| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.24560541735377706, | |
| "global_step": 3500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.1428571428571429e-05, | |
| "loss": 1.2994, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.2857142857142858e-05, | |
| "loss": 1.2625, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.428571428571429e-05, | |
| "loss": 1.1582, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.5714285714285716e-05, | |
| "loss": 0.9934, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.714285714285714e-05, | |
| "loss": 0.8862, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.857142857142858e-05, | |
| "loss": 0.7613, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8e-05, | |
| "loss": 0.6832, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.142857142857143e-05, | |
| "loss": 0.5934, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00010285714285714286, | |
| "loss": 0.5305, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00011428571428571428, | |
| "loss": 0.5195, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00012571428571428572, | |
| "loss": 0.5001, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00013714285714285716, | |
| "loss": 0.4812, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00014857142857142857, | |
| "loss": 0.4624, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00016, | |
| "loss": 0.4699, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00017142857142857143, | |
| "loss": 0.4355, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00018285714285714286, | |
| "loss": 0.4341, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00019314285714285717, | |
| "loss": 0.4243, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00020457142857142858, | |
| "loss": 0.4255, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00021600000000000002, | |
| "loss": 0.4116, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00022742857142857146, | |
| "loss": 0.425, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0002388571428571429, | |
| "loss": 0.4121, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002502857142857143, | |
| "loss": 0.4066, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002617142857142857, | |
| "loss": 0.4019, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00027314285714285716, | |
| "loss": 0.3961, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00028457142857142857, | |
| "loss": 0.3947, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000296, | |
| "loss": 0.4015, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00030742857142857145, | |
| "loss": 0.3862, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00031885714285714286, | |
| "loss": 0.4013, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00033028571428571433, | |
| "loss": 0.3913, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00034171428571428574, | |
| "loss": 0.4038, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00035314285714285715, | |
| "loss": 0.3962, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0003645714285714286, | |
| "loss": 0.3897, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000376, | |
| "loss": 0.3875, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00038742857142857144, | |
| "loss": 0.378, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00039885714285714286, | |
| "loss": 0.3843, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00039999194323417424, | |
| "loss": 0.3827, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000399964093519005, | |
| "loss": 0.3828, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0003999163541576902, | |
| "loss": 0.3905, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00039984872989867726, | |
| "loss": 0.3938, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0003997612274682868, | |
| "loss": 0.3811, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0003996538555700433, | |
| "loss": 0.3942, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0003995266248838093, | |
| "loss": 0.3746, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00039937954806472354, | |
| "loss": 0.3785, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0003992126397419419, | |
| "loss": 0.3807, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00039902591651718225, | |
| "loss": 0.3765, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0003988193969630734, | |
| "loss": 0.375, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0003985931016213074, | |
| "loss": 0.3886, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00039834705300059676, | |
| "loss": 0.3796, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0003980812755744352, | |
| "loss": 0.3778, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00039779579577866366, | |
| "loss": 0.3801, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0003974906420088406, | |
| "loss": 0.382, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00039716584461741783, | |
| "loss": 0.3779, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0003968214359107213, | |
| "loss": 0.3811, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0003964574501457378, | |
| "loss": 0.3837, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0003960739235267074, | |
| "loss": 0.3798, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00039567089420152265, | |
| "loss": 0.3801, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00039524840225793395, | |
| "loss": 0.3761, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00039480648971956214, | |
| "loss": 0.3702, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00039434520054171875, | |
| "loss": 0.3787, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0003938645806070337, | |
| "loss": 0.364, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00039336467772089194, | |
| "loss": 0.3669, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0003928455416066777, | |
| "loss": 0.3713, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00039230722390082984, | |
| "loss": 0.3676, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0003917497781477043, | |
| "loss": 0.3715, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00039117325979424984, | |
| "loss": 0.3626, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00039057772618449163, | |
| "loss": 0.3744, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0003899632365538282, | |
| "loss": 0.3686, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0003893298520231393, | |
| "loss": 0.3683, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0003886776355927065, | |
| "loss": 0.3745, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00038800665213594663, | |
| "loss": 0.3701, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0003873169683929594, | |
| "loss": 0.3685, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0003866086529638889, | |
| "loss": 0.3649, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0003858817763020997, | |
| "loss": 0.3621, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00038513641070717016, | |
| "loss": 0.3678, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0003843726303177001, | |
| "loss": 0.362, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0003835905111039371, | |
| "loss": 0.3551, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00038279013086021973, | |
| "loss": 0.3675, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00038197156919723984, | |
| "loss": 0.3666, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000381134907534124, | |
| "loss": 0.3719, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0003802802290903348, | |
| "loss": 0.3675, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00037940761887739373, | |
| "loss": 0.3746, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00037851716369042513, | |
| "loss": 0.3677, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000377608952099523, | |
| "loss": 0.3689, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00037668307444094153, | |
| "loss": 0.3606, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00037573962280810944, | |
| "loss": 0.3657, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0003747786910424701, | |
| "loss": 0.3601, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00037380037472414704, | |
| "loss": 0.3732, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0003728047711624374, | |
| "loss": 0.3765, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0003717919793861329, | |
| "loss": 0.3641, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00037076210013366943, | |
| "loss": 0.3621, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00036971523584310764, | |
| "loss": 0.3635, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0003686514906419431, | |
| "loss": 0.3716, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0003675709703367499, | |
| "loss": 0.3613, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0003664737824026558, | |
| "loss": 0.3728, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0003653600359726526, | |
| "loss": 0.362, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00036422984182674084, | |
| "loss": 0.3518, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00036308331238091095, | |
| "loss": 0.3672, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00036192056167596193, | |
| "loss": 0.3563, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00036074170536615777, | |
| "loss": 0.358, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00035954686070772404, | |
| "loss": 0.3594, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00035833614654718493, | |
| "loss": 0.3635, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00035710968330954164, | |
| "loss": 0.354, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0003558675929862949, | |
| "loss": 0.3721, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00035460999912330994, | |
| "loss": 0.3588, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0003533370268085289, | |
| "loss": 0.3608, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0003520488026595282, | |
| "loss": 0.3539, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00035074545481092456, | |
| "loss": 0.3597, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00034942711290162997, | |
| "loss": 0.3493, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00034809390806195685, | |
| "loss": 0.3648, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0003467459729005753, | |
| "loss": 0.3553, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0003453834414913229, | |
| "loss": 0.3497, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00034400644935986876, | |
| "loss": 0.3496, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0003426151334702336, | |
| "loss": 0.3612, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00034120963221116637, | |
| "loss": 0.3578, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00033979008538237914, | |
| "loss": 0.3502, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0003383566341806417, | |
| "loss": 0.3556, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00033690942118573774, | |
| "loss": 0.3525, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0003354485903462825, | |
| "loss": 0.3524, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000333974286965405, | |
| "loss": 0.353, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0003324866576862952, | |
| "loss": 0.362, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00033098585047761797, | |
| "loss": 0.3613, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003294720146187955, | |
| "loss": 0.3514, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00032794530068515874, | |
| "loss": 0.3638, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003264058605329702, | |
| "loss": 0.3511, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003248538472843198, | |
| "loss": 0.3619, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00032328941531189397, | |
| "loss": 0.3535, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00032171272022362134, | |
| "loss": 0.3523, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003201239188471943, | |
| "loss": 0.3581, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003185231692144706, | |
| "loss": 0.3555, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00031691063054575434, | |
| "loss": 0.3472, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003152864632339587, | |
| "loss": 0.3538, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003136508288286527, | |
| "loss": 0.3477, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003120038900199922, | |
| "loss": 0.3475, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00031034581062253786, | |
| "loss": 0.3521, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0003086767555589612, | |
| "loss": 0.3531, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0003069968908436401, | |
| "loss": 0.3448, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00030530638356614644, | |
| "loss": 0.3564, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00030360540187462606, | |
| "loss": 0.3473, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00030189411495907376, | |
| "loss": 0.3575, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0003001726930345046, | |
| "loss": 0.3478, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00029844130732402347, | |
| "loss": 0.3499, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000296700130041794, | |
| "loss": 0.3638, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00029494933437590926, | |
| "loss": 0.3494, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00029318909447116496, | |
| "loss": 0.3497, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002914195854117389, | |
| "loss": 0.3555, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002896409832037748, | |
| "loss": 0.352, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002878534647578768, | |
| "loss": 0.3486, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002860572078715121, | |
| "loss": 0.3528, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0002842523912113264, | |
| "loss": 0.3412, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00028243919429537265, | |
| "loss": 0.3384, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00028061779747525504, | |
| "loss": 0.3493, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00027878838191819024, | |
| "loss": 0.3416, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002769511295889872, | |
| "loss": 0.3485, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000275106223231948, | |
| "loss": 0.3436, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00027325384635269094, | |
| "loss": 0.3405, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00027139418319989785, | |
| "loss": 0.349, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002695274187469878, | |
| "loss": 0.3366, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002676537386737183, | |
| "loss": 0.3524, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00026577332934771665, | |
| "loss": 0.3405, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002638863778059426, | |
| "loss": 0.3437, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00026199307173608433, | |
| "loss": 0.3375, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002600935994578904, | |
| "loss": 0.3447, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002581881499044377, | |
| "loss": 0.3464, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002562769126033394, | |
| "loss": 0.3369, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00025436007765789327, | |
| "loss": 0.3423, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00025243783572817297, | |
| "loss": 0.3433, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002505103780120636, | |
| "loss": 0.3412, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002485778962262443, | |
| "loss": 0.3424, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00024664058258711853, | |
| "loss": 0.3452, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002446986297916954, | |
| "loss": 0.3493, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00024275223099842291, | |
| "loss": 0.3387, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00024080157980797484, | |
| "loss": 0.3421, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002388468702439944, | |
| "loss": 0.3365, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00023688829673379534, | |
| "loss": 0.3382, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00023492605408902297, | |
| "loss": 0.3443, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00023296033748627712, | |
| "loss": 0.3441, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0002309913424476986, | |
| "loss": 0.3425, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00022901926482152138, | |
| "loss": 0.3469, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00022704430076259246, | |
| "loss": 0.3409, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00022506664671286087, | |
| "loss": 0.3585, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00022308649938183864, | |
| "loss": 0.3327, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00022110405572703466, | |
| "loss": 0.3354, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00021911951293436416, | |
| "loss": 0.3424, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00021713306839853545, | |
| "loss": 0.3369, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002151449197034157, | |
| "loss": 0.3407, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002131552646023783, | |
| "loss": 0.3391, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00021116430099863277, | |
| "loss": 0.3358, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002091722269255404, | |
| "loss": 0.3364, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002071792405269165, | |
| "loss": 0.341, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00020518554003732167, | |
| "loss": 0.3378, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00020319132376234462, | |
| "loss": 0.3305, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00020119679005887702, | |
| "loss": 0.3352, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00019920213731538394, | |
| "loss": 0.3376, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00019720756393217098, | |
| "loss": 0.3283, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00019521326830164998, | |
| "loss": 0.3354, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00019321944878860587, | |
| "loss": 0.3445, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000191226303710466, | |
| "loss": 0.3401, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00018923403131757439, | |
| "loss": 0.3438, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00018724282977347235, | |
| "loss": 0.3341, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00018525289713518817, | |
| "loss": 0.3272, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00018326443133353693, | |
| "loss": 0.3488, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00018127763015343332, | |
| "loss": 0.3242, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00017929269121421857, | |
| "loss": 0.337, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00017730981195000406, | |
| "loss": 0.3386, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00017532918959003353, | |
| "loss": 0.3312, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00017335102113906505, | |
| "loss": 0.3382, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00017137550335777612, | |
| "loss": 0.3319, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001694028327431924, | |
| "loss": 0.3294, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001674332055091431, | |
| "loss": 0.3313, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001654668175667442, | |
| "loss": 0.3347, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00016350386450491208, | |
| "loss": 0.324, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00016154454157090884, | |
| "loss": 0.3302, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00015958904365092225, | |
| "loss": 0.334, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00015763756525068065, | |
| "loss": 0.3375, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00015569030047610656, | |
| "loss": 0.337, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001537474430140096, | |
| "loss": 0.3346, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001518091861128213, | |
| "loss": 0.3377, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00014987572256337336, | |
| "loss": 0.3247, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001479472446797216, | |
| "loss": 0.3194, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00014602394428001712, | |
| "loss": 0.3299, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00014410601266742691, | |
| "loss": 0.3202, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00014219364061110565, | |
| "loss": 0.3316, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00014028701832722104, | |
| "loss": 0.3304, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00013838633546003302, | |
| "loss": 0.3286, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00013649178106303115, | |
| "loss": 0.3267, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00013460354358013, | |
| "loss": 0.3295, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0001327218108269255, | |
| "loss": 0.3321, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00013084676997201342, | |
| "loss": 0.3302, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012897860751837263, | |
| "loss": 0.3246, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012711750928481443, | |
| "loss": 0.3305, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012526366038749956, | |
| "loss": 0.3274, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.000123417245221526, | |
| "loss": 0.3303, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00012157844744258722, | |
| "loss": 0.3265, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00011974744994870517, | |
| "loss": 0.3237, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00011792443486203788, | |
| "loss": 0.3288, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00011610958351076458, | |
| "loss": 0.3372, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00011430307641104971, | |
| "loss": 0.3257, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00011250509324908767, | |
| "loss": 0.3294, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00011071581286323, | |
| "loss": 0.3304, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00010893541322619732, | |
| "loss": 0.33, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00010716407142737659, | |
| "loss": 0.3251, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00010540196365520754, | |
| "loss": 0.3293, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00010364926517965692, | |
| "loss": 0.3269, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001019061503347858, | |
| "loss": 0.3187, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00010017279250140891, | |
| "loss": 0.3204, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.844936408984924e-05, | |
| "loss": 0.3151, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.673603652278904e-05, | |
| "loss": 0.3237, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.503298021821905e-05, | |
| "loss": 0.3224, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.334036457248774e-05, | |
| "loss": 0.3228, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.165835794345205e-05, | |
| "loss": 0.3186, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.99871276337315e-05, | |
| "loss": 0.3181, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.832683987406746e-05, | |
| "loss": 0.3233, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.667765980678851e-05, | |
| "loss": 0.3302, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.503975146938444e-05, | |
| "loss": 0.3314, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.341327777819035e-05, | |
| "loss": 0.3309, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.179840051218167e-05, | |
| "loss": 0.331, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.019528029688286e-05, | |
| "loss": 0.318, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.860407658839049e-05, | |
| "loss": 0.3273, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.702494765751285e-05, | |
| "loss": 0.3182, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.545805057402733e-05, | |
| "loss": 0.3237, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.390354119105722e-05, | |
| "loss": 0.3229, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.236157412956994e-05, | |
| "loss": 0.3304, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.0832302762997e-05, | |
| "loss": 0.3284, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.93158792019789e-05, | |
| "loss": 0.3251, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.781245427923522e-05, | |
| "loss": 0.3201, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.632217753456174e-05, | |
| "loss": 0.322, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.484519719995647e-05, | |
| "loss": 0.3265, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.338166018487555e-05, | |
| "loss": 0.3273, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.193171206162065e-05, | |
| "loss": 0.3217, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.0495497050859574e-05, | |
| "loss": 0.3247, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.907315800728106e-05, | |
| "loss": 0.3179, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.766483640538587e-05, | |
| "loss": 0.3116, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.62706723254145e-05, | |
| "loss": 0.315, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.489080443941415e-05, | |
| "loss": 0.3197, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.352536999744557e-05, | |
| "loss": 0.32, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.217450481393129e-05, | |
| "loss": 0.3195, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.083834325414667e-05, | |
| "loss": 0.3221, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.951701822085515e-05, | |
| "loss": 0.3165, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.821066114108892e-05, | |
| "loss": 0.3216, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.69194019530764e-05, | |
| "loss": 0.3122, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.564336909331768e-05, | |
| "loss": 0.3237, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.438268948380972e-05, | |
| "loss": 0.3169, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.3137488519421656e-05, | |
| "loss": 0.3146, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.1907890055422286e-05, | |
| "loss": 0.3256, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.069401639516075e-05, | |
| "loss": 0.3178, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.949598827790155e-05, | |
| "loss": 0.3151, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.831392486681495e-05, | |
| "loss": 0.3121, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.714794373712431e-05, | |
| "loss": 0.3199, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.5998160864411476e-05, | |
| "loss": 0.3226, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.486469061308093e-05, | |
| "loss": 0.3182, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.3747645724984544e-05, | |
| "loss": 0.3162, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.264713730820768e-05, | |
| "loss": 0.3133, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.156327482601742e-05, | |
| "loss": 0.318, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.0496166085974943e-05, | |
| "loss": 0.3107, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.9445917229212193e-05, | |
| "loss": 0.3188, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.8412632719874532e-05, | |
| "loss": 0.3278, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.7396415334729964e-05, | |
| "loss": 0.3161, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.6397366152946523e-05, | |
| "loss": 0.3151, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.5415584546038096e-05, | |
| "loss": 0.316, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.4451168167980497e-05, | |
| "loss": 0.3191, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.350421294549825e-05, | |
| "loss": 0.3152, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.2574813068522894e-05, | |
| "loss": 0.3195, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.166306098082451e-05, | |
| "loss": 0.3139, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.0769047370816553e-05, | |
| "loss": 0.3214, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.989286116253557e-05, | |
| "loss": 0.3114, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.903458950679613e-05, | |
| "loss": 0.3203, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8194317772522362e-05, | |
| "loss": 0.3132, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7372129538256667e-05, | |
| "loss": 0.3091, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.6568106583846378e-05, | |
| "loss": 0.3111, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.5782328882309484e-05, | |
| "loss": 0.3126, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.5014874591880157e-05, | |
| "loss": 0.3173, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.4265820048234447e-05, | |
| "loss": 0.3084, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.3535239756897566e-05, | |
| "loss": 0.3133, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.2823206385833187e-05, | |
| "loss": 0.3182, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.212979075821532e-05, | |
| "loss": 0.3109, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.1455061845383852e-05, | |
| "loss": 0.3169, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.0799086759984333e-05, | |
| "loss": 0.3185, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.016193074929237e-05, | |
| "loss": 0.3035, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.543657188723876e-06, | |
| "loss": 0.3117, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.944327575531275e-06, | |
| "loss": 0.3204, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.364001522686726e-06, | |
| "loss": 0.3144, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.802736752952533e-06, | |
| "loss": 0.321, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 7.260589093139736e-06, | |
| "loss": 0.316, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 6.737612468555221e-06, | |
| "loss": 0.3133, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 6.2338588976380115e-06, | |
| "loss": 0.3107, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.749378486785162e-06, | |
| "loss": 0.3076, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 5.2842194253679424e-06, | |
| "loss": 0.3137, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.8384279809385426e-06, | |
| "loss": 0.3121, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.41204849462804e-06, | |
| "loss": 0.3065, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.005123376735997e-06, | |
| "loss": 0.3172, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.617693102512032e-06, | |
| "loss": 0.3183, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.2497962081299514e-06, | |
| "loss": 0.3141, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.9014692868546633e-06, | |
| "loss": 0.3094, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.572746985402419e-06, | |
| "loss": 0.3139, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.2636620004946154e-06, | |
| "loss": 0.3091, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.974245075605574e-06, | |
| "loss": 0.3092, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7045249979046995e-06, | |
| "loss": 0.3108, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.4545285953929677e-06, | |
| "loss": 0.3098, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.224280734234573e-06, | |
| "loss": 0.3059, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.013804316283573e-06, | |
| "loss": 0.3073, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.231202768059332e-07, | |
| "loss": 0.306, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 6.522475823970808e-07, | |
| "loss": 0.3071, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.012032290955037e-07, | |
| "loss": 0.3033, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.7000224069216883e-07, | |
| "loss": 0.3109, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.586576672361396e-07, | |
| "loss": 0.3127, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.671805837365703e-07, | |
| "loss": 0.312, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.558008906112026e-08, | |
| "loss": 0.317, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.386330503090008e-08, | |
| "loss": 0.3092, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.2035375712105001e-08, | |
| "loss": 0.3142, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.946690433526584e-11, | |
| "loss": 0.3105, | |
| "step": 3500 | |
| } | |
| ], | |
| "max_steps": 3500, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.12189130145792e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |