| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 19.789564097058193, | |
| "eval_steps": 500, | |
| "global_step": 1440, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0001, | |
| "loss": 7.5882, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_bleu": 0.0872, | |
| "eval_bp": 0.6461, | |
| "eval_counts_1": 3993, | |
| "eval_counts_2": 105, | |
| "eval_counts_3": 0, | |
| "eval_counts_4": 0, | |
| "eval_exact_match": 0.0, | |
| "eval_f1": 0.1155, | |
| "eval_gen_len": 9.7105, | |
| "eval_loss": 5.682333469390869, | |
| "eval_precisions_1": 26.998, | |
| "eval_precisions_2": 0.8343, | |
| "eval_precisions_3": 0.0048, | |
| "eval_precisions_4": 0.0031, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.1101, | |
| "eval_rouge2": 0.0077, | |
| "eval_rougeL": 0.1078, | |
| "eval_rougeLsum": 0.1076, | |
| "eval_runtime": 1951.3051, | |
| "eval_samples_per_second": 1.13, | |
| "eval_steps_per_second": 0.565, | |
| "eval_sys_len": 14790, | |
| "eval_totals_1": 14790, | |
| "eval_totals_2": 12586, | |
| "eval_totals_3": 10382, | |
| "eval_totals_4": 8178, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.0001, | |
| "loss": 5.2903, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_bleu": 0.351, | |
| "eval_bp": 0.8828, | |
| "eval_counts_1": 3827, | |
| "eval_counts_2": 229, | |
| "eval_counts_3": 32, | |
| "eval_counts_4": 0, | |
| "eval_exact_match": 0.0, | |
| "eval_f1": 0.0964, | |
| "eval_gen_len": 16.7005, | |
| "eval_loss": 4.872079372406006, | |
| "eval_precisions_1": 20.2551, | |
| "eval_precisions_2": 1.3721, | |
| "eval_precisions_3": 0.2209, | |
| "eval_precisions_4": 0.0041, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.0924, | |
| "eval_rouge2": 0.015, | |
| "eval_rougeL": 0.091, | |
| "eval_rougeLsum": 0.0909, | |
| "eval_runtime": 3438.1674, | |
| "eval_samples_per_second": 0.641, | |
| "eval_steps_per_second": 0.321, | |
| "eval_sys_len": 18894, | |
| "eval_totals_1": 18894, | |
| "eval_totals_2": 16690, | |
| "eval_totals_3": 14486, | |
| "eval_totals_4": 12282, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.0001, | |
| "loss": 4.6636, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bleu": 0.2933, | |
| "eval_bp": 0.6758, | |
| "eval_counts_1": 3638, | |
| "eval_counts_2": 174, | |
| "eval_counts_3": 21, | |
| "eval_counts_4": 0, | |
| "eval_exact_match": 0.0, | |
| "eval_f1": 0.0925, | |
| "eval_gen_len": 8.9197, | |
| "eval_loss": 4.280586242675781, | |
| "eval_precisions_1": 23.8276, | |
| "eval_precisions_2": 1.3319, | |
| "eval_precisions_3": 0.1934, | |
| "eval_precisions_4": 0.0058, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.0884, | |
| "eval_rouge2": 0.012, | |
| "eval_rougeL": 0.0876, | |
| "eval_rougeLsum": 0.0874, | |
| "eval_runtime": 2326.5895, | |
| "eval_samples_per_second": 0.947, | |
| "eval_steps_per_second": 0.474, | |
| "eval_sys_len": 15268, | |
| "eval_totals_1": 15268, | |
| "eval_totals_2": 13064, | |
| "eval_totals_3": 10860, | |
| "eval_totals_4": 8656, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.0001, | |
| "loss": 4.2229, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bleu": 0.2288, | |
| "eval_bp": 1.0, | |
| "eval_counts_1": 4274, | |
| "eval_counts_2": 240, | |
| "eval_counts_3": 24, | |
| "eval_counts_4": 0, | |
| "eval_exact_match": 0.0, | |
| "eval_f1": 0.1023, | |
| "eval_gen_len": 24.7015, | |
| "eval_loss": 3.9210410118103027, | |
| "eval_precisions_1": 14.583, | |
| "eval_precisions_2": 0.8855, | |
| "eval_precisions_3": 0.0964, | |
| "eval_precisions_4": 0.0022, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.0894, | |
| "eval_rouge2": 0.0109, | |
| "eval_rougeL": 0.0849, | |
| "eval_rougeLsum": 0.0849, | |
| "eval_runtime": 2975.0462, | |
| "eval_samples_per_second": 0.741, | |
| "eval_steps_per_second": 0.37, | |
| "eval_sys_len": 29308, | |
| "eval_totals_1": 29308, | |
| "eval_totals_2": 27104, | |
| "eval_totals_3": 24900, | |
| "eval_totals_4": 22696, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 0.0001, | |
| "loss": 3.9434, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "eval_bleu": 0.4204, | |
| "eval_bp": 0.7465, | |
| "eval_counts_1": 3652, | |
| "eval_counts_2": 218, | |
| "eval_counts_3": 35, | |
| "eval_counts_4": 1, | |
| "eval_exact_match": 0.0, | |
| "eval_f1": 0.0898, | |
| "eval_gen_len": 12.3049, | |
| "eval_loss": 3.690653085708618, | |
| "eval_precisions_1": 22.2114, | |
| "eval_precisions_2": 1.5311, | |
| "eval_precisions_3": 0.2908, | |
| "eval_precisions_4": 0.0102, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.0856, | |
| "eval_rouge2": 0.0141, | |
| "eval_rougeL": 0.0843, | |
| "eval_rougeLsum": 0.0842, | |
| "eval_runtime": 3036.8902, | |
| "eval_samples_per_second": 0.726, | |
| "eval_steps_per_second": 0.363, | |
| "eval_sys_len": 16442, | |
| "eval_totals_1": 16442, | |
| "eval_totals_2": 14238, | |
| "eval_totals_3": 12034, | |
| "eval_totals_4": 9830, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 0.0001, | |
| "loss": 3.6152, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "eval_bleu": 1.0505, | |
| "eval_bp": 0.968, | |
| "eval_counts_1": 4103, | |
| "eval_counts_2": 341, | |
| "eval_counts_3": 77, | |
| "eval_counts_4": 11, | |
| "eval_exact_match": 0.0, | |
| "eval_f1": 0.112, | |
| "eval_gen_len": 14.3607, | |
| "eval_loss": 3.460298538208008, | |
| "eval_precisions_1": 19.9359, | |
| "eval_precisions_2": 1.8556, | |
| "eval_precisions_3": 0.4761, | |
| "eval_precisions_4": 0.0787, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.107, | |
| "eval_rouge2": 0.019, | |
| "eval_rougeL": 0.1023, | |
| "eval_rougeLsum": 0.1024, | |
| "eval_runtime": 3225.717, | |
| "eval_samples_per_second": 0.683, | |
| "eval_steps_per_second": 0.342, | |
| "eval_sys_len": 20581, | |
| "eval_totals_1": 20581, | |
| "eval_totals_2": 18377, | |
| "eval_totals_3": 16173, | |
| "eval_totals_4": 13969, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 0.0001, | |
| "loss": 3.3814, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_bleu": 2.3489, | |
| "eval_bp": 0.8218, | |
| "eval_counts_1": 4342, | |
| "eval_counts_2": 675, | |
| "eval_counts_3": 218, | |
| "eval_counts_4": 43, | |
| "eval_exact_match": 0.0005, | |
| "eval_f1": 0.1308, | |
| "eval_gen_len": 10.2418, | |
| "eval_loss": 3.2883455753326416, | |
| "eval_precisions_1": 24.4441, | |
| "eval_precisions_2": 4.3383, | |
| "eval_precisions_3": 1.6323, | |
| "eval_precisions_4": 0.3856, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.1264, | |
| "eval_rouge2": 0.0353, | |
| "eval_rougeL": 0.1234, | |
| "eval_rougeLsum": 0.1234, | |
| "eval_runtime": 2402.3288, | |
| "eval_samples_per_second": 0.917, | |
| "eval_steps_per_second": 0.459, | |
| "eval_sys_len": 17763, | |
| "eval_totals_1": 17763, | |
| "eval_totals_2": 15559, | |
| "eval_totals_3": 13355, | |
| "eval_totals_4": 11151, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 0.0001, | |
| "loss": 3.1711, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_bleu": 2.6207, | |
| "eval_bp": 0.9273, | |
| "eval_counts_1": 4820, | |
| "eval_counts_2": 856, | |
| "eval_counts_3": 246, | |
| "eval_counts_4": 44, | |
| "eval_exact_match": 0.0005, | |
| "eval_f1": 0.1547, | |
| "eval_gen_len": 14.3249, | |
| "eval_loss": 3.0987935066223145, | |
| "eval_precisions_1": 24.3939, | |
| "eval_precisions_2": 4.8761, | |
| "eval_precisions_3": 1.6025, | |
| "eval_precisions_4": 0.3347, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.1503, | |
| "eval_rouge2": 0.0465, | |
| "eval_rougeL": 0.1455, | |
| "eval_rougeLsum": 0.1457, | |
| "eval_runtime": 2969.3248, | |
| "eval_samples_per_second": 0.742, | |
| "eval_steps_per_second": 0.371, | |
| "eval_sys_len": 19759, | |
| "eval_totals_1": 19759, | |
| "eval_totals_2": 17555, | |
| "eval_totals_3": 15351, | |
| "eval_totals_4": 13147, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 0.0001, | |
| "loss": 3.0147, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "eval_bleu": 3.4764, | |
| "eval_bp": 0.8739, | |
| "eval_counts_1": 5167, | |
| "eval_counts_2": 1066, | |
| "eval_counts_3": 321, | |
| "eval_counts_4": 76, | |
| "eval_exact_match": 0.0018, | |
| "eval_f1": 0.1816, | |
| "eval_gen_len": 14.3067, | |
| "eval_loss": 2.9539589881896973, | |
| "eval_precisions_1": 27.5941, | |
| "eval_precisions_2": 6.4524, | |
| "eval_precisions_3": 2.2421, | |
| "eval_precisions_4": 0.6274, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.1773, | |
| "eval_rouge2": 0.0588, | |
| "eval_rougeL": 0.1721, | |
| "eval_rougeLsum": 0.1721, | |
| "eval_runtime": 2825.1201, | |
| "eval_samples_per_second": 0.78, | |
| "eval_steps_per_second": 0.39, | |
| "eval_sys_len": 18725, | |
| "eval_totals_1": 18725, | |
| "eval_totals_2": 16521, | |
| "eval_totals_3": 14317, | |
| "eval_totals_4": 12113, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 9.99, | |
| "learning_rate": 0.0001, | |
| "loss": 2.7829, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 9.99, | |
| "eval_bleu": 4.5099, | |
| "eval_bp": 0.7974, | |
| "eval_counts_1": 5625, | |
| "eval_counts_2": 1267, | |
| "eval_counts_3": 420, | |
| "eval_counts_4": 124, | |
| "eval_exact_match": 0.0045, | |
| "eval_f1": 0.2159, | |
| "eval_gen_len": 12.9741, | |
| "eval_loss": 2.8288302421569824, | |
| "eval_precisions_1": 32.4638, | |
| "eval_precisions_2": 8.378, | |
| "eval_precisions_3": 3.251, | |
| "eval_precisions_4": 1.1573, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.2127, | |
| "eval_rouge2": 0.0741, | |
| "eval_rougeL": 0.2067, | |
| "eval_rougeLsum": 0.2065, | |
| "eval_runtime": 2709.6941, | |
| "eval_samples_per_second": 0.813, | |
| "eval_steps_per_second": 0.407, | |
| "eval_sys_len": 17327, | |
| "eval_totals_1": 17327, | |
| "eval_totals_2": 15123, | |
| "eval_totals_3": 12919, | |
| "eval_totals_4": 10715, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 10.99, | |
| "learning_rate": 0.0001, | |
| "loss": 2.6093, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 10.99, | |
| "eval_bleu": 5.5051, | |
| "eval_bp": 0.8685, | |
| "eval_counts_1": 6005, | |
| "eval_counts_2": 1469, | |
| "eval_counts_3": 528, | |
| "eval_counts_4": 181, | |
| "eval_exact_match": 0.0064, | |
| "eval_f1": 0.231, | |
| "eval_gen_len": 14.4791, | |
| "eval_loss": 2.7177300453186035, | |
| "eval_precisions_1": 32.2416, | |
| "eval_precisions_2": 8.9459, | |
| "eval_precisions_3": 3.7139, | |
| "eval_precisions_4": 1.5067, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.229, | |
| "eval_rouge2": 0.0827, | |
| "eval_rougeL": 0.2215, | |
| "eval_rougeLsum": 0.2213, | |
| "eval_runtime": 1457.0803, | |
| "eval_samples_per_second": 1.513, | |
| "eval_steps_per_second": 0.756, | |
| "eval_sys_len": 18625, | |
| "eval_totals_1": 18625, | |
| "eval_totals_2": 16421, | |
| "eval_totals_3": 14217, | |
| "eval_totals_4": 12013, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 0.0001, | |
| "loss": 2.453, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_bleu": 6.6865, | |
| "eval_bp": 0.8515, | |
| "eval_counts_1": 6396, | |
| "eval_counts_2": 1744, | |
| "eval_counts_3": 664, | |
| "eval_counts_4": 246, | |
| "eval_exact_match": 0.0059, | |
| "eval_f1": 0.2565, | |
| "eval_gen_len": 13.7142, | |
| "eval_loss": 2.591360330581665, | |
| "eval_precisions_1": 34.9375, | |
| "eval_precisions_2": 10.8303, | |
| "eval_precisions_3": 4.7773, | |
| "eval_precisions_4": 2.1035, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.2553, | |
| "eval_rouge2": 0.0998, | |
| "eval_rougeL": 0.2479, | |
| "eval_rougeLsum": 0.2478, | |
| "eval_runtime": 1377.6536, | |
| "eval_samples_per_second": 1.6, | |
| "eval_steps_per_second": 0.8, | |
| "eval_sys_len": 18307, | |
| "eval_totals_1": 18307, | |
| "eval_totals_2": 16103, | |
| "eval_totals_3": 13899, | |
| "eval_totals_4": 11695, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 12.99, | |
| "learning_rate": 0.0001, | |
| "loss": 2.3329, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 12.99, | |
| "eval_bleu": 7.383, | |
| "eval_bp": 0.8592, | |
| "eval_counts_1": 6673, | |
| "eval_counts_2": 1888, | |
| "eval_counts_3": 741, | |
| "eval_counts_4": 291, | |
| "eval_exact_match": 0.0091, | |
| "eval_f1": 0.2749, | |
| "eval_gen_len": 14.1751, | |
| "eval_loss": 2.499257802963257, | |
| "eval_precisions_1": 36.1661, | |
| "eval_precisions_2": 11.6206, | |
| "eval_precisions_3": 5.2767, | |
| "eval_precisions_4": 2.458, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.2747, | |
| "eval_rouge2": 0.1114, | |
| "eval_rougeL": 0.2652, | |
| "eval_rougeLsum": 0.2652, | |
| "eval_runtime": 1427.0765, | |
| "eval_samples_per_second": 1.544, | |
| "eval_steps_per_second": 0.772, | |
| "eval_sys_len": 18451, | |
| "eval_totals_1": 18451, | |
| "eval_totals_2": 16247, | |
| "eval_totals_3": 14043, | |
| "eval_totals_4": 11839, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 13.99, | |
| "learning_rate": 0.0001, | |
| "loss": 2.1663, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 13.99, | |
| "eval_bleu": 8.1343, | |
| "eval_bp": 0.8635, | |
| "eval_counts_1": 6953, | |
| "eval_counts_2": 2052, | |
| "eval_counts_3": 834, | |
| "eval_counts_4": 337, | |
| "eval_exact_match": 0.0082, | |
| "eval_f1": 0.2889, | |
| "eval_gen_len": 14.6783, | |
| "eval_loss": 2.4196276664733887, | |
| "eval_precisions_1": 37.5209, | |
| "eval_precisions_2": 12.5681, | |
| "eval_precisions_3": 5.9053, | |
| "eval_precisions_4": 2.8274, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.2886, | |
| "eval_rouge2": 0.1215, | |
| "eval_rougeL": 0.2773, | |
| "eval_rougeLsum": 0.277, | |
| "eval_runtime": 1443.0194, | |
| "eval_samples_per_second": 1.527, | |
| "eval_steps_per_second": 0.764, | |
| "eval_sys_len": 18531, | |
| "eval_totals_1": 18531, | |
| "eval_totals_2": 16327, | |
| "eval_totals_3": 14123, | |
| "eval_totals_4": 11919, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 14.99, | |
| "learning_rate": 0.0001, | |
| "loss": 2.0422, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 14.99, | |
| "eval_bleu": 8.4322, | |
| "eval_bp": 0.8339, | |
| "eval_counts_1": 6968, | |
| "eval_counts_2": 2089, | |
| "eval_counts_3": 862, | |
| "eval_counts_4": 365, | |
| "eval_exact_match": 0.0113, | |
| "eval_f1": 0.2951, | |
| "eval_gen_len": 13.6987, | |
| "eval_loss": 2.3703055381774902, | |
| "eval_precisions_1": 38.7456, | |
| "eval_precisions_2": 13.2383, | |
| "eval_precisions_3": 6.3494, | |
| "eval_precisions_4": 3.2096, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.2961, | |
| "eval_rouge2": 0.1268, | |
| "eval_rougeL": 0.2858, | |
| "eval_rougeLsum": 0.2857, | |
| "eval_runtime": 1381.8523, | |
| "eval_samples_per_second": 1.595, | |
| "eval_steps_per_second": 0.797, | |
| "eval_sys_len": 17984, | |
| "eval_totals_1": 17984, | |
| "eval_totals_2": 15780, | |
| "eval_totals_3": 13576, | |
| "eval_totals_4": 11372, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 0.0001, | |
| "loss": 1.9245, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_bleu": 9.5973, | |
| "eval_bp": 0.8892, | |
| "eval_counts_1": 7500, | |
| "eval_counts_2": 2353, | |
| "eval_counts_3": 999, | |
| "eval_counts_4": 446, | |
| "eval_exact_match": 0.0132, | |
| "eval_f1": 0.314, | |
| "eval_gen_len": 14.77, | |
| "eval_loss": 2.3217406272888184, | |
| "eval_precisions_1": 39.4384, | |
| "eval_precisions_2": 13.9951, | |
| "eval_precisions_3": 6.8383, | |
| "eval_precisions_4": 3.5953, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.3149, | |
| "eval_rouge2": 0.1407, | |
| "eval_rougeL": 0.3017, | |
| "eval_rougeLsum": 0.3017, | |
| "eval_runtime": 1430.5793, | |
| "eval_samples_per_second": 1.541, | |
| "eval_steps_per_second": 0.77, | |
| "eval_sys_len": 19017, | |
| "eval_totals_1": 19017, | |
| "eval_totals_2": 16813, | |
| "eval_totals_3": 14609, | |
| "eval_totals_4": 12405, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 0.0001, | |
| "loss": 1.8216, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_bleu": 9.9557, | |
| "eval_bp": 0.8467, | |
| "eval_counts_1": 7444, | |
| "eval_counts_2": 2357, | |
| "eval_counts_3": 1044, | |
| "eval_counts_4": 488, | |
| "eval_exact_match": 0.0132, | |
| "eval_f1": 0.3181, | |
| "eval_gen_len": 13.8031, | |
| "eval_loss": 2.27047061920166, | |
| "eval_precisions_1": 40.8584, | |
| "eval_precisions_2": 14.7175, | |
| "eval_precisions_3": 7.5592, | |
| "eval_precisions_4": 4.2044, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.3201, | |
| "eval_rouge2": 0.1437, | |
| "eval_rougeL": 0.3081, | |
| "eval_rougeLsum": 0.3077, | |
| "eval_runtime": 1357.6078, | |
| "eval_samples_per_second": 1.623, | |
| "eval_steps_per_second": 0.812, | |
| "eval_sys_len": 18219, | |
| "eval_totals_1": 18219, | |
| "eval_totals_2": 16015, | |
| "eval_totals_3": 13811, | |
| "eval_totals_4": 11607, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7503, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "eval_bleu": 10.4354, | |
| "eval_bp": 0.8498, | |
| "eval_counts_1": 7571, | |
| "eval_counts_2": 2487, | |
| "eval_counts_3": 1114, | |
| "eval_counts_4": 515, | |
| "eval_exact_match": 0.0145, | |
| "eval_f1": 0.3265, | |
| "eval_gen_len": 13.9106, | |
| "eval_loss": 2.238603353500366, | |
| "eval_precisions_1": 41.4282, | |
| "eval_precisions_2": 15.4751, | |
| "eval_precisions_3": 8.0335, | |
| "eval_precisions_4": 4.4157, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.3289, | |
| "eval_rouge2": 0.1512, | |
| "eval_rougeL": 0.3153, | |
| "eval_rougeLsum": 0.3151, | |
| "eval_runtime": 1353.0462, | |
| "eval_samples_per_second": 1.629, | |
| "eval_steps_per_second": 0.814, | |
| "eval_sys_len": 18275, | |
| "eval_totals_1": 18275, | |
| "eval_totals_2": 16071, | |
| "eval_totals_3": 13867, | |
| "eval_totals_4": 11663, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 18.99, | |
| "learning_rate": 0.0001, | |
| "loss": 1.6342, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 18.99, | |
| "eval_bleu": 10.7447, | |
| "eval_bp": 0.8418, | |
| "eval_counts_1": 7697, | |
| "eval_counts_2": 2536, | |
| "eval_counts_3": 1155, | |
| "eval_counts_4": 537, | |
| "eval_exact_match": 0.0177, | |
| "eval_f1": 0.3313, | |
| "eval_gen_len": 13.8494, | |
| "eval_loss": 2.2182679176330566, | |
| "eval_precisions_1": 42.4568, | |
| "eval_precisions_2": 15.9246, | |
| "eval_precisions_3": 8.4178, | |
| "eval_precisions_4": 4.6627, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.3342, | |
| "eval_rouge2": 0.1559, | |
| "eval_rougeL": 0.3224, | |
| "eval_rougeLsum": 0.3222, | |
| "eval_runtime": 1333.2607, | |
| "eval_samples_per_second": 1.653, | |
| "eval_steps_per_second": 0.827, | |
| "eval_sys_len": 18129, | |
| "eval_totals_1": 18129, | |
| "eval_totals_2": 15925, | |
| "eval_totals_3": 13721, | |
| "eval_totals_4": 11517, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 19.79, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5474, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 19.79, | |
| "eval_bleu": 11.1066, | |
| "eval_bp": 0.8786, | |
| "eval_counts_1": 7879, | |
| "eval_counts_2": 2632, | |
| "eval_counts_3": 1187, | |
| "eval_counts_4": 570, | |
| "eval_exact_match": 0.0177, | |
| "eval_f1": 0.3375, | |
| "eval_gen_len": 14.5136, | |
| "eval_loss": 2.1956045627593994, | |
| "eval_precisions_1": 41.8762, | |
| "eval_precisions_2": 15.8449, | |
| "eval_precisions_3": 8.2391, | |
| "eval_precisions_4": 4.671, | |
| "eval_ref_len": 21250, | |
| "eval_rouge1": 0.3398, | |
| "eval_rouge2": 0.1607, | |
| "eval_rougeL": 0.326, | |
| "eval_rougeLsum": 0.326, | |
| "eval_runtime": 1394.5803, | |
| "eval_samples_per_second": 1.58, | |
| "eval_steps_per_second": 0.79, | |
| "eval_sys_len": 18815, | |
| "eval_totals_1": 18815, | |
| "eval_totals_2": 16611, | |
| "eval_totals_3": 14407, | |
| "eval_totals_4": 12203, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 19.79, | |
| "step": 1440, | |
| "total_flos": 8.496574887886848e+17, | |
| "train_loss": 3.111723126305474, | |
| "train_runtime": 93678.1212, | |
| "train_samples_per_second": 1.989, | |
| "train_steps_per_second": 0.015 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 1440, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 8.496574887886848e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |