| { | |
| "best_metric": 0.6102247834205627, | |
| "best_model_checkpoint": "./luke-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": 8.052495139338951e-07, | |
| "loss": 0.7856, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.6153596889176928e-06, | |
| "loss": 0.748, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.4254698639014908e-06, | |
| "loss": 0.7809, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.2355800388852886e-06, | |
| "loss": 0.7923, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.044069993519119e-06, | |
| "loss": 0.7799, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.854180168502917e-06, | |
| "loss": 0.8085, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 5.664290343486715e-06, | |
| "loss": 0.7759, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.474400518470513e-06, | |
| "loss": 0.7833, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.2828904731043424e-06, | |
| "loss": 0.7804, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.09300064808814e-06, | |
| "loss": 0.8169, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.903110823071939e-06, | |
| "loss": 0.8058, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.713220998055737e-06, | |
| "loss": 0.8265, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.972541528805813e-06, | |
| "loss": 0.7904, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.929904151175087e-06, | |
| "loss": 0.7829, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.88726677354436e-06, | |
| "loss": 0.8126, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.844629395913635e-06, | |
| "loss": 0.8133, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.802077293038171e-06, | |
| "loss": 0.7928, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.759525190162705e-06, | |
| "loss": 0.8564, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.716887812531979e-06, | |
| "loss": 0.7624, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.674250434901252e-06, | |
| "loss": 0.7662, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.631613057270527e-06, | |
| "loss": 0.8363, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.5889756796398e-06, | |
| "loss": 0.7793, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.546338302009073e-06, | |
| "loss": 0.7561, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.50378619913361e-06, | |
| "loss": 0.774, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_BLEU": 32.0811169920738, | |
| "eval_BLEU-Bigram-Precision": 37.1092045950269, | |
| "eval_BLEU-Trigram-Precision": 26.487573476332887, | |
| "eval_BLEU-Unigram-Precision": 56.60325589903054, | |
| "eval_ROUGE-2": 25.923225512351976, | |
| "eval_ROUGE-L": 40.443599801980355, | |
| "eval_Sacre-Bigram-Precision": 34.86367988711385, | |
| "eval_Sacre-Trigram-Precision": 26.20122541115769, | |
| "eval_Sacre-Unigram-Precision": 54.42125237191651, | |
| "eval_SacreBLEU": 30.767354925943692, | |
| "eval_loss": 0.9131098985671997, | |
| "eval_runtime": 174.9979, | |
| "eval_samples_per_second": 7.069, | |
| "eval_steps_per_second": 7.069, | |
| "step": 12344 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.461148821502884e-06, | |
| "loss": 0.7846, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.418511443872157e-06, | |
| "loss": 0.7368, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.37587406624143e-06, | |
| "loss": 0.7279, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.333236688610704e-06, | |
| "loss": 0.7253, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.290599310979979e-06, | |
| "loss": 0.7259, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.247961933349252e-06, | |
| "loss": 0.7432, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.205324555718525e-06, | |
| "loss": 0.7437, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 9.1626871780878e-06, | |
| "loss": 0.6641, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.120135075212335e-06, | |
| "loss": 0.7239, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.077497697581608e-06, | |
| "loss": 0.7007, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 9.034860319950883e-06, | |
| "loss": 0.7107, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.992222942320156e-06, | |
| "loss": 0.7033, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 8.949670839444692e-06, | |
| "loss": 0.6932, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 8.907118736569226e-06, | |
| "loss": 0.7351, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.8644813589385e-06, | |
| "loss": 0.7511, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 8.821843981307775e-06, | |
| "loss": 0.6994, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.779206603677048e-06, | |
| "loss": 0.6792, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.736569226046321e-06, | |
| "loss": 0.6701, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.693931848415596e-06, | |
| "loss": 0.6885, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.65129447078487e-06, | |
| "loss": 0.6796, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 8.608657093154144e-06, | |
| "loss": 0.6812, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 8.566104990278679e-06, | |
| "loss": 0.704, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.523467612647952e-06, | |
| "loss": 0.6443, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.480830235017227e-06, | |
| "loss": 0.6513, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.4381928573865e-06, | |
| "loss": 0.6532, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_BLEU": 34.14564345469897, | |
| "eval_BLEU-Bigram-Precision": 42.337954289996254, | |
| "eval_BLEU-Trigram-Precision": 32.169876203576344, | |
| "eval_BLEU-Unigram-Precision": 61.27510040160643, | |
| "eval_ROUGE-2": 28.971835230430752, | |
| "eval_ROUGE-L": 43.230417531342944, | |
| "eval_Sacre-Bigram-Precision": 40.19043401240035, | |
| "eval_Sacre-Trigram-Precision": 31.907054139181078, | |
| "eval_Sacre-Unigram-Precision": 59.04875395057251, | |
| "eval_SacreBLEU": 33.00279853845508, | |
| "eval_loss": 0.8338403701782227, | |
| "eval_runtime": 167.0515, | |
| "eval_samples_per_second": 7.405, | |
| "eval_steps_per_second": 7.405, | |
| "step": 24688 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.395555479755773e-06, | |
| "loss": 0.6209, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.35300337688031e-06, | |
| "loss": 0.6246, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 8.310365999249583e-06, | |
| "loss": 0.6061, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 8.267728621618858e-06, | |
| "loss": 0.6347, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 8.22509124398813e-06, | |
| "loss": 0.609, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.182539141112665e-06, | |
| "loss": 0.6059, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.13990176348194e-06, | |
| "loss": 0.6301, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 8.097264385851213e-06, | |
| "loss": 0.5859, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 8.054627008220487e-06, | |
| "loss": 0.6133, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 8.012074905345023e-06, | |
| "loss": 0.5813, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.969437527714296e-06, | |
| "loss": 0.6049, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.92680015008357e-06, | |
| "loss": 0.5982, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 7.884162772452844e-06, | |
| "loss": 0.5807, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.841525394822117e-06, | |
| "loss": 0.5809, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 7.798973291946653e-06, | |
| "loss": 0.5984, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 7.756335914315927e-06, | |
| "loss": 0.6308, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 7.7136985366852e-06, | |
| "loss": 0.5964, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 7.671061159054473e-06, | |
| "loss": 0.5783, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 7.62850905617901e-06, | |
| "loss": 0.5914, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 7.585871678548283e-06, | |
| "loss": 0.5716, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.543319575672818e-06, | |
| "loss": 0.613, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.500682198042092e-06, | |
| "loss": 0.561, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 7.458044820411367e-06, | |
| "loss": 0.5731, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 7.41540744278064e-06, | |
| "loss": 0.565, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.372855339905175e-06, | |
| "loss": 0.6118, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_BLEU": 39.5916929052258, | |
| "eval_BLEU-Bigram-Precision": 44.30874450415242, | |
| "eval_BLEU-Trigram-Precision": 34.66957832891385, | |
| "eval_BLEU-Unigram-Precision": 61.68977749113189, | |
| "eval_ROUGE-2": 33.461963127391705, | |
| "eval_ROUGE-L": 46.65653806735692, | |
| "eval_Sacre-Bigram-Precision": 42.369437106279214, | |
| "eval_Sacre-Trigram-Precision": 34.54220198406245, | |
| "eval_Sacre-Unigram-Precision": 59.68165957650208, | |
| "eval_SacreBLEU": 38.27100461755832, | |
| "eval_loss": 0.7759082317352295, | |
| "eval_runtime": 173.3948, | |
| "eval_samples_per_second": 7.134, | |
| "eval_steps_per_second": 7.134, | |
| "step": 37032 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.33030323702971e-06, | |
| "loss": 0.4871, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 7.287665859398984e-06, | |
| "loss": 0.4991, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 7.245028481768258e-06, | |
| "loss": 0.5436, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 7.202391104137532e-06, | |
| "loss": 0.5165, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 7.159753726506805e-06, | |
| "loss": 0.516, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 7.11711634887608e-06, | |
| "loss": 0.5173, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 7.074478971245353e-06, | |
| "loss": 0.5239, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 7.031841593614626e-06, | |
| "loss": 0.5112, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 6.9892042159839e-06, | |
| "loss": 0.516, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 6.9465668383531745e-06, | |
| "loss": 0.5384, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 6.903929460722449e-06, | |
| "loss": 0.5194, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 6.861292083091722e-06, | |
| "loss": 0.5006, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 6.818739980216257e-06, | |
| "loss": 0.5362, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.776102602585531e-06, | |
| "loss": 0.5505, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 6.733465224954805e-06, | |
| "loss": 0.5247, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 6.6908278473240785e-06, | |
| "loss": 0.5104, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 6.648275744448614e-06, | |
| "loss": 0.516, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 6.605638366817888e-06, | |
| "loss": 0.5097, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 6.563000989187162e-06, | |
| "loss": 0.5147, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 6.520363611556435e-06, | |
| "loss": 0.4988, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 6.47781150868097e-06, | |
| "loss": 0.5143, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 6.435174131050244e-06, | |
| "loss": 0.5146, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 6.3925367534195185e-06, | |
| "loss": 0.4949, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.349899375788792e-06, | |
| "loss": 0.5327, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_BLEU": 41.583977153720554, | |
| "eval_BLEU-Bigram-Precision": 47.79210232463709, | |
| "eval_BLEU-Trigram-Precision": 38.44570809639859, | |
| "eval_BLEU-Unigram-Precision": 64.70223028797936, | |
| "eval_ROUGE-2": 35.92233849645701, | |
| "eval_ROUGE-L": 48.89087193511837, | |
| "eval_Sacre-Bigram-Precision": 46.017043349388665, | |
| "eval_Sacre-Trigram-Precision": 38.29859537834164, | |
| "eval_Sacre-Unigram-Precision": 62.97069051167412, | |
| "eval_SacreBLEU": 40.202754965999326, | |
| "eval_loss": 0.734529972076416, | |
| "eval_runtime": 168.3041, | |
| "eval_samples_per_second": 7.35, | |
| "eval_steps_per_second": 7.35, | |
| "step": 49376 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 6.307347272913327e-06, | |
| "loss": 0.515, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.264709895282601e-06, | |
| "loss": 0.4603, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 6.222072517651875e-06, | |
| "loss": 0.4565, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 6.179435140021148e-06, | |
| "loss": 0.4557, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 6.136797762390423e-06, | |
| "loss": 0.4586, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 6.094245659514958e-06, | |
| "loss": 0.4295, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 6.051608281884232e-06, | |
| "loss": 0.4774, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 6.009056179008766e-06, | |
| "loss": 0.4555, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 5.966418801378041e-06, | |
| "loss": 0.4448, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 5.923781423747314e-06, | |
| "loss": 0.4481, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 5.881144046116588e-06, | |
| "loss": 0.4288, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 5.838506668485862e-06, | |
| "loss": 0.4459, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 5.7958692908551366e-06, | |
| "loss": 0.4616, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 5.75323191322441e-06, | |
| "loss": 0.4775, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 5.710594535593683e-06, | |
| "loss": 0.4414, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 5.668042432718218e-06, | |
| "loss": 0.483, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 5.625405055087493e-06, | |
| "loss": 0.4671, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 5.5827676774567664e-06, | |
| "loss": 0.4696, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 5.54013029982604e-06, | |
| "loss": 0.4628, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 5.497578196950575e-06, | |
| "loss": 0.4591, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 5.45494081931985e-06, | |
| "loss": 0.4362, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.412303441689123e-06, | |
| "loss": 0.4767, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 5.369666064058396e-06, | |
| "loss": 0.4642, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.3271139611829315e-06, | |
| "loss": 0.4348, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 5.2844765835522064e-06, | |
| "loss": 0.4365, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_BLEU": 43.53907049710156, | |
| "eval_BLEU-Bigram-Precision": 49.90573248407643, | |
| "eval_BLEU-Trigram-Precision": 40.825538394605175, | |
| "eval_BLEU-Unigram-Precision": 66.37906241012367, | |
| "eval_ROUGE-2": 38.22782051808487, | |
| "eval_ROUGE-L": 51.36307879612767, | |
| "eval_Sacre-Bigram-Precision": 48.25423818887177, | |
| "eval_Sacre-Trigram-Precision": 40.784982935153586, | |
| "eval_Sacre-Unigram-Precision": 64.79006681958711, | |
| "eval_SacreBLEU": 42.18614400441328, | |
| "eval_loss": 0.7037733197212219, | |
| "eval_runtime": 168.6409, | |
| "eval_samples_per_second": 7.335, | |
| "eval_steps_per_second": 7.335, | |
| "step": 61720 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 5.241924480676741e-06, | |
| "loss": 0.4345, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 5.199372377801276e-06, | |
| "loss": 0.4124, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 5.156735000170549e-06, | |
| "loss": 0.3992, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 5.114097622539824e-06, | |
| "loss": 0.3928, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 5.0714602449090974e-06, | |
| "loss": 0.4428, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 5.0288228672783715e-06, | |
| "loss": 0.4003, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 4.986185489647646e-06, | |
| "loss": 0.4077, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 4.943548112016919e-06, | |
| "loss": 0.4076, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 4.900910734386193e-06, | |
| "loss": 0.4138, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 4.858273356755466e-06, | |
| "loss": 0.3901, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 4.81563597912474e-06, | |
| "loss": 0.4412, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 4.7729986014940135e-06, | |
| "loss": 0.4128, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 4.730361223863288e-06, | |
| "loss": 0.4382, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 4.687809120987823e-06, | |
| "loss": 0.3855, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 4.645171743357097e-06, | |
| "loss": 0.4116, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 4.60253436572637e-06, | |
| "loss": 0.3977, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 4.559896988095644e-06, | |
| "loss": 0.4232, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 4.517259610464918e-06, | |
| "loss": 0.4194, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 4.4746222328341924e-06, | |
| "loss": 0.389, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 4.4319848552034665e-06, | |
| "loss": 0.4051, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 4.38934747757274e-06, | |
| "loss": 0.4337, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 4.346795374697275e-06, | |
| "loss": 0.4143, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 4.304157997066549e-06, | |
| "loss": 0.4146, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 4.261520619435823e-06, | |
| "loss": 0.3878, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 4.218883241805096e-06, | |
| "loss": 0.3955, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_BLEU": 46.18598569567297, | |
| "eval_BLEU-Bigram-Precision": 51.835734293717486, | |
| "eval_BLEU-Trigram-Precision": 42.84724073580379, | |
| "eval_BLEU-Unigram-Precision": 67.89297658862876, | |
| "eval_ROUGE-2": 40.77288879803008, | |
| "eval_ROUGE-L": 53.47443573476642, | |
| "eval_Sacre-Bigram-Precision": 50.16359387172163, | |
| "eval_Sacre-Trigram-Precision": 42.76279276279276, | |
| "eval_Sacre-Unigram-Precision": 66.28928362287722, | |
| "eval_SacreBLEU": 44.9656942267261, | |
| "eval_loss": 0.6645232439041138, | |
| "eval_runtime": 169.8377, | |
| "eval_samples_per_second": 7.283, | |
| "eval_steps_per_second": 7.283, | |
| "step": 74064 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 4.176331138929632e-06, | |
| "loss": 0.3766, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 4.133693761298906e-06, | |
| "loss": 0.3556, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 4.09105638366818e-06, | |
| "loss": 0.3511, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 4.048504280792714e-06, | |
| "loss": 0.4021, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 4.005952177917249e-06, | |
| "loss": 0.3931, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 3.9633148002865234e-06, | |
| "loss": 0.3704, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 3.9206774226557975e-06, | |
| "loss": 0.3515, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 3.878040045025071e-06, | |
| "loss": 0.3907, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 3.835402667394345e-06, | |
| "loss": 0.3643, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 3.7927652897636185e-06, | |
| "loss": 0.3708, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 3.7501279121328926e-06, | |
| "loss": 0.3697, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 3.7074905345021663e-06, | |
| "loss": 0.3486, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 3.6648531568714404e-06, | |
| "loss": 0.3552, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 3.6222157792407136e-06, | |
| "loss": 0.3812, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 3.5795784016099877e-06, | |
| "loss": 0.3622, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 3.5369410239792614e-06, | |
| "loss": 0.391, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 3.494388921103797e-06, | |
| "loss": 0.4055, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 6.72, | |
| "learning_rate": 3.4517515434730702e-06, | |
| "loss": 0.3967, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 3.4091141658423443e-06, | |
| "loss": 0.3949, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 3.3664767882116184e-06, | |
| "loss": 0.3898, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 3.3239246853361536e-06, | |
| "loss": 0.3997, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 3.281287307705427e-06, | |
| "loss": 0.4075, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 3.238649930074701e-06, | |
| "loss": 0.3704, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 3.196012552443975e-06, | |
| "loss": 0.3809, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_BLEU": 48.53041153339425, | |
| "eval_BLEU-Bigram-Precision": 52.751822317890515, | |
| "eval_BLEU-Trigram-Precision": 44.54280358258696, | |
| "eval_BLEU-Unigram-Precision": 67.82913552910786, | |
| "eval_ROUGE-2": 42.93136082972705, | |
| "eval_ROUGE-L": 54.935588774300534, | |
| "eval_Sacre-Bigram-Precision": 51.321117635148816, | |
| "eval_Sacre-Trigram-Precision": 44.494843134078515, | |
| "eval_Sacre-Unigram-Precision": 66.2363645024532, | |
| "eval_SacreBLEU": 47.488510994186036, | |
| "eval_loss": 0.6420064568519592, | |
| "eval_runtime": 175.4955, | |
| "eval_samples_per_second": 7.049, | |
| "eval_steps_per_second": 7.049, | |
| "step": 86408 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 3.1534604495685103e-06, | |
| "loss": 0.3682, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 3.110908346693045e-06, | |
| "loss": 0.3759, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 3.068270969062319e-06, | |
| "loss": 0.3607, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 3.025633591431593e-06, | |
| "loss": 0.3478, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 2.982996213800867e-06, | |
| "loss": 0.3474, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 2.94035883617014e-06, | |
| "loss": 0.3824, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 2.8977214585394142e-06, | |
| "loss": 0.3469, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 2.8550840809086883e-06, | |
| "loss": 0.3599, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 2.8124467032779616e-06, | |
| "loss": 0.3332, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 2.7698093256472356e-06, | |
| "loss": 0.3508, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 2.727257222771771e-06, | |
| "loss": 0.3382, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 2.684619845141045e-06, | |
| "loss": 0.3583, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 2.641982467510318e-06, | |
| "loss": 0.3392, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 2.5993450898795923e-06, | |
| "loss": 0.3548, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 2.556707712248866e-06, | |
| "loss": 0.3372, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 2.5141556093734016e-06, | |
| "loss": 0.3386, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 2.471518231742675e-06, | |
| "loss": 0.362, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 2.428880854111949e-06, | |
| "loss": 0.3454, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 2.3862434764812226e-06, | |
| "loss": 0.3757, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 2.3436913736057578e-06, | |
| "loss": 0.3785, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 2.3010539959750314e-06, | |
| "loss": 0.3656, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 2.2584166183443055e-06, | |
| "loss": 0.36, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 2.2157792407135796e-06, | |
| "loss": 0.3522, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 2.1732271378381144e-06, | |
| "loss": 0.3511, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 7.98, | |
| "learning_rate": 2.130589760207388e-06, | |
| "loss": 0.3349, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_BLEU": 49.57199736076417, | |
| "eval_BLEU-Bigram-Precision": 55.58697298661954, | |
| "eval_BLEU-Trigram-Precision": 47.143087942269375, | |
| "eval_BLEU-Unigram-Precision": 70.68244463454045, | |
| "eval_ROUGE-2": 44.4865922111988, | |
| "eval_ROUGE-L": 56.71133192128113, | |
| "eval_Sacre-Bigram-Precision": 53.98119122257053, | |
| "eval_Sacre-Trigram-Precision": 47.064737753449144, | |
| "eval_Sacre-Unigram-Precision": 69.11714187564411, | |
| "eval_SacreBLEU": 48.47735653922784, | |
| "eval_loss": 0.6274014711380005, | |
| "eval_runtime": 169.1165, | |
| "eval_samples_per_second": 7.314, | |
| "eval_steps_per_second": 7.314, | |
| "step": 98752 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.087952382576662e-06, | |
| "loss": 0.3635, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 2.0453150049459363e-06, | |
| "loss": 0.3149, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 2.00267762731521e-06, | |
| "loss": 0.3348, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 1.9600402496844836e-06, | |
| "loss": 0.3358, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 1.9174028720537573e-06, | |
| "loss": 0.3405, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 1.8747654944230311e-06, | |
| "loss": 0.328, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 1.8322133915475666e-06, | |
| "loss": 0.3487, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 1.7895760139168402e-06, | |
| "loss": 0.3012, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 1.746938636286114e-06, | |
| "loss": 0.3456, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 1.7043012586553878e-06, | |
| "loss": 0.3472, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 1.6617491557799232e-06, | |
| "loss": 0.3476, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 1.6191117781491968e-06, | |
| "loss": 0.3262, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 8.51, | |
| "learning_rate": 1.5765596752737323e-06, | |
| "loss": 0.3607, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 1.533922297643006e-06, | |
| "loss": 0.341, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 1.4912849200122798e-06, | |
| "loss": 0.3164, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 1.4486475423815535e-06, | |
| "loss": 0.354, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 1.4060101647508274e-06, | |
| "loss": 0.3282, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 1.363372787120101e-06, | |
| "loss": 0.3419, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 1.320735409489375e-06, | |
| "loss": 0.3524, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 1.2780980318586486e-06, | |
| "loss": 0.3372, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 1.2354606542279224e-06, | |
| "loss": 0.358, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 1.1929085513524577e-06, | |
| "loss": 0.3426, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 1.150356448476993e-06, | |
| "loss": 0.3449, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 1.1077190708462667e-06, | |
| "loss": 0.3547, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 1.0650816932155406e-06, | |
| "loss": 0.347, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_BLEU": 50.42765483523289, | |
| "eval_BLEU-Bigram-Precision": 55.31231468669697, | |
| "eval_BLEU-Trigram-Precision": 47.11826938259908, | |
| "eval_BLEU-Unigram-Precision": 69.82722488706749, | |
| "eval_ROUGE-2": 44.880113409125535, | |
| "eval_ROUGE-L": 56.82064108206701, | |
| "eval_Sacre-Bigram-Precision": 53.78361795684631, | |
| "eval_Sacre-Trigram-Precision": 47.04437530702472, | |
| "eval_Sacre-Unigram-Precision": 68.36258716037509, | |
| "eval_SacreBLEU": 49.36901330124516, | |
| "eval_loss": 0.6147586703300476, | |
| "eval_runtime": 169.9129, | |
| "eval_samples_per_second": 7.28, | |
| "eval_steps_per_second": 7.28, | |
| "step": 111096 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 1.0224443155848143e-06, | |
| "loss": 0.3398, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 9.798069379540882e-07, | |
| "loss": 0.3255, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 9.37169560323362e-07, | |
| "loss": 0.3178, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 8.946174574478971e-07, | |
| "loss": 0.3281, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 8.520653545724324e-07, | |
| "loss": 0.3261, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 8.094279769417062e-07, | |
| "loss": 0.3275, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 7.6679059931098e-07, | |
| "loss": 0.3359, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 7.241532216802538e-07, | |
| "loss": 0.3165, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 6.815158440495275e-07, | |
| "loss": 0.3199, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 6.388784664188015e-07, | |
| "loss": 0.332, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 5.962410887880752e-07, | |
| "loss": 0.3227, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 5.536037111573491e-07, | |
| "loss": 0.3203, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 9.52, | |
| "learning_rate": 5.109663335266228e-07, | |
| "loss": 0.347, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 4.6832895589589667e-07, | |
| "loss": 0.3485, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 4.2569157826517044e-07, | |
| "loss": 0.3244, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 3.830542006344442e-07, | |
| "loss": 0.3351, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 3.4041682300371804e-07, | |
| "loss": 0.3341, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 2.9786472012825325e-07, | |
| "loss": 0.328, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 2.5522734249752707e-07, | |
| "loss": 0.3447, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 2.1258996486680084e-07, | |
| "loss": 0.3326, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 1.6995258723607467e-07, | |
| "loss": 0.3259, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 1.274004843606099e-07, | |
| "loss": 0.3355, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 8.47631067298837e-08, | |
| "loss": 0.3653, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 4.212572909915749e-08, | |
| "loss": 0.3359, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_BLEU": 50.8713809636162, | |
| "eval_BLEU-Bigram-Precision": 55.25529935869192, | |
| "eval_BLEU-Trigram-Precision": 47.15997915581032, | |
| "eval_BLEU-Unigram-Precision": 69.56702363367799, | |
| "eval_ROUGE-2": 45.59251780348448, | |
| "eval_ROUGE-L": 57.14390377702154, | |
| "eval_Sacre-Bigram-Precision": 53.736780853109344, | |
| "eval_Sacre-Trigram-Precision": 47.052790672568285, | |
| "eval_Sacre-Unigram-Precision": 68.06666666666666, | |
| "eval_SacreBLEU": 49.82287256259556, | |
| "eval_loss": 0.6102247834205627, | |
| "eval_runtime": 172.6687, | |
| "eval_samples_per_second": 7.164, | |
| "eval_steps_per_second": 7.164, | |
| "step": 123440 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 123440, | |
| "total_flos": 3771355808563200.0, | |
| "train_loss": 0.4878676912032082, | |
| "train_runtime": 11854.8374, | |
| "train_samples_per_second": 20.824, | |
| "train_steps_per_second": 10.413 | |
| } | |
| ], | |
| "max_steps": 123440, | |
| "num_train_epochs": 10, | |
| "total_flos": 3771355808563200.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |