval_ROUGE-L": 90.77000002140947, | |
| "eval_Sacre-Bigram-Precision": 82.65461672473867, | |
| "eval_Sacre-Trigram-Precision": 78.53486104339346, | |
| "eval_Sacre-Unigram-Precision": 88.0695208169678, | |
| "eval_SacreBLEU": 81.05092176649303, | |
| "eval_loss": 0.2411227524280548, | |
| "eval_runtime": 1329.304, | |
| "eval_samples_per_second": 0.752, | |
| "eval_steps_per_second": 0.752, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 4.180394736842105e-06, | |
| "loss": 0.0561, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 4.1475000000000005e-06, | |
| "loss": 0.0484, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 4.114605263157895e-06, | |
| "loss": 0.0649, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 4.08171052631579e-06, | |
| "loss": 0.0489, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 4.048815789473684e-06, | |
| "loss": 0.0418, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 4.01592105263158e-06, | |
| "loss": 0.0671, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 3.983026315789474e-06, | |
| "loss": 0.0496, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 3.950131578947369e-06, | |
| "loss": 0.064, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 3.9173026315789475e-06, | |
| "loss": 0.0633, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 3.884407894736842e-06, | |
| "loss": 0.0578, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 3.8515131578947375e-06, | |
| "loss": 0.0501, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 3.818618421052632e-06, | |
| "loss": 0.0517, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 3.7857236842105266e-06, | |
| "loss": 0.0502, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 3.7528947368421054e-06, | |
| "loss": 0.0603, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 3.7200000000000004e-06, | |
| "loss": 0.0485, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 3.6871052631578954e-06, | |
| "loss": 0.0532, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 3.654276315789474e-06, | |
| "loss": 0.0551, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 3.6213815789473687e-06, | |
| "loss": 0.0429, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 3.5884868421052633e-06, | |
| "loss": 0.0485, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 3.5555921052631583e-06, | |
| "loss": 0.0606, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 3.5226973684210532e-06, | |
| "loss": 0.046, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 3.4898026315789474e-06, | |
| "loss": 0.0398, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 3.4569078947368424e-06, | |
| "loss": 0.0516, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 3.4240131578947373e-06, | |
| "loss": 0.06, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 3.391118421052632e-06, | |
| "loss": 0.0507, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 3.3582236842105265e-06, | |
| "loss": 0.0651, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 3.3253947368421057e-06, | |
| "loss": 0.0502, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 3.2925000000000002e-06, | |
| "loss": 0.0455, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 3.2596052631578952e-06, | |
| "loss": 0.0467, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 6.94, | |
| "learning_rate": 3.2267105263157898e-06, | |
| "loss": 0.0571, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 3.1938157894736843e-06, | |
| "loss": 0.0546, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.1609210526315793e-06, | |
| "loss": 0.043, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_BLEU": 80.56639568650905, | |
| "eval_BLEU-Bigram-Precision": 82.55929838548933, | |
| "eval_BLEU-Trigram-Precision": 77.6121051468964, | |
| "eval_BLEU-Unigram-Precision": 88.20010875475802, | |
| "eval_ROUGE-2": 77.17811658990045, | |
| "eval_ROUGE-L": 90.74508068067605, | |
| "eval_Sacre-Bigram-Precision": 81.82111373898087, | |
| "eval_Sacre-Trigram-Precision": 77.62557077625571, | |
| "eval_Sacre-Unigram-Precision": 87.54610755193167, | |
| "eval_SacreBLEU": 80.26422111075566, | |
| "eval_loss": 0.2396911382675171, | |
| "eval_runtime": 1545.8252, | |
| "eval_samples_per_second": 0.647, | |
| "eval_steps_per_second": 0.647, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 3.1280263157894743e-06, | |
| "loss": 0.0353, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 3.0951315789473685e-06, | |
| "loss": 0.05, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 3.0623026315789477e-06, | |
| "loss": 0.0403, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 3.0294078947368422e-06, | |
| "loss": 0.0384, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 2.996513157894737e-06, | |
| "loss": 0.0492, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 2.9636184210526318e-06, | |
| "loss": 0.0382, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 2.9307236842105263e-06, | |
| "loss": 0.0582, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 2.8978289473684213e-06, | |
| "loss": 0.037, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 2.865e-06, | |
| "loss": 0.0326, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 2.832105263157895e-06, | |
| "loss": 0.0447, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 2.7992105263157896e-06, | |
| "loss": 0.0531, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "learning_rate": 2.7663815789473684e-06, | |
| "loss": 0.0426, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 2.7334868421052634e-06, | |
| "loss": 0.0464, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 7.44, | |
| "learning_rate": 2.700592105263158e-06, | |
| "loss": 0.0491, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 7.47, | |
| "learning_rate": 2.667697368421053e-06, | |
| "loss": 0.0324, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 2.6348026315789475e-06, | |
| "loss": 0.0405, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 2.6019078947368425e-06, | |
| "loss": 0.0328, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 2.569013157894737e-06, | |
| "loss": 0.0456, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 7.59, | |
| "learning_rate": 2.536118421052632e-06, | |
| "loss": 0.0321, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 2.5032236842105266e-06, | |
| "loss": 0.0385, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 2.470328947368421e-06, | |
| "loss": 0.0385, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 2.437434210526316e-06, | |
| "loss": 0.0403, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 2.404605263157895e-06, | |
| "loss": 0.0563, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 2.3717105263157895e-06, | |
| "loss": 0.057, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 2.3388815789473687e-06, | |
| "loss": 0.0463, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 2.3059868421052633e-06, | |
| "loss": 0.0449, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 2.273092105263158e-06, | |
| "loss": 0.04, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 2.240197368421053e-06, | |
| "loss": 0.0447, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 7.91, | |
| "learning_rate": 2.2073026315789474e-06, | |
| "loss": 0.0557, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 2.1744078947368424e-06, | |
| "loss": 0.0602, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 2.141513157894737e-06, | |
| "loss": 0.0383, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 2.108618421052632e-06, | |
| "loss": 0.0491, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_BLEU": 78.11044908486078, | |
| "eval_BLEU-Bigram-Precision": 80.32311115410661, | |
| "eval_BLEU-Trigram-Precision": 75.1309180292829, | |
| "eval_BLEU-Unigram-Precision": 85.89573961365441, | |
| "eval_ROUGE-2": 76.73896747691722, | |
| "eval_ROUGE-L": 90.4316320530482, | |
| "eval_Sacre-Bigram-Precision": 79.53764448609809, | |
| "eval_Sacre-Trigram-Precision": 75.15945726545286, | |
| "eval_Sacre-Unigram-Precision": 85.18343864943884, | |
| "eval_SacreBLEU": 77.77903164926069, | |
| "eval_loss": 0.24439701437950134, | |
| "eval_runtime": 1491.254, | |
| "eval_samples_per_second": 0.671, | |
| "eval_steps_per_second": 0.671, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 2.0757236842105265e-06, | |
| "loss": 0.0393, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 2.0428289473684215e-06, | |
| "loss": 0.0466, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 2.009934210526316e-06, | |
| "loss": 0.0377, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 1.977105263157895e-06, | |
| "loss": 0.0413, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 8.16, | |
| "learning_rate": 1.9442105263157894e-06, | |
| "loss": 0.0394, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 1.9113157894736843e-06, | |
| "loss": 0.03, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 1.8784210526315791e-06, | |
| "loss": 0.0286, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 1.8455263157894737e-06, | |
| "loss": 0.0403, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "learning_rate": 1.8126973684210527e-06, | |
| "loss": 0.0543, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 1.7798026315789474e-06, | |
| "loss": 0.0362, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 1.7469078947368422e-06, | |
| "loss": 0.0321, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 1.714013157894737e-06, | |
| "loss": 0.0306, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 1.6811184210526318e-06, | |
| "loss": 0.0385, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 1.6482236842105265e-06, | |
| "loss": 0.0347, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 1.615328947368421e-06, | |
| "loss": 0.0352, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 1.5825e-06, | |
| "loss": 0.0437, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 8.53, | |
| "learning_rate": 1.5496052631578949e-06, | |
| "loss": 0.0324, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 1.5167105263157896e-06, | |
| "loss": 0.0354, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 1.4838157894736844e-06, | |
| "loss": 0.0367, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 1.450921052631579e-06, | |
| "loss": 0.0408, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 8.66, | |
| "learning_rate": 1.418026315789474e-06, | |
| "loss": 0.0543, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 1.3851315789473685e-06, | |
| "loss": 0.0245, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 1.3523026315789475e-06, | |
| "loss": 0.0283, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 1.319407894736842e-06, | |
| "loss": 0.0353, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 8.78, | |
| "learning_rate": 1.286513157894737e-06, | |
| "loss": 0.0442, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 1.2536184210526316e-06, | |
| "loss": 0.0396, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 1.2207236842105264e-06, | |
| "loss": 0.0382, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 1.1878289473684212e-06, | |
| "loss": 0.0393, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 1.154934210526316e-06, | |
| "loss": 0.0279, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 8.94, | |
| "learning_rate": 1.1220394736842107e-06, | |
| "loss": 0.0368, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 1.0891447368421055e-06, | |
| "loss": 0.0339, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 1.0563157894736843e-06, | |
| "loss": 0.0367, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_BLEU": 78.72613610038634, | |
| "eval_BLEU-Bigram-Precision": 80.66982867099023, | |
| "eval_BLEU-Trigram-Precision": 75.83146187573522, | |
| "eval_BLEU-Unigram-Precision": 86.24128497043509, | |
| "eval_ROUGE-2": 77.10014496108923, | |
| "eval_ROUGE-L": 90.75201858392994, | |
| "eval_Sacre-Bigram-Precision": 79.9499791579825, | |
| "eval_Sacre-Trigram-Precision": 75.83565459610028, | |
| "eval_Sacre-Unigram-Precision": 85.47565118912797, | |
| "eval_SacreBLEU": 78.39445859677504, | |
| "eval_loss": 0.2408515363931656, | |
| "eval_runtime": 1473.0938, | |
| "eval_samples_per_second": 0.679, | |
| "eval_steps_per_second": 0.679, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 1.0234868421052633e-06, | |
| "loss": 0.0404, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 9.90592105263158e-07, | |
| "loss": 0.0337, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 9.576973684210528e-07, | |
| "loss": 0.0426, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 9.248026315789474e-07, | |
| "loss": 0.0408, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 8.919078947368421e-07, | |
| "loss": 0.0347, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 8.59078947368421e-07, | |
| "loss": 0.0392, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 8.261842105263158e-07, | |
| "loss": 0.0339, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 7.932894736842106e-07, | |
| "loss": 0.0371, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 7.603947368421054e-07, | |
| "loss": 0.0375, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 7.275e-07, | |
| "loss": 0.0359, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 6.946052631578948e-07, | |
| "loss": 0.0241, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 6.617105263157896e-07, | |
| "loss": 0.031, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 6.288815789473685e-07, | |
| "loss": 0.0294, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 5.959868421052632e-07, | |
| "loss": 0.0371, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 5.630921052631579e-07, | |
| "loss": 0.0263, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 5.301973684210527e-07, | |
| "loss": 0.049, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 4.973026315789474e-07, | |
| "loss": 0.033, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 4.6440789473684216e-07, | |
| "loss": 0.045, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 4.315131578947369e-07, | |
| "loss": 0.0464, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 3.986842105263158e-07, | |
| "loss": 0.0353, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 3.6578947368421055e-07, | |
| "loss": 0.0365, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 9.69, | |
| "learning_rate": 3.328947368421053e-07, | |
| "loss": 0.0335, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 0.0404, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 2.67171052631579e-07, | |
| "loss": 0.0374, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 2.3427631578947373e-07, | |
| "loss": 0.0366, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 9.81, | |
| "learning_rate": 2.0138157894736842e-07, | |
| "loss": 0.0245, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 1.6848684210526317e-07, | |
| "loss": 0.039, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 1.3559210526315791e-07, | |
| "loss": 0.0292, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 1.0269736842105265e-07, | |
| "loss": 0.0361, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 9.94, | |
| "learning_rate": 6.980263157894737e-08, | |
| "loss": 0.0372, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 3.690789473684211e-08, | |
| "loss": 0.0211, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 4.0131578947368425e-09, | |
| "loss": 0.0354, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_BLEU": 78.74942323410538, | |
| "eval_BLEU-Bigram-Precision": 80.77783179387458, | |
| "eval_BLEU-Trigram-Precision": 75.84094572810316, | |
| "eval_BLEU-Unigram-Precision": 86.30926007975188, | |
| "eval_ROUGE-2": 77.17672035094623, | |
| "eval_ROUGE-L": 90.75273654124103, | |
| "eval_Sacre-Bigram-Precision": 80.08164119740422, | |
| "eval_Sacre-Trigram-Precision": 75.89223233030091, | |
| "eval_Sacre-Unigram-Precision": 85.5694523403449, | |
| "eval_SacreBLEU": 78.4509112113324, | |
| "eval_loss": 0.2423561066389084, | |
| "eval_runtime": 1552.9635, | |
| "eval_samples_per_second": 0.644, | |
| "eval_steps_per_second": 0.644, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 160000, | |
| "total_flos": 2093466255360000.0, | |
| "train_loss": 0.08072224444150924, | |
| "train_runtime": 26837.0061, | |
| "train_samples_per_second": 5.962, | |
| "train_steps_per_second": 5.962 | |
| } | |
| ], | |
| "max_steps": 160000, | |
| "num_train_epochs": 10, | |
| "total_flos": 2093466255360000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |