| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 48.63813229571984, | |
| "eval_steps": 500, | |
| "global_step": 12500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_bp": 0.4559528592458481, | |
| "eval_counts": [ | |
| 3210, | |
| 1930, | |
| 1488, | |
| 1066 | |
| ], | |
| "eval_loss": 2.412109375, | |
| "eval_precisions": [ | |
| 71.17516629711751, | |
| 48.44377510040161, | |
| 43.03065355696935, | |
| 36.24617477048623 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 118.4312, | |
| "eval_samples_per_second": 4.441, | |
| "eval_score": 21.9569286964753, | |
| "eval_steps_per_second": 0.279, | |
| "eval_sys_len": 4510, | |
| "eval_totals": [ | |
| 4510, | |
| 3984, | |
| 3458, | |
| 2941 | |
| ], | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.9221789883268484e-05, | |
| "loss": 2.8948, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bp": 0.43532380297987505, | |
| "eval_counts": [ | |
| 1708, | |
| 541, | |
| 376, | |
| 228 | |
| ], | |
| "eval_loss": 1.5029296875, | |
| "eval_precisions": [ | |
| 38.853503184713375, | |
| 13.979328165374676, | |
| 11.24401913875598, | |
| 7.497533706017757 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 115.6212, | |
| "eval_samples_per_second": 4.549, | |
| "eval_score": 6.3679777301051494, | |
| "eval_steps_per_second": 0.285, | |
| "eval_sys_len": 4396, | |
| "eval_totals": [ | |
| 4396, | |
| 3870, | |
| 3344, | |
| 3041 | |
| ], | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bp": 0.5596896112039585, | |
| "eval_counts": [ | |
| 2795, | |
| 1858, | |
| 1416, | |
| 991 | |
| ], | |
| "eval_loss": 0.65576171875, | |
| "eval_precisions": [ | |
| 54.85770363101079, | |
| 40.66535346903042, | |
| 35.02349740291862, | |
| 28.177423940858688 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 115.5065, | |
| "eval_samples_per_second": 4.554, | |
| "eval_score": 21.558969055160425, | |
| "eval_steps_per_second": 0.286, | |
| "eval_sys_len": 5095, | |
| "eval_totals": [ | |
| 5095, | |
| 4569, | |
| 4043, | |
| 3517 | |
| ], | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.8443579766536967e-05, | |
| "loss": 0.8924, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bp": 0.6650198090145658, | |
| "eval_counts": [ | |
| 3257, | |
| 2161, | |
| 1704, | |
| 1256 | |
| ], | |
| "eval_loss": 0.485107421875, | |
| "eval_precisions": [ | |
| 56.950515824444835, | |
| 41.61371076449066, | |
| 36.51167773730448, | |
| 30.330837961844964 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 115.5544, | |
| "eval_samples_per_second": 4.552, | |
| "eval_score": 26.766843398496384, | |
| "eval_steps_per_second": 0.286, | |
| "eval_sys_len": 5719, | |
| "eval_totals": [ | |
| 5719, | |
| 5193, | |
| 4667, | |
| 4141 | |
| ], | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bp": 0.41788118238391686, | |
| "eval_counts": [ | |
| 3699, | |
| 2841, | |
| 2368, | |
| 1900 | |
| ], | |
| "eval_loss": 0.293701171875, | |
| "eval_precisions": [ | |
| 86.02325581395348, | |
| 75.27821939586646, | |
| 72.9064039408867, | |
| 69.80161645848641 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 115.1699, | |
| "eval_samples_per_second": 4.567, | |
| "eval_score": 31.661530724736487, | |
| "eval_steps_per_second": 0.287, | |
| "eval_sys_len": 4300, | |
| "eval_totals": [ | |
| 4300, | |
| 3774, | |
| 3248, | |
| 2722 | |
| ], | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 1.766536964980545e-05, | |
| "loss": 0.4295, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bp": 0.4089581075583404, | |
| "eval_counts": [ | |
| 3783, | |
| 2928, | |
| 2446, | |
| 1971 | |
| ], | |
| "eval_loss": 0.2445068359375, | |
| "eval_precisions": [ | |
| 88.9908256880734, | |
| 78.60402684563758, | |
| 76.46139418568302, | |
| 73.73737373737374 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 115.449, | |
| "eval_samples_per_second": 4.556, | |
| "eval_score": 32.408490251125635, | |
| "eval_steps_per_second": 0.286, | |
| "eval_sys_len": 4251, | |
| "eval_totals": [ | |
| 4251, | |
| 3725, | |
| 3199, | |
| 2673 | |
| ], | |
| "step": 1542 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_bp": 0.417153226107242, | |
| "eval_counts": [ | |
| 3818, | |
| 2965, | |
| 2480, | |
| 2002 | |
| ], | |
| "eval_loss": 0.22021484375, | |
| "eval_precisions": [ | |
| 88.8733705772812, | |
| 78.64721485411141, | |
| 76.44882860665845, | |
| 73.6571008094187 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 115.4073, | |
| "eval_samples_per_second": 4.558, | |
| "eval_score": 33.04119304311829, | |
| "eval_steps_per_second": 0.286, | |
| "eval_sys_len": 4296, | |
| "eval_totals": [ | |
| 4296, | |
| 3770, | |
| 3244, | |
| 2718 | |
| ], | |
| "step": 1799 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 1.6887159533073932e-05, | |
| "loss": 0.2991, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_bp": 0.42679131632296613, | |
| "eval_counts": [ | |
| 3874, | |
| 3019, | |
| 2524, | |
| 2038 | |
| ], | |
| "eval_loss": 0.2076416015625, | |
| "eval_precisions": [ | |
| 89.07794895378248, | |
| 78.9693957624902, | |
| 76.55444343342432, | |
| 73.54745579213281 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 114.0107, | |
| "eval_samples_per_second": 4.614, | |
| "eval_score": 33.85769159645968, | |
| "eval_steps_per_second": 0.289, | |
| "eval_sys_len": 4349, | |
| "eval_totals": [ | |
| 4349, | |
| 3823, | |
| 3297, | |
| 2771 | |
| ], | |
| "step": 2056 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_bp": 0.46028228872696303, | |
| "eval_counts": [ | |
| 4065, | |
| 3225, | |
| 2700, | |
| 2186 | |
| ], | |
| "eval_loss": 0.1663818359375, | |
| "eval_precisions": [ | |
| 89.65593295103662, | |
| 80.46407185628742, | |
| 77.54164273406089, | |
| 73.9512855209743 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 114.8195, | |
| "eval_samples_per_second": 4.581, | |
| "eval_score": 36.91388078489759, | |
| "eval_steps_per_second": 0.287, | |
| "eval_sys_len": 4534, | |
| "eval_totals": [ | |
| 4534, | |
| 4008, | |
| 3482, | |
| 2956 | |
| ], | |
| "step": 2313 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 1.6108949416342414e-05, | |
| "loss": 0.2277, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_bp": 0.5426087135017283, | |
| "eval_counts": [ | |
| 4419, | |
| 3611, | |
| 3062, | |
| 2525 | |
| ], | |
| "eval_loss": 0.1044921875, | |
| "eval_precisions": [ | |
| 88.43305983590155, | |
| 80.76492954596287, | |
| 77.617237008872, | |
| 73.85200350979818 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 114.7152, | |
| "eval_samples_per_second": 4.585, | |
| "eval_score": 43.40364936643555, | |
| "eval_steps_per_second": 0.288, | |
| "eval_sys_len": 4997, | |
| "eval_totals": [ | |
| 4997, | |
| 4471, | |
| 3945, | |
| 3419 | |
| ], | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_bp": 0.5841943959505824, | |
| "eval_counts": [ | |
| 4717, | |
| 3950, | |
| 3372, | |
| 2808 | |
| ], | |
| "eval_loss": 0.08892822265625, | |
| "eval_precisions": [ | |
| 90.07065113614665, | |
| 83.846317130121, | |
| 80.57347670250896, | |
| 76.74227931128723 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 113.7863, | |
| "eval_samples_per_second": 4.623, | |
| "eval_score": 48.29263576279789, | |
| "eval_steps_per_second": 0.29, | |
| "eval_sys_len": 5237, | |
| "eval_totals": [ | |
| 5237, | |
| 4711, | |
| 4185, | |
| 3659 | |
| ], | |
| "step": 2827 | |
| }, | |
| { | |
| "epoch": 11.67, | |
| "learning_rate": 1.5330739299610897e-05, | |
| "loss": 0.1405, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_bp": 0.5586477230994942, | |
| "eval_counts": [ | |
| 4630, | |
| 3875, | |
| 3303, | |
| 2749 | |
| ], | |
| "eval_loss": 0.08489990234375, | |
| "eval_precisions": [ | |
| 90.98054627628218, | |
| 84.9222003068157, | |
| 81.81818181818181, | |
| 78.29678154371973 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.8868, | |
| "eval_samples_per_second": 4.875, | |
| "eval_score": 46.85747814062412, | |
| "eval_steps_per_second": 0.306, | |
| "eval_sys_len": 5089, | |
| "eval_totals": [ | |
| 5089, | |
| 4563, | |
| 4037, | |
| 3511 | |
| ], | |
| "step": 3084 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_bp": 0.5695616786732568, | |
| "eval_counts": [ | |
| 4747, | |
| 4034, | |
| 3464, | |
| 2904 | |
| ], | |
| "eval_loss": 0.08123779296875, | |
| "eval_precisions": [ | |
| 92.1389751552795, | |
| 87.20276696930394, | |
| 84.48780487804878, | |
| 81.2534974818131 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 108.2759, | |
| "eval_samples_per_second": 4.858, | |
| "eval_score": 49.08436700451685, | |
| "eval_steps_per_second": 0.305, | |
| "eval_sys_len": 5152, | |
| "eval_totals": [ | |
| 5152, | |
| 4626, | |
| 4100, | |
| 3574 | |
| ], | |
| "step": 3341 | |
| }, | |
| { | |
| "epoch": 13.62, | |
| "learning_rate": 1.4552529182879378e-05, | |
| "loss": 0.1241, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_bp": 0.5695616786732568, | |
| "eval_counts": [ | |
| 4738, | |
| 4024, | |
| 3452, | |
| 2894 | |
| ], | |
| "eval_loss": 0.07525634765625, | |
| "eval_precisions": [ | |
| 91.96428571428571, | |
| 86.98659749243407, | |
| 84.1951219512195, | |
| 80.97369893676553 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.2784, | |
| "eval_samples_per_second": 4.903, | |
| "eval_score": 48.94590635934059, | |
| "eval_steps_per_second": 0.308, | |
| "eval_sys_len": 5152, | |
| "eval_totals": [ | |
| 5152, | |
| 4626, | |
| 4100, | |
| 3574 | |
| ], | |
| "step": 3598 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_bp": 0.5961628688829712, | |
| "eval_counts": [ | |
| 4741, | |
| 4006, | |
| 3444, | |
| 2891 | |
| ], | |
| "eval_loss": 0.07562255859375, | |
| "eval_precisions": [ | |
| 89.33484077633314, | |
| 83.79000209161264, | |
| 80.94007050528789, | |
| 77.52748726200053 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.2867, | |
| "eval_samples_per_second": 4.949, | |
| "eval_score": 49.354074470195435, | |
| "eval_steps_per_second": 0.31, | |
| "eval_sys_len": 5307, | |
| "eval_totals": [ | |
| 5307, | |
| 4781, | |
| 4255, | |
| 3729 | |
| ], | |
| "step": 3855 | |
| }, | |
| { | |
| "epoch": 15.56, | |
| "learning_rate": 1.377431906614786e-05, | |
| "loss": 0.1147, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_bp": 0.5655838797151567, | |
| "eval_counts": [ | |
| 4743, | |
| 4039, | |
| 3477, | |
| 2925 | |
| ], | |
| "eval_loss": 0.06915283203125, | |
| "eval_precisions": [ | |
| 92.47416650419186, | |
| 87.74712144253748, | |
| 85.28329654157469, | |
| 82.37116305266122 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.836, | |
| "eval_samples_per_second": 4.878, | |
| "eval_score": 49.143959340541095, | |
| "eval_steps_per_second": 0.306, | |
| "eval_sys_len": 5129, | |
| "eval_totals": [ | |
| 5129, | |
| 4603, | |
| 4077, | |
| 3551 | |
| ], | |
| "step": 4112 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_bp": 0.5932631592602093, | |
| "eval_counts": [ | |
| 4727, | |
| 3996, | |
| 3439, | |
| 2892 | |
| ], | |
| "eval_loss": 0.070068359375, | |
| "eval_precisions": [ | |
| 89.35727788279773, | |
| 83.87909319899245, | |
| 81.14676734308637, | |
| 77.90948275862068 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.6266, | |
| "eval_samples_per_second": 4.933, | |
| "eval_score": 49.221934768405774, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5290, | |
| "eval_totals": [ | |
| 5290, | |
| 4764, | |
| 4238, | |
| 3712 | |
| ], | |
| "step": 4369 | |
| }, | |
| { | |
| "epoch": 17.51, | |
| "learning_rate": 1.2996108949416343e-05, | |
| "loss": 0.1065, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_bp": 0.5610779943992972, | |
| "eval_counts": [ | |
| 4753, | |
| 4064, | |
| 3505, | |
| 2956 | |
| ], | |
| "eval_loss": 0.0623779296875, | |
| "eval_precisions": [ | |
| 93.14128943758574, | |
| 88.7917850120166, | |
| 86.52184645766478, | |
| 83.8581560283688 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.949, | |
| "eval_samples_per_second": 4.918, | |
| "eval_score": 49.382124037917905, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5103, | |
| "eval_totals": [ | |
| 5103, | |
| 4577, | |
| 4051, | |
| 3525 | |
| ], | |
| "step": 4626 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_bp": 0.5788702549376445, | |
| "eval_counts": [ | |
| 4784, | |
| 4087, | |
| 3529, | |
| 2977 | |
| ], | |
| "eval_loss": 0.060699462890625, | |
| "eval_precisions": [ | |
| 91.89396849788706, | |
| 87.32905982905983, | |
| 84.95426095329803, | |
| 82.05622932745314 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.4771, | |
| "eval_samples_per_second": 4.94, | |
| "eval_score": 50.0629990425284, | |
| "eval_steps_per_second": 0.31, | |
| "eval_sys_len": 5206, | |
| "eval_totals": [ | |
| 5206, | |
| 4680, | |
| 4154, | |
| 3628 | |
| ], | |
| "step": 4883 | |
| }, | |
| { | |
| "epoch": 19.46, | |
| "learning_rate": 1.2217898832684827e-05, | |
| "loss": 0.0964, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_bp": 0.5826501698750266, | |
| "eval_counts": [ | |
| 4773, | |
| 4068, | |
| 3509, | |
| 2957 | |
| ], | |
| "eval_loss": 0.0595703125, | |
| "eval_precisions": [ | |
| 91.29686304514155, | |
| 86.51637601020842, | |
| 84.02777777777777, | |
| 81.01369863013699 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.5688, | |
| "eval_samples_per_second": 4.936, | |
| "eval_score": 49.89324555557292, | |
| "eval_steps_per_second": 0.31, | |
| "eval_sys_len": 5228, | |
| "eval_totals": [ | |
| 5228, | |
| 4702, | |
| 4176, | |
| 3650 | |
| ], | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_bp": 0.5824785136401668, | |
| "eval_counts": [ | |
| 4780, | |
| 4078, | |
| 3521, | |
| 2972 | |
| ], | |
| "eval_loss": 0.057952880859375, | |
| "eval_precisions": [ | |
| 91.4482494738856, | |
| 86.74750053180175, | |
| 84.33532934131736, | |
| 81.44697177308852 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 105.7144, | |
| "eval_samples_per_second": 4.976, | |
| "eval_score": 50.04481904653482, | |
| "eval_steps_per_second": 0.312, | |
| "eval_sys_len": 5227, | |
| "eval_totals": [ | |
| 5227, | |
| 4701, | |
| 4175, | |
| 3649 | |
| ], | |
| "step": 5397 | |
| }, | |
| { | |
| "epoch": 21.4, | |
| "learning_rate": 1.1439688715953308e-05, | |
| "loss": 0.0925, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_bp": 0.6060221334079605, | |
| "eval_counts": [ | |
| 4800, | |
| 4076, | |
| 3514, | |
| 2962 | |
| ], | |
| "eval_loss": 0.060546875, | |
| "eval_precisions": [ | |
| 89.46877912395153, | |
| 84.23227939656954, | |
| 81.47461163923023, | |
| 78.21494586744124 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 105.9417, | |
| "eval_samples_per_second": 4.965, | |
| "eval_score": 50.4491686657978, | |
| "eval_steps_per_second": 0.311, | |
| "eval_sys_len": 5365, | |
| "eval_totals": [ | |
| 5365, | |
| 4839, | |
| 4313, | |
| 3787 | |
| ], | |
| "step": 5654 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_bp": 0.5761166700049626, | |
| "eval_counts": [ | |
| 4832, | |
| 4155, | |
| 3593, | |
| 3036 | |
| ], | |
| "eval_loss": 0.053558349609375, | |
| "eval_precisions": [ | |
| 93.10211946050096, | |
| 89.08662092624357, | |
| 86.82938617689705, | |
| 84.0531561461794 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.564, | |
| "eval_samples_per_second": 4.936, | |
| "eval_score": 50.81695573260325, | |
| "eval_steps_per_second": 0.31, | |
| "eval_sys_len": 5190, | |
| "eval_totals": [ | |
| 5190, | |
| 4664, | |
| 4138, | |
| 3612 | |
| ], | |
| "step": 5911 | |
| }, | |
| { | |
| "epoch": 23.35, | |
| "learning_rate": 1.066147859922179e-05, | |
| "loss": 0.0871, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_bp": 0.5719791556804446, | |
| "eval_counts": [ | |
| 4807, | |
| 4125, | |
| 3565, | |
| 3012 | |
| ], | |
| "eval_loss": 0.052276611328125, | |
| "eval_precisions": [ | |
| 93.05071622144793, | |
| 88.90086206896552, | |
| 86.65532328633932, | |
| 83.94648829431438 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.1172, | |
| "eval_samples_per_second": 4.911, | |
| "eval_score": 50.37743722047891, | |
| "eval_steps_per_second": 0.308, | |
| "eval_sys_len": 5166, | |
| "eval_totals": [ | |
| 5166, | |
| 4640, | |
| 4114, | |
| 3588 | |
| ], | |
| "step": 6168 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_bp": 0.579214183971878, | |
| "eval_counts": [ | |
| 4838, | |
| 4161, | |
| 3602, | |
| 3050 | |
| ], | |
| "eval_loss": 0.050567626953125, | |
| "eval_precisions": [ | |
| 92.89554531490015, | |
| 88.87227680478428, | |
| 86.66987487969202, | |
| 84.02203856749311 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.5446, | |
| "eval_samples_per_second": 4.937, | |
| "eval_score": 51.0028956058344, | |
| "eval_steps_per_second": 0.31, | |
| "eval_sys_len": 5208, | |
| "eval_totals": [ | |
| 5208, | |
| 4682, | |
| 4156, | |
| 3630 | |
| ], | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 25.29, | |
| "learning_rate": 9.883268482490273e-06, | |
| "loss": 0.0843, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_bp": 0.5607309845734951, | |
| "eval_counts": [ | |
| 4817, | |
| 4157, | |
| 3596, | |
| 3042 | |
| ], | |
| "eval_loss": 0.051177978515625, | |
| "eval_precisions": [ | |
| 94.43246422270143, | |
| 90.86338797814207, | |
| 88.81205235860706, | |
| 86.34686346863468 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.5728, | |
| "eval_samples_per_second": 4.89, | |
| "eval_score": 50.50236718840154, | |
| "eval_steps_per_second": 0.307, | |
| "eval_sys_len": 5101, | |
| "eval_totals": [ | |
| 5101, | |
| 4575, | |
| 4049, | |
| 3523 | |
| ], | |
| "step": 6682 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_bp": 0.5869366550146455, | |
| "eval_counts": [ | |
| 4855, | |
| 4170, | |
| 3608, | |
| 3055 | |
| ], | |
| "eval_loss": 0.0489501953125, | |
| "eval_precisions": [ | |
| 92.42337711783743, | |
| 88.21662788237784, | |
| 85.88431325874792, | |
| 83.12925170068027 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 109.9833, | |
| "eval_samples_per_second": 4.783, | |
| "eval_score": 51.26739301541927, | |
| "eval_steps_per_second": 0.3, | |
| "eval_sys_len": 5253, | |
| "eval_totals": [ | |
| 5253, | |
| 4727, | |
| 4201, | |
| 3675 | |
| ], | |
| "step": 6939 | |
| }, | |
| { | |
| "epoch": 27.24, | |
| "learning_rate": 9.105058365758756e-06, | |
| "loss": 0.0813, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_bp": 0.5641984935077309, | |
| "eval_counts": [ | |
| 4838, | |
| 4184, | |
| 3624, | |
| 3070 | |
| ], | |
| "eval_loss": 0.047760009765625, | |
| "eval_precisions": [ | |
| 94.47373559851592, | |
| 91.05549510337323, | |
| 89.06365200294913, | |
| 86.64973186565058 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 109.0318, | |
| "eval_samples_per_second": 4.824, | |
| "eval_score": 50.9275946797506, | |
| "eval_steps_per_second": 0.303, | |
| "eval_sys_len": 5121, | |
| "eval_totals": [ | |
| 5121, | |
| 4595, | |
| 4069, | |
| 3543 | |
| ], | |
| "step": 7196 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_bp": 0.5711161019095474, | |
| "eval_counts": [ | |
| 4838, | |
| 4179, | |
| 3625, | |
| 3079 | |
| ], | |
| "eval_loss": 0.0462646484375, | |
| "eval_precisions": [ | |
| 93.74152296066654, | |
| 90.16181229773463, | |
| 88.22097834022877, | |
| 85.93357521629919 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.2897, | |
| "eval_samples_per_second": 4.903, | |
| "eval_score": 51.09715872720289, | |
| "eval_steps_per_second": 0.308, | |
| "eval_sys_len": 5161, | |
| "eval_totals": [ | |
| 5161, | |
| 4635, | |
| 4109, | |
| 3583 | |
| ], | |
| "step": 7453 | |
| }, | |
| { | |
| "epoch": 29.18, | |
| "learning_rate": 8.326848249027239e-06, | |
| "loss": 0.0778, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_bp": 0.587279163676868, | |
| "eval_counts": [ | |
| 4863, | |
| 4185, | |
| 3626, | |
| 3075 | |
| ], | |
| "eval_loss": 0.04534912109375, | |
| "eval_precisions": [ | |
| 92.54043767840152, | |
| 88.49651089025164, | |
| 86.27171068284558, | |
| 83.62795757410933 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 108.186, | |
| "eval_samples_per_second": 4.862, | |
| "eval_score": 51.488944843891275, | |
| "eval_steps_per_second": 0.305, | |
| "eval_sys_len": 5255, | |
| "eval_totals": [ | |
| 5255, | |
| 4729, | |
| 4203, | |
| 3677 | |
| ], | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_bp": 0.587279163676868, | |
| "eval_counts": [ | |
| 4847, | |
| 4168, | |
| 3612, | |
| 3064 | |
| ], | |
| "eval_loss": 0.044677734375, | |
| "eval_precisions": [ | |
| 92.23596574690771, | |
| 88.137026855572, | |
| 85.93861527480371, | |
| 83.32880065270601 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.6208, | |
| "eval_samples_per_second": 4.933, | |
| "eval_score": 51.298555626377826, | |
| "eval_steps_per_second": 0.31, | |
| "eval_sys_len": 5255, | |
| "eval_totals": [ | |
| 5255, | |
| 4729, | |
| 4203, | |
| 3677 | |
| ], | |
| "step": 7967 | |
| }, | |
| { | |
| "epoch": 31.13, | |
| "learning_rate": 7.54863813229572e-06, | |
| "loss": 0.0753, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_bp": 0.5690432735111319, | |
| "eval_counts": [ | |
| 4866, | |
| 4219, | |
| 3661, | |
| 3111 | |
| ], | |
| "eval_loss": 0.0438232421875, | |
| "eval_precisions": [ | |
| 94.50378714313459, | |
| 91.26108587497296, | |
| 89.35806687820356, | |
| 87.11845421450575 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.5192, | |
| "eval_samples_per_second": 4.892, | |
| "eval_score": 51.50981459551784, | |
| "eval_steps_per_second": 0.307, | |
| "eval_sys_len": 5149, | |
| "eval_totals": [ | |
| 5149, | |
| 4623, | |
| 4097, | |
| 3571 | |
| ], | |
| "step": 8224 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_bp": 0.5881351685074624, | |
| "eval_counts": [ | |
| 4869, | |
| 4201, | |
| 3645, | |
| 3097 | |
| ], | |
| "eval_loss": 0.04400634765625, | |
| "eval_precisions": [ | |
| 92.56653992395437, | |
| 88.74102239121251, | |
| 86.62072243346007, | |
| 84.11189570885388 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.8688, | |
| "eval_samples_per_second": 4.922, | |
| "eval_score": 51.729891771805434, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5260, | |
| "eval_totals": [ | |
| 5260, | |
| 4734, | |
| 4208, | |
| 3682 | |
| ], | |
| "step": 8481 | |
| }, | |
| { | |
| "epoch": 33.07, | |
| "learning_rate": 6.770428015564204e-06, | |
| "loss": 0.0714, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_bp": 0.5823068423133116, | |
| "eval_counts": [ | |
| 4881, | |
| 4226, | |
| 3674, | |
| 3130 | |
| ], | |
| "eval_loss": 0.041656494140625, | |
| "eval_precisions": [ | |
| 93.398392652124, | |
| 89.91489361702128, | |
| 88.02108289410637, | |
| 85.80043859649123 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.0835, | |
| "eval_samples_per_second": 4.912, | |
| "eval_score": 51.96533200156475, | |
| "eval_steps_per_second": 0.308, | |
| "eval_sys_len": 5226, | |
| "eval_totals": [ | |
| 5226, | |
| 4700, | |
| 4174, | |
| 3648 | |
| ], | |
| "step": 8738 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_bp": 0.5862514549555176, | |
| "eval_counts": [ | |
| 4902, | |
| 4242, | |
| 3685, | |
| 3133 | |
| ], | |
| "eval_loss": 0.042633056640625, | |
| "eval_precisions": [ | |
| 93.38921699371309, | |
| 89.81579504552191, | |
| 87.80081010245414, | |
| 85.34459275401798 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.8404, | |
| "eval_samples_per_second": 4.923, | |
| "eval_score": 52.19933016750815, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5249, | |
| "eval_totals": [ | |
| 5249, | |
| 4723, | |
| 4197, | |
| 3671 | |
| ], | |
| "step": 8995 | |
| }, | |
| { | |
| "epoch": 35.02, | |
| "learning_rate": 5.992217898832685e-06, | |
| "loss": 0.0697, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_bp": 0.5807611221368078, | |
| "eval_counts": [ | |
| 4907, | |
| 4257, | |
| 3699, | |
| 3149 | |
| ], | |
| "eval_loss": 0.04095458984375, | |
| "eval_precisions": [ | |
| 94.05788767490895, | |
| 90.74824131315285, | |
| 88.81152460984394, | |
| 86.53476229733444 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.3976, | |
| "eval_samples_per_second": 4.944, | |
| "eval_score": 52.266194224133834, | |
| "eval_steps_per_second": 0.31, | |
| "eval_sys_len": 5217, | |
| "eval_totals": [ | |
| 5217, | |
| 4691, | |
| 4165, | |
| 3639 | |
| ], | |
| "step": 9252 | |
| }, | |
| { | |
| "epoch": 36.96, | |
| "learning_rate": 5.214007782101168e-06, | |
| "loss": 0.0686, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_bp": 0.5983772718445015, | |
| "eval_counts": [ | |
| 4899, | |
| 4227, | |
| 3672, | |
| 3123 | |
| ], | |
| "eval_loss": 0.042388916015625, | |
| "eval_precisions": [ | |
| 92.08646616541354, | |
| 88.17271589486859, | |
| 86.03561387066541, | |
| 83.45804382683058 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.906, | |
| "eval_samples_per_second": 4.92, | |
| "eval_score": 52.28705860616529, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5320, | |
| "eval_totals": [ | |
| 5320, | |
| 4794, | |
| 4268, | |
| 3742 | |
| ], | |
| "step": 9509 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_bp": 0.5780101703802235, | |
| "eval_counts": [ | |
| 4913, | |
| 4273, | |
| 3718, | |
| 3172 | |
| ], | |
| "eval_loss": 0.0394287109375, | |
| "eval_precisions": [ | |
| 94.46260334551047, | |
| 91.40106951871658, | |
| 89.61195468787659, | |
| 87.55175269113994 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.7621, | |
| "eval_samples_per_second": 4.927, | |
| "eval_score": 52.43798269679689, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5201, | |
| "eval_totals": [ | |
| 5201, | |
| 4675, | |
| 4149, | |
| 3623 | |
| ], | |
| "step": 9766 | |
| }, | |
| { | |
| "epoch": 38.91, | |
| "learning_rate": 4.43579766536965e-06, | |
| "loss": 0.0664, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_bp": 0.5975258891581067, | |
| "eval_counts": [ | |
| 4912, | |
| 4243, | |
| 3689, | |
| 3141 | |
| ], | |
| "eval_loss": 0.040374755859375, | |
| "eval_precisions": [ | |
| 92.41768579492003, | |
| 88.59887241595322, | |
| 86.5353037766831, | |
| 84.05137811078406 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.9692, | |
| "eval_samples_per_second": 4.917, | |
| "eval_score": 52.491270926490635, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5315, | |
| "eval_totals": [ | |
| 5315, | |
| 4789, | |
| 4263, | |
| 3737 | |
| ], | |
| "step": 10023 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_bp": 0.587279163676868, | |
| "eval_counts": [ | |
| 4913, | |
| 4259, | |
| 3711, | |
| 3170 | |
| ], | |
| "eval_loss": 0.0382080078125, | |
| "eval_precisions": [ | |
| 93.4919124643197, | |
| 90.0613237470924, | |
| 88.29407566024268, | |
| 86.21158553168344 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.0316, | |
| "eval_samples_per_second": 4.914, | |
| "eval_score": 52.5468859983503, | |
| "eval_steps_per_second": 0.308, | |
| "eval_sys_len": 5255, | |
| "eval_totals": [ | |
| 5255, | |
| 4729, | |
| 4203, | |
| 3677 | |
| ], | |
| "step": 10280 | |
| }, | |
| { | |
| "epoch": 40.86, | |
| "learning_rate": 3.6575875486381323e-06, | |
| "loss": 0.0658, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_bp": 0.5811047209098391, | |
| "eval_counts": [ | |
| 4921, | |
| 4278, | |
| 3725, | |
| 3179 | |
| ], | |
| "eval_loss": 0.0377197265625, | |
| "eval_precisions": [ | |
| 94.29009388771796, | |
| 91.1570424035798, | |
| 89.39284857211423, | |
| 87.31117824773413 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.7065, | |
| "eval_samples_per_second": 4.929, | |
| "eval_score": 52.59102479681527, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5219, | |
| "eval_totals": [ | |
| 5219, | |
| 4693, | |
| 4167, | |
| 3641 | |
| ], | |
| "step": 10537 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_bp": 0.5817917378355022, | |
| "eval_counts": [ | |
| 4908, | |
| 4261, | |
| 3712, | |
| 3169 | |
| ], | |
| "eval_loss": 0.037109375, | |
| "eval_precisions": [ | |
| 93.96898334290637, | |
| 90.7174792420694, | |
| 88.99544473747302, | |
| 86.94101508916324 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.4197, | |
| "eval_samples_per_second": 4.897, | |
| "eval_score": 52.43056600057888, | |
| "eval_steps_per_second": 0.307, | |
| "eval_sys_len": 5223, | |
| "eval_totals": [ | |
| 5223, | |
| 4697, | |
| 4171, | |
| 3645 | |
| ], | |
| "step": 10794 | |
| }, | |
| { | |
| "epoch": 42.8, | |
| "learning_rate": 2.879377431906615e-06, | |
| "loss": 0.0643, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_bp": 0.5804174632159932, | |
| "eval_counts": [ | |
| 4905, | |
| 4264, | |
| 3714, | |
| 3172 | |
| ], | |
| "eval_loss": 0.037017822265625, | |
| "eval_precisions": [ | |
| 94.0556088207095, | |
| 90.936233738537, | |
| 89.21450876771559, | |
| 87.21473742095134 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 106.7676, | |
| "eval_samples_per_second": 4.927, | |
| "eval_score": 52.42364666449266, | |
| "eval_steps_per_second": 0.309, | |
| "eval_sys_len": 5215, | |
| "eval_totals": [ | |
| 5215, | |
| 4689, | |
| 4163, | |
| 3637 | |
| ], | |
| "step": 11051 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_bp": 0.5961628688829712, | |
| "eval_counts": [ | |
| 4930, | |
| 4270, | |
| 3718, | |
| 3173 | |
| ], | |
| "eval_loss": 0.0380859375, | |
| "eval_precisions": [ | |
| 92.89617486338798, | |
| 89.31185944363104, | |
| 87.37955346650999, | |
| 85.08983641727005 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.4748, | |
| "eval_samples_per_second": 4.894, | |
| "eval_score": 52.834006019511406, | |
| "eval_steps_per_second": 0.307, | |
| "eval_sys_len": 5307, | |
| "eval_totals": [ | |
| 5307, | |
| 4781, | |
| 4255, | |
| 3729 | |
| ], | |
| "step": 11308 | |
| }, | |
| { | |
| "epoch": 44.75, | |
| "learning_rate": 2.1011673151750974e-06, | |
| "loss": 0.0608, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_bp": 0.5757722034899391, | |
| "eval_counts": [ | |
| 4915, | |
| 4280, | |
| 3729, | |
| 3186 | |
| ], | |
| "eval_loss": 0.036224365234375, | |
| "eval_precisions": [ | |
| 94.7378565921357, | |
| 91.8060918060918, | |
| 90.15957446808511, | |
| 88.25484764542936 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.743, | |
| "eval_samples_per_second": 4.882, | |
| "eval_score": 52.515446703245765, | |
| "eval_steps_per_second": 0.306, | |
| "eval_sys_len": 5188, | |
| "eval_totals": [ | |
| 5188, | |
| 4662, | |
| 4136, | |
| 3610 | |
| ], | |
| "step": 11565 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_bp": 0.5843659009664612, | |
| "eval_counts": [ | |
| 4924, | |
| 4278, | |
| 3730, | |
| 3188 | |
| ], | |
| "eval_loss": 0.036651611328125, | |
| "eval_precisions": [ | |
| 94.0053455517373, | |
| 90.78947368421052, | |
| 89.10654562828476, | |
| 87.10382513661202 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 107.889, | |
| "eval_samples_per_second": 4.875, | |
| "eval_score": 52.71917275684773, | |
| "eval_steps_per_second": 0.306, | |
| "eval_sys_len": 5238, | |
| "eval_totals": [ | |
| 5238, | |
| 4712, | |
| 4186, | |
| 3660 | |
| ], | |
| "step": 11822 | |
| }, | |
| { | |
| "epoch": 46.69, | |
| "learning_rate": 1.32295719844358e-06, | |
| "loss": 0.0622, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_bp": 0.586080116901772, | |
| "eval_counts": [ | |
| 4938, | |
| 4295, | |
| 3745, | |
| 3201 | |
| ], | |
| "eval_loss": 0.036529541015625, | |
| "eval_precisions": [ | |
| 94.09298780487805, | |
| 90.95722151630665, | |
| 89.2516682554814, | |
| 87.22070844686648 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 109.2986, | |
| "eval_samples_per_second": 4.813, | |
| "eval_score": 52.949832085516945, | |
| "eval_steps_per_second": 0.302, | |
| "eval_sys_len": 5248, | |
| "eval_totals": [ | |
| 5248, | |
| 4722, | |
| 4196, | |
| 3670 | |
| ], | |
| "step": 12079 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_bp": 0.5817917378355022, | |
| "eval_counts": [ | |
| 4925, | |
| 4285, | |
| 3733, | |
| 3189 | |
| ], | |
| "eval_loss": 0.036285400390625, | |
| "eval_precisions": [ | |
| 94.29446678154318, | |
| 91.22844368746009, | |
| 89.49892112203308, | |
| 87.48971193415638 | |
| ], | |
| "eval_ref_len": 8052, | |
| "eval_runtime": 108.6659, | |
| "eval_samples_per_second": 4.841, | |
| "eval_score": 52.70664408353883, | |
| "eval_steps_per_second": 0.304, | |
| "eval_sys_len": 5223, | |
| "eval_totals": [ | |
| 5223, | |
| 4697, | |
| 4171, | |
| 3645 | |
| ], | |
| "step": 12336 | |
| }, | |
| { | |
| "epoch": 48.64, | |
| "learning_rate": 5.447470817120623e-07, | |
| "loss": 0.0625, | |
| "step": 12500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 12850, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 50, | |
| "save_steps": 500, | |
| "total_flos": 8.680648839008256e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |