| { | |
| "best_metric": 0.06889855116605759, | |
| "best_model_checkpoint": "./models/1/checkpoint-9737", | |
| "epoch": 10.0, | |
| "global_step": 13910, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.591954022988506e-05, | |
| "loss": 0.0876, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.884970485848343e-05, | |
| "loss": 0.0879, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_BLEU": 8.064725756490624, | |
| "eval_BLEU-Bigram-Precision": 10.432874977383753, | |
| "eval_BLEU-Trigram-Precision": 5.407291933269737, | |
| "eval_BLEU-Unigram-Precision": 21.631135458605595, | |
| "eval_ROUGE-2": 12.39926519031995, | |
| "eval_ROUGE-L": 24.731962013373813, | |
| "eval_Sacre-Bigram-Precision": 8.552870639534884, | |
| "eval_Sacre-Trigram-Precision": 4.851033999298983, | |
| "eval_Sacre-Unigram-Precision": 19.295765314011796, | |
| "eval_SacreBLEU": 7.1292786955103695, | |
| "eval_loss": 0.0844602882862091, | |
| "eval_runtime": 523.4808, | |
| "eval_samples_per_second": 2.363, | |
| "eval_steps_per_second": 0.296, | |
| "step": 1391 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.6961555925533526e-05, | |
| "loss": 0.0829, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.506962312698653e-05, | |
| "loss": 0.0747, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.317769032843954e-05, | |
| "loss": 0.0738, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_BLEU": 7.726645595861129, | |
| "eval_BLEU-Bigram-Precision": 9.779983147645352, | |
| "eval_BLEU-Trigram-Precision": 5.376859377396105, | |
| "eval_BLEU-Unigram-Precision": 19.00918707221551, | |
| "eval_ROUGE-2": 14.567403936228773, | |
| "eval_ROUGE-L": 26.962145889007623, | |
| "eval_Sacre-Bigram-Precision": 8.340425531914894, | |
| "eval_Sacre-Trigram-Precision": 5.06410008133545, | |
| "eval_Sacre-Unigram-Precision": 17.33626552335896, | |
| "eval_SacreBLEU": 7.08253441311445, | |
| "eval_loss": 0.07733357697725296, | |
| "eval_runtime": 594.316, | |
| "eval_samples_per_second": 2.081, | |
| "eval_steps_per_second": 0.261, | |
| "step": 2782 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 4.1285757529892535e-05, | |
| "loss": 0.0669, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.939382473134555e-05, | |
| "loss": 0.0631, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.7505675798395645e-05, | |
| "loss": 0.0616, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_BLEU": 13.75944614514078, | |
| "eval_BLEU-Bigram-Precision": 17.05563909774436, | |
| "eval_BLEU-Trigram-Precision": 9.895979758223222, | |
| "eval_BLEU-Unigram-Precision": 31.226259170122074, | |
| "eval_ROUGE-2": 16.762154637582928, | |
| "eval_ROUGE-L": 29.733231785857146, | |
| "eval_Sacre-Bigram-Precision": 14.784847370356749, | |
| "eval_Sacre-Trigram-Precision": 9.445382434455736, | |
| "eval_Sacre-Unigram-Precision": 28.888232688616565, | |
| "eval_SacreBLEU": 12.80566461172721, | |
| "eval_loss": 0.07288952171802521, | |
| "eval_runtime": 656.3315, | |
| "eval_samples_per_second": 1.885, | |
| "eval_steps_per_second": 0.236, | |
| "step": 4173 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.561374299984865e-05, | |
| "loss": 0.0556, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 3.372181020130165e-05, | |
| "loss": 0.0522, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.1829877402754655e-05, | |
| "loss": 0.0541, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_BLEU": 12.584990506585802, | |
| "eval_BLEU-Bigram-Precision": 15.284490474363894, | |
| "eval_BLEU-Trigram-Precision": 9.189817259955635, | |
| "eval_BLEU-Unigram-Precision": 27.802290416935204, | |
| "eval_ROUGE-2": 18.2427503756888, | |
| "eval_ROUGE-L": 31.248445460832464, | |
| "eval_Sacre-Bigram-Precision": 13.375350140056023, | |
| "eval_Sacre-Trigram-Precision": 8.759613065730592, | |
| "eval_Sacre-Unigram-Precision": 25.743221169552434, | |
| "eval_SacreBLEU": 11.731554660360798, | |
| "eval_loss": 0.07118678092956543, | |
| "eval_runtime": 463.4231, | |
| "eval_samples_per_second": 2.669, | |
| "eval_steps_per_second": 0.334, | |
| "step": 5564 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 2.9937944604207657e-05, | |
| "loss": 0.0464, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.8046011805660665e-05, | |
| "loss": 0.0459, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_BLEU": 15.1776190124192, | |
| "eval_BLEU-Bigram-Precision": 18.32776890695311, | |
| "eval_BLEU-Trigram-Precision": 11.247289809684414, | |
| "eval_BLEU-Unigram-Precision": 32.42811501597444, | |
| "eval_ROUGE-2": 17.936785005872903, | |
| "eval_ROUGE-L": 31.307371141735135, | |
| "eval_Sacre-Bigram-Precision": 16.26334519572954, | |
| "eval_Sacre-Trigram-Precision": 10.753317119724164, | |
| "eval_Sacre-Unigram-Precision": 30.265755070515205, | |
| "eval_SacreBLEU": 14.242434580808037, | |
| "eval_loss": 0.06986960768699646, | |
| "eval_runtime": 398.2357, | |
| "eval_samples_per_second": 3.106, | |
| "eval_steps_per_second": 0.389, | |
| "step": 6955 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.6154079007113667e-05, | |
| "loss": 0.0455, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 2.4262146208566673e-05, | |
| "loss": 0.0393, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 2.2370213410019678e-05, | |
| "loss": 0.0405, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_BLEU": 17.03804435280673, | |
| "eval_BLEU-Bigram-Precision": 20.16791440608614, | |
| "eval_BLEU-Trigram-Precision": 12.76924045849603, | |
| "eval_BLEU-Unigram-Precision": 34.79696172947707, | |
| "eval_ROUGE-2": 20.00404092356461, | |
| "eval_ROUGE-L": 33.44410880996549, | |
| "eval_Sacre-Bigram-Precision": 18.140892297184017, | |
| "eval_Sacre-Trigram-Precision": 12.388609066253391, | |
| "eval_Sacre-Unigram-Precision": 32.66758356754171, | |
| "eval_SacreBLEU": 16.183245622226913, | |
| "eval_loss": 0.06919773668050766, | |
| "eval_runtime": 389.7744, | |
| "eval_samples_per_second": 3.174, | |
| "eval_steps_per_second": 0.398, | |
| "step": 8346 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 2.047828061147268e-05, | |
| "loss": 0.0395, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 1.8586347812925686e-05, | |
| "loss": 0.0351, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 1.669441501437869e-05, | |
| "loss": 0.0356, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_BLEU": 18.45635959837559, | |
| "eval_BLEU-Bigram-Precision": 21.599746072052056, | |
| "eval_BLEU-Trigram-Precision": 13.98176291793313, | |
| "eval_BLEU-Unigram-Precision": 36.946429662207564, | |
| "eval_ROUGE-2": 20.663766964343406, | |
| "eval_ROUGE-L": 33.85196384000605, | |
| "eval_Sacre-Bigram-Precision": 19.43390720747009, | |
| "eval_Sacre-Trigram-Precision": 13.581706410862664, | |
| "eval_Sacre-Unigram-Precision": 34.678927680798004, | |
| "eval_SacreBLEU": 17.534804534318486, | |
| "eval_loss": 0.06889855116605759, | |
| "eval_runtime": 289.9351, | |
| "eval_samples_per_second": 4.266, | |
| "eval_steps_per_second": 0.535, | |
| "step": 9737 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 1.4802482215831695e-05, | |
| "loss": 0.0324, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 1.2910549417284698e-05, | |
| "loss": 0.032, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 7.91, | |
| "learning_rate": 1.1018616618737702e-05, | |
| "loss": 0.0316, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_BLEU": 23.036610051092996, | |
| "eval_BLEU-Bigram-Precision": 26.881516587677723, | |
| "eval_BLEU-Trigram-Precision": 17.706261436868488, | |
| "eval_BLEU-Unigram-Precision": 44.332174416920175, | |
| "eval_ROUGE-2": 21.29088469404566, | |
| "eval_ROUGE-L": 35.036834725440904, | |
| "eval_Sacre-Bigram-Precision": 24.360966513868387, | |
| "eval_Sacre-Trigram-Precision": 17.249540910018364, | |
| "eval_Sacre-Unigram-Precision": 41.78435079135624, | |
| "eval_SacreBLEU": 21.988824762940034, | |
| "eval_loss": 0.0689924880862236, | |
| "eval_runtime": 259.2001, | |
| "eval_samples_per_second": 4.772, | |
| "eval_steps_per_second": 0.598, | |
| "step": 11128 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 9.126683820190707e-06, | |
| "loss": 0.0295, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 7.238534887240806e-06, | |
| "loss": 0.0283, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 5.3466020886938095e-06, | |
| "loss": 0.0287, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_BLEU": 26.995788994612766, | |
| "eval_BLEU-Bigram-Precision": 30.950083340643914, | |
| "eval_BLEU-Trigram-Precision": 20.98232920550995, | |
| "eval_BLEU-Unigram-Precision": 50.522155190347405, | |
| "eval_ROUGE-2": 22.105956241069418, | |
| "eval_ROUGE-L": 35.85099346344104, | |
| "eval_Sacre-Bigram-Precision": 28.175123119324088, | |
| "eval_Sacre-Trigram-Precision": 20.573315467075037, | |
| "eval_Sacre-Unigram-Precision": 47.89045785194694, | |
| "eval_SacreBLEU": 25.924953054404842, | |
| "eval_loss": 0.06922509521245956, | |
| "eval_runtime": 198.6398, | |
| "eval_samples_per_second": 6.227, | |
| "eval_steps_per_second": 0.78, | |
| "step": 12519 | |
| }, | |
| { | |
| "epoch": 9.35, | |
| "learning_rate": 3.454669290146814e-06, | |
| "loss": 0.0271, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 1.5627364915998186e-06, | |
| "loss": 0.0265, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_BLEU": 26.712396231914138, | |
| "eval_BLEU-Bigram-Precision": 30.61851211072664, | |
| "eval_BLEU-Trigram-Precision": 20.719279806242287, | |
| "eval_BLEU-Unigram-Precision": 49.895307303855155, | |
| "eval_ROUGE-2": 22.458649124626863, | |
| "eval_ROUGE-L": 36.256812107261126, | |
| "eval_Sacre-Bigram-Precision": 27.962698554345884, | |
| "eval_Sacre-Trigram-Precision": 20.387249114521843, | |
| "eval_Sacre-Unigram-Precision": 47.333925324538036, | |
| "eval_SacreBLEU": 25.70732578879903, | |
| "eval_loss": 0.06933248043060303, | |
| "eval_runtime": 190.4819, | |
| "eval_samples_per_second": 6.494, | |
| "eval_steps_per_second": 0.814, | |
| "step": 13910 | |
| } | |
| ], | |
| "max_steps": 13910, | |
| "num_train_epochs": 10, | |
| "total_flos": 3.016954478592e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |