| { | |
| "best_metric": 0.4481608271598816, | |
| "best_model_checkpoint": "./distilbert-marian-training-1/checkpoint-123440", | |
| "epoch": 10.0, | |
| "global_step": 123440, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.068697342838627e-07, | |
| "loss": 0.5266, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.6169799092676605e-06, | |
| "loss": 0.497, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.4270900842514583e-06, | |
| "loss": 0.5398, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.2372002592352565e-06, | |
| "loss": 0.5439, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.047310434219054e-06, | |
| "loss": 0.5357, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.857420609202852e-06, | |
| "loss": 0.5693, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5.66753078418665e-06, | |
| "loss": 0.5311, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.477640959170448e-06, | |
| "loss": 0.5433, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.286130913804278e-06, | |
| "loss": 0.5403, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.094620868438109e-06, | |
| "loss": 0.5478, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.904731043421905e-06, | |
| "loss": 0.5556, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.714841218405703e-06, | |
| "loss": 0.5835, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.972370979295291e-06, | |
| "loss": 0.5578, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.929733601664564e-06, | |
| "loss": 0.5389, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.887181498789099e-06, | |
| "loss": 0.5645, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.844544121158372e-06, | |
| "loss": 0.5672, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.801906743527647e-06, | |
| "loss": 0.559, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.759354640652181e-06, | |
| "loss": 0.5927, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.716717263021456e-06, | |
| "loss": 0.5241, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.67407988539073e-06, | |
| "loss": 0.5458, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.631442507760004e-06, | |
| "loss": 0.5902, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.588805130129278e-06, | |
| "loss": 0.5425, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.546167752498551e-06, | |
| "loss": 0.5225, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.503615649623087e-06, | |
| "loss": 0.5301, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_BLEU": 39.52568484782476, | |
| "eval_BLEU-Bigram-Precision": 44.566237100676446, | |
| "eval_BLEU-Trigram-Precision": 34.507782919646615, | |
| "eval_BLEU-Unigram-Precision": 63.26198231735691, | |
| "eval_ROUGE-2": 33.35721035298087, | |
| "eval_ROUGE-L": 47.29001112126434, | |
| "eval_Sacre-Bigram-Precision": 42.86996462699963, | |
| "eval_Sacre-Trigram-Precision": 34.29733393583371, | |
| "eval_Sacre-Unigram-Precision": 61.48775894538606, | |
| "eval_SacreBLEU": 38.73636411085535, | |
| "eval_loss": 0.7005472183227539, | |
| "eval_runtime": 848.4996, | |
| "eval_samples_per_second": 1.458, | |
| "eval_steps_per_second": 1.458, | |
| "step": 12344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.46097827199236e-06, | |
| "loss": 0.5475, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.418340894361634e-06, | |
| "loss": 0.5107, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.375703516730907e-06, | |
| "loss": 0.5036, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.333066139100182e-06, | |
| "loss": 0.4865, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.290428761469455e-06, | |
| "loss": 0.5065, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.247791383838728e-06, | |
| "loss": 0.5159, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.205154006208003e-06, | |
| "loss": 0.5221, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 9.16260190333254e-06, | |
| "loss": 0.4597, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.119964525701812e-06, | |
| "loss": 0.5024, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.077327148071086e-06, | |
| "loss": 0.4852, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 9.034689770440359e-06, | |
| "loss": 0.4887, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.992052392809634e-06, | |
| "loss": 0.4796, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 8.949415015178907e-06, | |
| "loss": 0.4754, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 8.90677763754818e-06, | |
| "loss": 0.5062, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.864140259917455e-06, | |
| "loss": 0.5204, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 8.821502882286729e-06, | |
| "loss": 0.4799, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.778950779411265e-06, | |
| "loss": 0.4666, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.736313401780538e-06, | |
| "loss": 0.4484, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.693676024149813e-06, | |
| "loss": 0.4721, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.651038646519086e-06, | |
| "loss": 0.4867, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 8.60840126888836e-06, | |
| "loss": 0.47, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 8.565763891257632e-06, | |
| "loss": 0.4848, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.523126513626907e-06, | |
| "loss": 0.4429, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.480574410751442e-06, | |
| "loss": 0.4501, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.437937033120715e-06, | |
| "loss": 0.4363, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_BLEU": 44.07579241900836, | |
| "eval_BLEU-Bigram-Precision": 48.638266436838215, | |
| "eval_BLEU-Trigram-Precision": 39.453534717851895, | |
| "eval_BLEU-Unigram-Precision": 65.81561600594188, | |
| "eval_ROUGE-2": 37.93490065934162, | |
| "eval_ROUGE-L": 51.37484166047172, | |
| "eval_Sacre-Bigram-Precision": 47.097215642928575, | |
| "eval_Sacre-Trigram-Precision": 39.10595653642608, | |
| "eval_Sacre-Unigram-Precision": 64.20241154378336, | |
| "eval_SacreBLEU": 43.28181370216754, | |
| "eval_loss": 0.6306515336036682, | |
| "eval_runtime": 969.349, | |
| "eval_samples_per_second": 1.276, | |
| "eval_steps_per_second": 1.276, | |
| "step": 24688 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.395299655489988e-06, | |
| "loss": 0.4153, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.352662277859263e-06, | |
| "loss": 0.4241, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 8.310024900228536e-06, | |
| "loss": 0.401, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 8.267387522597811e-06, | |
| "loss": 0.4351, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 8.224750144967085e-06, | |
| "loss": 0.4075, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.18211276733636e-06, | |
| "loss": 0.4106, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.139560664460894e-06, | |
| "loss": 0.4175, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.096923286830167e-06, | |
| "loss": 0.3897, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 8.05428590919944e-06, | |
| "loss": 0.4152, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 8.011648531568715e-06, | |
| "loss": 0.3917, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.96909642869325e-06, | |
| "loss": 0.4081, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.926459051062525e-06, | |
| "loss": 0.3964, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 7.883821673431798e-06, | |
| "loss": 0.3833, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.841184295801073e-06, | |
| "loss": 0.3884, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 7.798546918170346e-06, | |
| "loss": 0.4027, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 7.75590954053962e-06, | |
| "loss": 0.4272, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 7.713272162908894e-06, | |
| "loss": 0.3868, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 7.670634785278167e-06, | |
| "loss": 0.3848, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 7.628082682402702e-06, | |
| "loss": 0.3919, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 7.585445304771976e-06, | |
| "loss": 0.3803, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.54280792714125e-06, | |
| "loss": 0.3986, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.500170549510524e-06, | |
| "loss": 0.3722, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 7.4576184466350585e-06, | |
| "loss": 0.3795, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 7.4149810690043326e-06, | |
| "loss": 0.3768, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.372343691373607e-06, | |
| "loss": 0.3983, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_BLEU": 47.64065251911936, | |
| "eval_BLEU-Bigram-Precision": 52.275310044962694, | |
| "eval_BLEU-Trigram-Precision": 43.41648247552889, | |
| "eval_BLEU-Unigram-Precision": 68.46246973365618, | |
| "eval_ROUGE-2": 41.91844760109249, | |
| "eval_ROUGE-L": 54.326234865452385, | |
| "eval_Sacre-Bigram-Precision": 50.88183888565224, | |
| "eval_Sacre-Trigram-Precision": 43.238494672409885, | |
| "eval_Sacre-Unigram-Precision": 67.02455512476389, | |
| "eval_SacreBLEU": 46.828668155215276, | |
| "eval_loss": 0.5844058990478516, | |
| "eval_runtime": 950.5601, | |
| "eval_samples_per_second": 1.301, | |
| "eval_steps_per_second": 1.301, | |
| "step": 37032 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.329706313742881e-06, | |
| "loss": 0.3116, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 7.287154210867415e-06, | |
| "loss": 0.3262, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 7.244516833236689e-06, | |
| "loss": 0.3429, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 7.201879455605963e-06, | |
| "loss": 0.3415, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 7.159242077975237e-06, | |
| "loss": 0.3407, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 7.116689975099772e-06, | |
| "loss": 0.3288, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 7.074052597469046e-06, | |
| "loss": 0.3499, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 7.03141521983832e-06, | |
| "loss": 0.3396, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 6.988863116962855e-06, | |
| "loss": 0.3395, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 6.946225739332128e-06, | |
| "loss": 0.3462, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 6.9035883617014025e-06, | |
| "loss": 0.3362, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 6.8609509840706765e-06, | |
| "loss": 0.3284, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 6.818313606439951e-06, | |
| "loss": 0.3446, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.775676228809224e-06, | |
| "loss": 0.36, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 6.733038851178497e-06, | |
| "loss": 0.3368, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 6.690401473547772e-06, | |
| "loss": 0.3362, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 6.647764095917045e-06, | |
| "loss": 0.338, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 6.6052119930415805e-06, | |
| "loss": 0.3206, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 6.562659890166116e-06, | |
| "loss": 0.339, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 6.52002251253539e-06, | |
| "loss": 0.3226, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 6.477385134904664e-06, | |
| "loss": 0.3386, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 6.434747757273937e-06, | |
| "loss": 0.3406, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 6.39211037964321e-06, | |
| "loss": 0.3231, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.349473002012485e-06, | |
| "loss": 0.3418, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_BLEU": 51.32592897573737, | |
| "eval_BLEU-Bigram-Precision": 55.485615933120236, | |
| "eval_BLEU-Trigram-Precision": 47.36897219749725, | |
| "eval_BLEU-Unigram-Precision": 70.36899684776562, | |
| "eval_ROUGE-2": 45.802241938436794, | |
| "eval_ROUGE-L": 57.842922925143036, | |
| "eval_Sacre-Bigram-Precision": 54.30648474951272, | |
| "eval_Sacre-Trigram-Precision": 47.28686538569899, | |
| "eval_Sacre-Unigram-Precision": 69.06033630069238, | |
| "eval_SacreBLEU": 50.59726409455209, | |
| "eval_loss": 0.5526171326637268, | |
| "eval_runtime": 916.9598, | |
| "eval_samples_per_second": 1.349, | |
| "eval_steps_per_second": 1.349, | |
| "step": 49376 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 6.3069208991370205e-06, | |
| "loss": 0.326, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.264283521506294e-06, | |
| "loss": 0.2931, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 6.221646143875567e-06, | |
| "loss": 0.2846, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 6.179008766244842e-06, | |
| "loss": 0.2896, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 6.136371388614115e-06, | |
| "loss": 0.2961, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 6.0937340109833884e-06, | |
| "loss": 0.2679, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 6.051096633352663e-06, | |
| "loss": 0.3005, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 6.0085445304771986e-06, | |
| "loss": 0.2921, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 5.965907152846472e-06, | |
| "loss": 0.2871, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 5.923269775215745e-06, | |
| "loss": 0.2853, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 5.88063239758502e-06, | |
| "loss": 0.2759, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 5.837995019954293e-06, | |
| "loss": 0.2793, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 5.7953576423235665e-06, | |
| "loss": 0.2924, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 5.7527202646928406e-06, | |
| "loss": 0.3162, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 5.710168161817377e-06, | |
| "loss": 0.2815, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 5.66753078418665e-06, | |
| "loss": 0.308, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 5.624893406555923e-06, | |
| "loss": 0.2984, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 5.582256028925197e-06, | |
| "loss": 0.2928, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 5.539618651294471e-06, | |
| "loss": 0.2903, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 5.496981273663745e-06, | |
| "loss": 0.2971, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 5.454343896033019e-06, | |
| "loss": 0.2815, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.411706518402292e-06, | |
| "loss": 0.3028, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 5.369154415526828e-06, | |
| "loss": 0.2931, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.326517037896102e-06, | |
| "loss": 0.2859, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 5.283879660265375e-06, | |
| "loss": 0.2838, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_BLEU": 53.29609495182536, | |
| "eval_BLEU-Bigram-Precision": 57.167285518858705, | |
| "eval_BLEU-Trigram-Precision": 49.0354115750611, | |
| "eval_BLEU-Unigram-Precision": 71.76226675881134, | |
| "eval_ROUGE-2": 48.21696209970032, | |
| "eval_ROUGE-L": 59.61148640581645, | |
| "eval_Sacre-Bigram-Precision": 56.00606885005755, | |
| "eval_Sacre-Trigram-Precision": 48.951166303070984, | |
| "eval_Sacre-Unigram-Precision": 70.51250552798388, | |
| "eval_SacreBLEU": 52.60559213285124, | |
| "eval_loss": 0.5165597796440125, | |
| "eval_runtime": 915.3947, | |
| "eval_samples_per_second": 1.351, | |
| "eval_steps_per_second": 1.351, | |
| "step": 61720 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 5.2412422826346485e-06, | |
| "loss": 0.2663, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 5.1986901797591846e-06, | |
| "loss": 0.25, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 5.156052802128459e-06, | |
| "loss": 0.2474, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 5.113415424497732e-06, | |
| "loss": 0.2469, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 5.070778046867005e-06, | |
| "loss": 0.277, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 5.02814066923628e-06, | |
| "loss": 0.2478, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 4.985588566360815e-06, | |
| "loss": 0.2593, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 4.9429511887300885e-06, | |
| "loss": 0.2565, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 4.900313811099363e-06, | |
| "loss": 0.2678, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 4.857676433468637e-06, | |
| "loss": 0.2434, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 4.815124330593172e-06, | |
| "loss": 0.2776, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 4.772486952962445e-06, | |
| "loss": 0.2678, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 4.729849575331719e-06, | |
| "loss": 0.2784, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 4.687212197700993e-06, | |
| "loss": 0.2404, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 4.6445748200702666e-06, | |
| "loss": 0.26, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 4.601937442439541e-06, | |
| "loss": 0.242, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 4.559300064808814e-06, | |
| "loss": 0.2636, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 4.516662687178088e-06, | |
| "loss": 0.2703, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 4.474110584302623e-06, | |
| "loss": 0.2457, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 4.431473206671897e-06, | |
| "loss": 0.2492, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 4.3888358290411705e-06, | |
| "loss": 0.2677, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 4.346198451410445e-06, | |
| "loss": 0.2557, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 4.303561073779719e-06, | |
| "loss": 0.262, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 4.261008970904254e-06, | |
| "loss": 0.247, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 4.218371593273527e-06, | |
| "loss": 0.2461, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_BLEU": 56.19871507978884, | |
| "eval_BLEU-Bigram-Precision": 58.651824957003626, | |
| "eval_BLEU-Trigram-Precision": 51.109418634171114, | |
| "eval_BLEU-Unigram-Precision": 72.28111326627273, | |
| "eval_ROUGE-2": 50.26575052573452, | |
| "eval_ROUGE-L": 61.216031502511356, | |
| "eval_Sacre-Bigram-Precision": 57.60808534531162, | |
| "eval_Sacre-Trigram-Precision": 51.07878391631252, | |
| "eval_Sacre-Unigram-Precision": 71.06299212598425, | |
| "eval_SacreBLEU": 55.641908183294326, | |
| "eval_loss": 0.49508827924728394, | |
| "eval_runtime": 1114.4052, | |
| "eval_samples_per_second": 1.11, | |
| "eval_steps_per_second": 1.11, | |
| "step": 74064 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 4.175734215642801e-06, | |
| "loss": 0.2288, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 4.133096838012075e-06, | |
| "loss": 0.2143, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 4.0904594603813494e-06, | |
| "loss": 0.2089, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 4.047907357505884e-06, | |
| "loss": 0.2401, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 4.00535525463042e-06, | |
| "loss": 0.2384, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 3.962717876999693e-06, | |
| "loss": 0.2268, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 3.920080499368967e-06, | |
| "loss": 0.2197, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 3.87744312173824e-06, | |
| "loss": 0.2373, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 3.8348910188627765e-06, | |
| "loss": 0.2261, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 3.7922536412320497e-06, | |
| "loss": 0.2299, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 3.749616263601324e-06, | |
| "loss": 0.2267, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 3.7069788859705975e-06, | |
| "loss": 0.2196, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 3.6643415083398716e-06, | |
| "loss": 0.2192, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 3.621704130709145e-06, | |
| "loss": 0.2365, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 3.579066753078419e-06, | |
| "loss": 0.2187, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 3.5364293754476926e-06, | |
| "loss": 0.242, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 3.493877272572228e-06, | |
| "loss": 0.2483, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 3.4512398949415014e-06, | |
| "loss": 0.2399, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 3.4086025173107755e-06, | |
| "loss": 0.2409, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 3.365965139680049e-06, | |
| "loss": 0.242, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 3.323413036804585e-06, | |
| "loss": 0.2484, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 3.280775659173858e-06, | |
| "loss": 0.2562, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 3.238138281543132e-06, | |
| "loss": 0.2283, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 3.1955009039124062e-06, | |
| "loss": 0.236, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_BLEU": 57.78119869959703, | |
| "eval_BLEU-Bigram-Precision": 61.05482573076737, | |
| "eval_BLEU-Trigram-Precision": 53.63036303630363, | |
| "eval_BLEU-Unigram-Precision": 74.23397054788255, | |
| "eval_ROUGE-2": 52.1883256479329, | |
| "eval_ROUGE-L": 63.12941898321425, | |
| "eval_Sacre-Bigram-Precision": 60.065305276251685, | |
| "eval_Sacre-Trigram-Precision": 53.55263886581381, | |
| "eval_Sacre-Unigram-Precision": 73.0894744532658, | |
| "eval_SacreBLEU": 57.19500795208099, | |
| "eval_loss": 0.4768661558628082, | |
| "eval_runtime": 1205.3601, | |
| "eval_samples_per_second": 1.026, | |
| "eval_steps_per_second": 1.026, | |
| "step": 86408 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 3.1529488010369415e-06, | |
| "loss": 0.2257, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 3.1103114234062147e-06, | |
| "loss": 0.2245, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 3.0676740457754888e-06, | |
| "loss": 0.215, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 3.025036668144763e-06, | |
| "loss": 0.2102, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 2.982484565269298e-06, | |
| "loss": 0.211, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 2.9398471876385713e-06, | |
| "loss": 0.236, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 2.8972098100078454e-06, | |
| "loss": 0.2069, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 2.8545724323771195e-06, | |
| "loss": 0.2169, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 2.8120203295016547e-06, | |
| "loss": 0.209, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 2.769382951870928e-06, | |
| "loss": 0.2195, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 2.726745574240202e-06, | |
| "loss": 0.2074, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 2.6841934713647373e-06, | |
| "loss": 0.2112, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 2.6415560937340113e-06, | |
| "loss": 0.2145, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 2.5989187161032846e-06, | |
| "loss": 0.213, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 2.5562813384725587e-06, | |
| "loss": 0.1966, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 2.5136439608418328e-06, | |
| "loss": 0.2097, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 2.4710065832111064e-06, | |
| "loss": 0.2167, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 2.42836920558038e-06, | |
| "loss": 0.2144, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 2.3858171027049153e-06, | |
| "loss": 0.2277, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 2.343179725074189e-06, | |
| "loss": 0.2347, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 2.300542347443463e-06, | |
| "loss": 0.2235, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 2.2579049698127367e-06, | |
| "loss": 0.2197, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 2.2152675921820104e-06, | |
| "loss": 0.2219, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 2.1726302145512845e-06, | |
| "loss": 0.21, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 7.98, | |
| "learning_rate": 2.129992836920558e-06, | |
| "loss": 0.2014, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_BLEU": 59.598575339554586, | |
| "eval_BLEU-Bigram-Precision": 63.173579559333625, | |
| "eval_BLEU-Trigram-Precision": 56.19508137056102, | |
| "eval_BLEU-Unigram-Precision": 75.73942688657515, | |
| "eval_ROUGE-2": 54.010901558700496, | |
| "eval_ROUGE-L": 64.55563198822112, | |
| "eval_Sacre-Bigram-Precision": 62.145801805562805, | |
| "eval_Sacre-Trigram-Precision": 56.11089418196017, | |
| "eval_Sacre-Unigram-Precision": 74.65196078431373, | |
| "eval_SacreBLEU": 59.05906731573519, | |
| "eval_loss": 0.4607221782207489, | |
| "eval_runtime": 1200.6288, | |
| "eval_samples_per_second": 1.03, | |
| "eval_steps_per_second": 1.03, | |
| "step": 98752 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.0873554592898322e-06, | |
| "loss": 0.2166, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 2.0448033564143674e-06, | |
| "loss": 0.1819, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 2.002165978783641e-06, | |
| "loss": 0.1905, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 1.9595286011529148e-06, | |
| "loss": 0.2006, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 1.916891223522189e-06, | |
| "loss": 0.2078, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 1.8742538458914625e-06, | |
| "loss": 0.2051, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 1.8317017430159977e-06, | |
| "loss": 0.207, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 1.7890643653852716e-06, | |
| "loss": 0.1757, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 1.7464269877545453e-06, | |
| "loss": 0.2117, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 1.7037896101238192e-06, | |
| "loss": 0.2073, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 1.6611522324930928e-06, | |
| "loss": 0.2056, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 1.6186001296176283e-06, | |
| "loss": 0.1992, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 8.51, | |
| "learning_rate": 1.575962751986902e-06, | |
| "loss": 0.2169, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 1.5333253743561758e-06, | |
| "loss": 0.2039, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 1.4906879967254495e-06, | |
| "loss": 0.1897, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 1.4480506190947233e-06, | |
| "loss": 0.2109, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 1.4054985162192586e-06, | |
| "loss": 0.2002, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 1.3628611385885324e-06, | |
| "loss": 0.2071, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 1.3203090357130676e-06, | |
| "loss": 0.2247, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 1.2776716580823415e-06, | |
| "loss": 0.2092, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 1.2350342804516152e-06, | |
| "loss": 0.2184, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 1.1924821775761504e-06, | |
| "loss": 0.2012, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 1.1498447999454243e-06, | |
| "loss": 0.2024, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 1.1072074223146981e-06, | |
| "loss": 0.2178, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 1.0645700446839718e-06, | |
| "loss": 0.2087, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_BLEU": 59.228700736361574, | |
| "eval_BLEU-Bigram-Precision": 62.02898550724638, | |
| "eval_BLEU-Trigram-Precision": 55.13255240443896, | |
| "eval_BLEU-Unigram-Precision": 74.83247481424078, | |
| "eval_ROUGE-2": 53.99481361299642, | |
| "eval_ROUGE-L": 64.48941872388136, | |
| "eval_Sacre-Bigram-Precision": 61.05263157894737, | |
| "eval_Sacre-Trigram-Precision": 55.05952380952381, | |
| "eval_Sacre-Unigram-Precision": 73.72556628025416, | |
| "eval_SacreBLEU": 58.70218229892996, | |
| "eval_loss": 0.4529183506965637, | |
| "eval_runtime": 1120.8668, | |
| "eval_samples_per_second": 1.104, | |
| "eval_steps_per_second": 1.104, | |
| "step": 111096 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 1.0219326670532457e-06, | |
| "loss": 0.2001, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 9.792952894225194e-07, | |
| "loss": 0.1909, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 9.366579117917933e-07, | |
| "loss": 0.1929, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 8.940205341610671e-07, | |
| "loss": 0.1847, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 8.514684312856022e-07, | |
| "loss": 0.1883, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 8.08831053654876e-07, | |
| "loss": 0.1995, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 7.662789507794113e-07, | |
| "loss": 0.2046, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 7.236415731486851e-07, | |
| "loss": 0.1945, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 6.810041955179588e-07, | |
| "loss": 0.2007, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 6.383668178872326e-07, | |
| "loss": 0.1966, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 5.957294402565065e-07, | |
| "loss": 0.1924, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 5.530920626257804e-07, | |
| "loss": 0.1974, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 9.52, | |
| "learning_rate": 5.104546849950541e-07, | |
| "loss": 0.2053, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 4.6781730736432786e-07, | |
| "loss": 0.2116, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 4.251799297336017e-07, | |
| "loss": 0.2003, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 3.8254255210287546e-07, | |
| "loss": 0.1966, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 3.3990517447214934e-07, | |
| "loss": 0.1966, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 2.9735307159668455e-07, | |
| "loss": 0.2016, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 2.547156939659583e-07, | |
| "loss": 0.2098, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 2.1207831633523215e-07, | |
| "loss": 0.2038, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 1.6944093870450592e-07, | |
| "loss": 0.1956, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 1.2680356107377972e-07, | |
| "loss": 0.2036, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 8.416618344305353e-08, | |
| "loss": 0.22, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 4.1528805812327316e-08, | |
| "loss": 0.2003, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_BLEU": 60.284861769008316, | |
| "eval_BLEU-Bigram-Precision": 62.98832834957076, | |
| "eval_BLEU-Trigram-Precision": 56.25192327418197, | |
| "eval_BLEU-Unigram-Precision": 75.38118428838013, | |
| "eval_ROUGE-2": 54.858141016126524, | |
| "eval_ROUGE-L": 65.28596759853026, | |
| "eval_Sacre-Bigram-Precision": 62.07553975369712, | |
| "eval_Sacre-Trigram-Precision": 56.22145176379946, | |
| "eval_Sacre-Unigram-Precision": 74.30246076341794, | |
| "eval_SacreBLEU": 59.77673945128229, | |
| "eval_loss": 0.4481608271598816, | |
| "eval_runtime": 1123.8109, | |
| "eval_samples_per_second": 1.101, | |
| "eval_steps_per_second": 1.101, | |
| "step": 123440 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 123440, | |
| "total_flos": 3230087590379520.0, | |
| "train_loss": 0.31656123802249336, | |
| "train_runtime": 24698.2128, | |
| "train_samples_per_second": 9.995, | |
| "train_steps_per_second": 4.998 | |
| } | |
| ], | |
| "max_steps": 123440, | |
| "num_train_epochs": 10, | |
| "total_flos": 3230087590379520.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |