{ "batch_size": 1, "best_eval_metric_checkpoint_number": 4, "best_eval_metric_epoch": 3, "best_eval_metric_steps": 2800, "best_eval_metric_value": 1.3660465478897095, "best_eval_test_metrics": { "combined": { "loss": 1.4971884489059448 }, "word": { "bleu": 0.0, "char_error_rate": 2.4850947856903076, "loss": 1.4971884489059448, "next_token_perplexity": 16168.8603515625, "perplexity": 28562.080078125, "rouge1_fmeasure": 0.054999999701976776, "rouge1_precision": 0.03666666895151138, "rouge1_recall": 0.10999999940395355, "rouge2_fmeasure": 0.0, "rouge2_precision": 0.0, "rouge2_recall": 0.0, "rougeL_fmeasure": 0.054999999701976776, "rougeL_precision": 0.03666666895151138, "rougeL_recall": 0.10999999940395355, "rougeLsum_fmeasure": 0.054999999701976776, "rougeLsum_precision": 0.03666666895151138, "rougeLsum_recall": 0.10999999940395355, "sequence_accuracy": 0.0, "token_accuracy": 0.0, "word_error_rate": 2.880000114440918 } }, "best_eval_train_metrics": { "combined": { "loss": 1.2540030479431152 }, "word": { "bleu": 0.0, "char_error_rate": 2.419161558151245, "loss": 1.2540030479431152, "next_token_perplexity": 15712.271484375, "perplexity": 28806.5546875, "rouge1_fmeasure": 0.12121212482452393, "rouge1_precision": 0.0833333358168602, "rouge1_recall": 0.22727273404598236, "rouge2_fmeasure": 0.0, "rouge2_precision": 0.0, "rouge2_recall": 0.0, "rougeL_fmeasure": 0.12121212482452393, "rougeL_precision": 0.0833333358168602, "rougeL_recall": 0.22727273404598236, "rougeLsum_fmeasure": 0.12121212482452393, "rougeLsum_precision": 0.0833333358168602, "rougeLsum_recall": 0.22727273404598236, "sequence_accuracy": 0.0, "token_accuracy": 0.0, "word_error_rate": 2.590909004211426 } }, "best_eval_validation_metrics": { "combined": { "loss": 1.3660465478897095 }, "word": { "bleu": 0.0, "char_error_rate": 2.5209367275238037, "loss": 1.3660465478897095, "next_token_perplexity": 16073.732421875, "perplexity": 28901.515625, "rouge1_fmeasure": 0.07750000059604645, "rouge1_precision": 0.05166666582226753, "rouge1_recall": 0.1550000011920929, "rouge2_fmeasure": 0.0, "rouge2_precision": 0.0, "rouge2_recall": 0.0, "rougeL_fmeasure": 0.07750000059604645, "rougeL_precision": 0.05166666582226753, "rougeL_recall": 0.1550000011920929, "rougeLsum_fmeasure": 0.07750000059604645, "rougeLsum_precision": 0.05166666582226753, "rougeLsum_recall": 0.1550000011920929, "sequence_accuracy": 0.0, "token_accuracy": 0.0, "word_error_rate": 2.8299999237060547 } }, "best_increase_batch_size_eval_metric": Infinity, "checkpoint_number": 5, "epoch": 5, "last_improvement_steps": 700, "last_increase_batch_size": 0, "last_increase_batch_size_eval_metric_improvement": 0, "last_increase_batch_size_steps": 0, "last_learning_rate_reduction": 0, "last_learning_rate_reduction_steps": 0, "learning_rate": 1.1103994690380681e-05, "num_increases_batch_size": 0, "num_reductions_learning_rate": 0, "steps": 3500, "test_metrics": { "combined": { "loss": [ [ 0, 700, 2.1768786907196045 ], [ 1, 1400, 1.6330327987670898 ], [ 2, 2100, 1.5191197395324707 ], [ 3, 2800, 1.4971884489059448 ], [ 4, 3500, 1.5195976495742798 ] ] }, "word": { "bleu": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "char_error_rate": [ [ 0, 700, 2.1924118995666504 ], [ 1, 1400, 1.803523063659668 ], [ 2, 2100, 2.697831869125366 ], [ 3, 2800, 2.4850947856903076 ], [ 4, 3500, 2.531165361404419 ] ], "loss": [ [ 0, 700, 2.1768786907196045 ], [ 1, 1400, 1.6330327987670898 ], [ 2, 2100, 1.5191197395324707 ], [ 3, 2800, 1.4971884489059448 ], [ 4, 3500, 1.5195976495742798 ] ], "next_token_perplexity": [ [ 0, 700, 16999.853515625 ], [ 1, 1400, 16360.1416015625 ], [ 2, 2100, 16165.0322265625 ], [ 3, 2800, 16168.8603515625 ], [ 4, 3500, 16030.0927734375 ] ], "perplexity": [ [ 0, 700, 28644.078125 ], [ 1, 1400, 26310.689453125 ], [ 2, 2100, 29572.255859375 ], [ 3, 2800, 28562.080078125 ], [ 4, 3500, 26435.541015625 ] ], "rouge1_fmeasure": [ [ 0, 700, 0.004999999888241291 ], [ 1, 1400, 0.04500000178813934 ], [ 2, 2100, 0.058000002056360245 ], [ 3, 2800, 0.054999999701976776 ], [ 4, 3500, 0.0806666687130928 ] ], "rouge1_precision": [ [ 0, 700, 0.0033333334140479565 ], [ 1, 1400, 0.03333333134651184 ], [ 2, 2100, 0.038333334028720856 ], [ 3, 2800, 0.03666666895151138 ], [ 4, 3500, 0.05750000476837158 ] ], "rouge1_recall": [ [ 0, 700, 0.009999999776482582 ], [ 1, 1400, 0.07000000029802322 ], [ 2, 2100, 0.11999999731779099 ], [ 3, 2800, 0.10999999940395355 ], [ 4, 3500, 0.14000000059604645 ] ], "rouge2_fmeasure": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rouge2_precision": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rouge2_recall": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rougeL_fmeasure": [ [ 0, 700, 0.004999999888241291 ], [ 1, 1400, 0.04500000178813934 ], [ 2, 2100, 0.058000002056360245 ], [ 3, 2800, 0.054999999701976776 ], [ 4, 3500, 0.0806666687130928 ] ], "rougeL_precision": [ [ 0, 700, 0.0033333334140479565 ], [ 1, 1400, 0.03333333134651184 ], [ 2, 2100, 0.038333334028720856 ], [ 3, 2800, 0.03666666895151138 ], [ 4, 3500, 0.05750000476837158 ] ], "rougeL_recall": [ [ 0, 700, 0.009999999776482582 ], [ 1, 1400, 0.07000000029802322 ], [ 2, 2100, 0.11999999731779099 ], [ 3, 2800, 0.10999999940395355 ], [ 4, 3500, 0.14000000059604645 ] ], "rougeLsum_fmeasure": [ [ 0, 700, 0.004999999888241291 ], [ 1, 1400, 0.04500000178813934 ], [ 2, 2100, 0.058000002056360245 ], [ 3, 2800, 0.054999999701976776 ], [ 4, 3500, 0.0806666687130928 ] ], "rougeLsum_precision": [ [ 0, 700, 0.0033333334140479565 ], [ 1, 1400, 0.03333333134651184 ], [ 2, 2100, 0.038333334028720856 ], [ 3, 2800, 0.03666666895151138 ], [ 4, 3500, 0.05750000476837158 ] ], "rougeLsum_recall": [ [ 0, 700, 0.009999999776482582 ], [ 1, 1400, 0.07000000029802322 ], [ 2, 2100, 0.11999999731779099 ], [ 3, 2800, 0.10999999940395355 ], [ 4, 3500, 0.14000000059604645 ] ], "sequence_accuracy": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "token_accuracy": [ [ 0, 700, 0.00326680694706738 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "word_error_rate": [ [ 0, 700, 3.6700000762939453 ], [ 1, 1400, 2.190000057220459 ], [ 2, 2100, 2.9700000286102295 ], [ 3, 2800, 2.880000114440918 ], [ 4, 3500, 2.619999885559082 ] ] } }, "train_metrics": { "combined": { "loss": [ [ 0, 700, 4.781581401824951 ], [ 1, 1400, 1.7923258543014526 ], [ 2, 2100, 1.4238678216934204 ], [ 3, 2800, 1.2540030479431152 ], [ 4, 3500, 1.0201646089553833 ] ] }, "word": { "bleu": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "char_error_rate": [ [ 0, 700, 2.5638630390167236 ], [ 1, 1400, 2.706451654434204 ], [ 2, 2100, 2.662576675415039 ], [ 3, 2800, 2.419161558151245 ], [ 4, 3500, 2.712538242340088 ] ], "loss": [ [ 0, 700, 4.781581401824951 ], [ 1, 1400, 1.7923258543014526 ], [ 2, 2100, 1.4238678216934204 ], [ 3, 2800, 1.2540030479431152 ], [ 4, 3500, 1.0201646089553833 ] ], "next_token_perplexity": [ [ 0, 700, 21023.48046875 ], [ 1, 1400, 16821.6953125 ], [ 2, 2100, 16183.7099609375 ], [ 3, 2800, 15712.271484375 ], [ 4, 3500, 15665.1103515625 ] ], "perplexity": [ [ 0, 700, 28891.017578125 ], [ 1, 1400, 27432.072265625 ], [ 2, 2100, 28224.69140625 ], [ 3, 2800, 28806.5546875 ], [ 4, 3500, 27271.728515625 ] ], "rouge1_fmeasure": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.07467532902956009 ], [ 2, 2100, 0.03030303120613098 ], [ 3, 2800, 0.12121212482452393 ], [ 4, 3500, 0.1401515156030655 ] ], "rouge1_precision": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.053030308336019516 ], [ 2, 2100, 0.019696969538927078 ], [ 3, 2800, 0.0833333358168602 ], [ 4, 3500, 0.09469697624444962 ] ], "rouge1_recall": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.13636364042758942 ], [ 2, 2100, 0.06818182021379471 ], [ 3, 2800, 0.22727273404598236 ], [ 4, 3500, 0.27272728085517883 ] ], "rouge2_fmeasure": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rouge2_precision": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rouge2_recall": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rougeL_fmeasure": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.07467532902956009 ], [ 2, 2100, 0.03030303120613098 ], [ 3, 2800, 0.12121212482452393 ], [ 4, 3500, 0.1401515156030655 ] ], "rougeL_precision": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.053030308336019516 ], [ 2, 2100, 0.019696969538927078 ], [ 3, 2800, 0.0833333358168602 ], [ 4, 3500, 0.09469697624444962 ] ], "rougeL_recall": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.13636364042758942 ], [ 2, 2100, 0.06818182021379471 ], [ 3, 2800, 0.22727273404598236 ], [ 4, 3500, 0.27272728085517883 ] ], "rougeLsum_fmeasure": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.07467532902956009 ], [ 2, 2100, 0.03030303120613098 ], [ 3, 2800, 0.12121212482452393 ], [ 4, 3500, 0.1401515156030655 ] ], "rougeLsum_precision": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.053030308336019516 ], [ 2, 2100, 0.019696969538927078 ], [ 3, 2800, 0.0833333358168602 ], [ 4, 3500, 0.09469697624444962 ] ], "rougeLsum_recall": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.13636364042758942 ], [ 2, 2100, 0.06818182021379471 ], [ 3, 2800, 0.22727273404598236 ], [ 4, 3500, 0.27272728085517883 ] ], "sequence_accuracy": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "token_accuracy": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "word_error_rate": [ [ 0, 700, 3.5227272510528564 ], [ 1, 1400, 3.4318182468414307 ], [ 2, 2100, 3.159090995788574 ], [ 3, 2800, 2.590909004211426 ], [ 4, 3500, 2.590909004211426 ] ] } }, "tune_checkpoint_num": 0, "validation_metrics": { "combined": { "loss": [ [ 0, 700, 2.016148567199707 ], [ 1, 1400, 1.5244907140731812 ], [ 2, 2100, 1.4093765020370483 ], [ 3, 2800, 1.3660465478897095 ], [ 4, 3500, 1.404098629951477 ] ] }, "word": { "bleu": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "char_error_rate": [ [ 0, 700, 2.315826892852783 ], [ 1, 1400, 1.851667881011963 ], [ 2, 2100, 2.8899929523468018 ], [ 3, 2800, 2.5209367275238037 ], [ 4, 3500, 2.6408801078796387 ] ], "loss": [ [ 0, 700, 2.016148567199707 ], [ 1, 1400, 1.5244907140731812 ], [ 2, 2100, 1.4093765020370483 ], [ 3, 2800, 1.3660465478897095 ], [ 4, 3500, 1.404098629951477 ] ], "next_token_perplexity": [ [ 0, 700, 17030.87109375 ], [ 1, 1400, 16436.427734375 ], [ 2, 2100, 16175.5751953125 ], [ 3, 2800, 16073.732421875 ], [ 4, 3500, 16021.54296875 ] ], "perplexity": [ [ 0, 700, 28752.59375 ], [ 1, 1400, 26383.328125 ], [ 2, 2100, 29749.330078125 ], [ 3, 2800, 28901.515625 ], [ 4, 3500, 26440.935546875 ] ], "rouge1_fmeasure": [ [ 0, 700, 0.020999999716877937 ], [ 1, 1400, 0.07999999821186066 ], [ 2, 2100, 0.06599999219179153 ], [ 3, 2800, 0.07750000059604645 ], [ 4, 3500, 0.07000000774860382 ] ], "rouge1_precision": [ [ 0, 700, 0.013749999925494194 ], [ 1, 1400, 0.05999999865889549 ], [ 2, 2100, 0.04375000298023224 ], [ 3, 2800, 0.05166666582226753 ], [ 4, 3500, 0.04833333194255829 ] ], "rouge1_recall": [ [ 0, 700, 0.04500000178813934 ], [ 1, 1400, 0.11999999731779099 ], [ 2, 2100, 0.13500000536441803 ], [ 3, 2800, 0.1550000011920929 ], [ 4, 3500, 0.12999999523162842 ] ], "rouge2_fmeasure": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rouge2_precision": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rouge2_recall": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "rougeL_fmeasure": [ [ 0, 700, 0.020999999716877937 ], [ 1, 1400, 0.07999999821186066 ], [ 2, 2100, 0.06599999219179153 ], [ 3, 2800, 0.07750000059604645 ], [ 4, 3500, 0.07000000774860382 ] ], "rougeL_precision": [ [ 0, 700, 0.013749999925494194 ], [ 1, 1400, 0.05999999865889549 ], [ 2, 2100, 0.04375000298023224 ], [ 3, 2800, 0.05166666582226753 ], [ 4, 3500, 0.04833333194255829 ] ], "rougeL_recall": [ [ 0, 700, 0.04500000178813934 ], [ 1, 1400, 0.11999999731779099 ], [ 2, 2100, 0.13500000536441803 ], [ 3, 2800, 0.1550000011920929 ], [ 4, 3500, 0.12999999523162842 ] ], "rougeLsum_fmeasure": [ [ 0, 700, 0.020999999716877937 ], [ 1, 1400, 0.07999999821186066 ], [ 2, 2100, 0.06599999219179153 ], [ 3, 2800, 0.07750000059604645 ], [ 4, 3500, 0.07000000774860382 ] ], "rougeLsum_precision": [ [ 0, 700, 0.013749999925494194 ], [ 1, 1400, 0.05999999865889549 ], [ 2, 2100, 0.04375000298023224 ], [ 3, 2800, 0.05166666582226753 ], [ 4, 3500, 0.04833333194255829 ] ], "rougeLsum_recall": [ [ 0, 700, 0.04500000178813934 ], [ 1, 1400, 0.11999999731779099 ], [ 2, 2100, 0.13500000536441803 ], [ 3, 2800, 0.1550000011920929 ], [ 4, 3500, 0.12999999523162842 ] ], "sequence_accuracy": [ [ 0, 700, 0.0 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "token_accuracy": [ [ 0, 700, 0.0013417367590591311 ], [ 1, 1400, 0.0 ], [ 2, 2100, 0.0 ], [ 3, 2800, 0.0 ], [ 4, 3500, 0.0 ] ], "word_error_rate": [ [ 0, 700, 3.5350000858306885 ], [ 1, 1400, 2.0950000286102295 ], [ 2, 2100, 2.944999933242798 ], [ 3, 2800, 2.8299999237060547 ], [ 4, 3500, 2.640000104904175 ] ] } } }