| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.4657383657017222, | |
| "global_step": 8000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.999389275680958e-05, | |
| "loss": 0.5939, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.993892756809577e-05, | |
| "loss": 0.7863, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 9.987785513619152e-05, | |
| "loss": 0.7268, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.98167827042873e-05, | |
| "loss": 0.779, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.975571027238305e-05, | |
| "loss": 0.7798, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.969463784047881e-05, | |
| "loss": 0.7235, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.963356540857457e-05, | |
| "loss": 0.7335, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.957249297667034e-05, | |
| "loss": 0.7096, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.95114205447661e-05, | |
| "loss": 0.6699, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.945034811286185e-05, | |
| "loss": 0.6999, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.938927568095762e-05, | |
| "loss": 0.7156, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.932820324905339e-05, | |
| "loss": 0.6295, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.926713081714914e-05, | |
| "loss": 0.6705, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.92060583852449e-05, | |
| "loss": 0.7581, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.914498595334066e-05, | |
| "loss": 0.6588, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.908391352143643e-05, | |
| "loss": 0.7137, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.90228410895322e-05, | |
| "loss": 0.7107, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.896176865762795e-05, | |
| "loss": 0.6286, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.890069622572371e-05, | |
| "loss": 0.8044, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.883962379381948e-05, | |
| "loss": 0.6621, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.877855136191524e-05, | |
| "loss": 0.6852, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.8717478930011e-05, | |
| "loss": 0.5985, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.865640649810675e-05, | |
| "loss": 0.7163, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.859533406620252e-05, | |
| "loss": 0.7209, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.853426163429828e-05, | |
| "loss": 0.6967, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.847318920239404e-05, | |
| "loss": 0.6473, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.841211677048981e-05, | |
| "loss": 0.6732, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.835104433858556e-05, | |
| "loss": 0.7574, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.828997190668134e-05, | |
| "loss": 0.6736, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.822889947477709e-05, | |
| "loss": 0.7509, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.816782704287285e-05, | |
| "loss": 0.5934, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.810675461096861e-05, | |
| "loss": 0.6957, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.804568217906438e-05, | |
| "loss": 0.6866, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.798460974716014e-05, | |
| "loss": 0.6404, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.792353731525589e-05, | |
| "loss": 0.5957, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.786246488335166e-05, | |
| "loss": 0.6305, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.780139245144743e-05, | |
| "loss": 0.6549, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.774032001954318e-05, | |
| "loss": 0.6323, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.767924758763895e-05, | |
| "loss": 0.5348, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.761817515573471e-05, | |
| "loss": 0.6242, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.755710272383046e-05, | |
| "loss": 0.6373, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.749603029192624e-05, | |
| "loss": 0.6608, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.743495786002199e-05, | |
| "loss": 0.7526, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.737388542811775e-05, | |
| "loss": 0.5854, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.731281299621352e-05, | |
| "loss": 0.4817, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.725174056430928e-05, | |
| "loss": 0.5875, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.719066813240504e-05, | |
| "loss": 0.6753, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.712959570050079e-05, | |
| "loss": 0.5999, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.706852326859656e-05, | |
| "loss": 0.6461, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.700745083669232e-05, | |
| "loss": 0.5637, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.694637840478808e-05, | |
| "loss": 0.67, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_loss": 0.5656798481941223, | |
| "eval_runtime": 8.0823, | |
| "eval_samples_per_second": 54.316, | |
| "eval_steps_per_second": 6.805, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.688530597288385e-05, | |
| "loss": 0.5756, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.68242335409796e-05, | |
| "loss": 0.6361, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.676316110907538e-05, | |
| "loss": 0.6295, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.670208867717113e-05, | |
| "loss": 0.6445, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.664101624526689e-05, | |
| "loss": 0.6181, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.657994381336265e-05, | |
| "loss": 0.6923, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.65188713814584e-05, | |
| "loss": 0.6842, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.645779894955418e-05, | |
| "loss": 0.6028, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.639672651764993e-05, | |
| "loss": 0.6811, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.63356540857457e-05, | |
| "loss": 0.6204, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.627458165384146e-05, | |
| "loss": 0.7021, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.621350922193722e-05, | |
| "loss": 0.4966, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.615243679003299e-05, | |
| "loss": 0.538, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.609136435812875e-05, | |
| "loss": 0.4867, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.60302919262245e-05, | |
| "loss": 0.6774, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.596921949432028e-05, | |
| "loss": 0.604, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.590814706241603e-05, | |
| "loss": 0.6152, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.584707463051179e-05, | |
| "loss": 0.6309, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.578600219860756e-05, | |
| "loss": 0.6678, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.572492976670332e-05, | |
| "loss": 0.6345, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.566385733479908e-05, | |
| "loss": 0.5487, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.560278490289483e-05, | |
| "loss": 0.7764, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.55417124709906e-05, | |
| "loss": 0.6665, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.548064003908636e-05, | |
| "loss": 0.701, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.541956760718212e-05, | |
| "loss": 0.5862, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.535849517527789e-05, | |
| "loss": 0.7384, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.529742274337364e-05, | |
| "loss": 0.5901, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.52363503114694e-05, | |
| "loss": 0.653, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.517527787956517e-05, | |
| "loss": 0.5761, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.511420544766093e-05, | |
| "loss": 0.6226, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.505313301575669e-05, | |
| "loss": 0.6058, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.499206058385244e-05, | |
| "loss": 0.6013, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.493098815194822e-05, | |
| "loss": 0.697, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.486991572004398e-05, | |
| "loss": 0.6018, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.480884328813973e-05, | |
| "loss": 0.6812, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.47477708562355e-05, | |
| "loss": 0.6619, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.468669842433126e-05, | |
| "loss": 0.6633, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.462562599242703e-05, | |
| "loss": 0.5281, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.456455356052279e-05, | |
| "loss": 0.6583, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.450348112861854e-05, | |
| "loss": 0.6724, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.444240869671432e-05, | |
| "loss": 0.5381, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.438133626481007e-05, | |
| "loss": 0.7141, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.432026383290583e-05, | |
| "loss": 0.6254, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.42591914010016e-05, | |
| "loss": 0.6187, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.419811896909735e-05, | |
| "loss": 0.5858, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.413704653719312e-05, | |
| "loss": 0.5568, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.407597410528887e-05, | |
| "loss": 0.6704, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.401490167338464e-05, | |
| "loss": 0.6384, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.39538292414804e-05, | |
| "loss": 0.6731, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.389275680957616e-05, | |
| "loss": 0.6543, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_loss": 0.5430613160133362, | |
| "eval_runtime": 8.0616, | |
| "eval_samples_per_second": 54.456, | |
| "eval_steps_per_second": 6.823, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.383168437767193e-05, | |
| "loss": 0.6808, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.377061194576768e-05, | |
| "loss": 0.7629, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.370953951386344e-05, | |
| "loss": 0.7063, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.364846708195922e-05, | |
| "loss": 0.6896, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.358739465005497e-05, | |
| "loss": 0.6373, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.352632221815073e-05, | |
| "loss": 0.5895, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.346524978624648e-05, | |
| "loss": 0.6133, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.340417735434226e-05, | |
| "loss": 0.5843, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.334310492243802e-05, | |
| "loss": 0.586, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.328203249053377e-05, | |
| "loss": 0.6214, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.322096005862954e-05, | |
| "loss": 0.6515, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.315988762672529e-05, | |
| "loss": 0.5755, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.309881519482107e-05, | |
| "loss": 0.5883, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.303774276291683e-05, | |
| "loss": 0.6867, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.297667033101258e-05, | |
| "loss": 0.6216, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.291559789910834e-05, | |
| "loss": 0.6517, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.285452546720411e-05, | |
| "loss": 0.5527, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.279345303529987e-05, | |
| "loss": 0.6887, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.273238060339563e-05, | |
| "loss": 0.6357, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.267130817149138e-05, | |
| "loss": 0.5963, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.261023573958716e-05, | |
| "loss": 0.5073, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.254916330768291e-05, | |
| "loss": 0.664, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.248809087577868e-05, | |
| "loss": 0.6244, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.242701844387444e-05, | |
| "loss": 0.539, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.23659460119702e-05, | |
| "loss": 0.6082, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.230487358006597e-05, | |
| "loss": 0.5748, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.224380114816172e-05, | |
| "loss": 0.6449, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.218272871625748e-05, | |
| "loss": 0.4955, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.212165628435326e-05, | |
| "loss": 0.6673, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.206058385244901e-05, | |
| "loss": 0.7262, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.199951142054477e-05, | |
| "loss": 0.6273, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.193843898864052e-05, | |
| "loss": 0.7038, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.187736655673629e-05, | |
| "loss": 0.6306, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.181629412483206e-05, | |
| "loss": 0.5542, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.175522169292781e-05, | |
| "loss": 0.6046, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.169414926102358e-05, | |
| "loss": 0.5442, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.163307682911934e-05, | |
| "loss": 0.6006, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.15720043972151e-05, | |
| "loss": 0.5902, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.151093196531087e-05, | |
| "loss": 0.5492, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.144985953340662e-05, | |
| "loss": 0.567, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.138878710150238e-05, | |
| "loss": 0.6177, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.132771466959815e-05, | |
| "loss": 0.526, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.126664223769391e-05, | |
| "loss": 0.6082, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.120556980578967e-05, | |
| "loss": 0.6598, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.114449737388542e-05, | |
| "loss": 0.6655, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.10834249419812e-05, | |
| "loss": 0.6724, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.102235251007695e-05, | |
| "loss": 0.5951, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.096128007817272e-05, | |
| "loss": 0.6242, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.090020764626848e-05, | |
| "loss": 0.5449, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.083913521436423e-05, | |
| "loss": 0.7236, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_loss": 0.5334087610244751, | |
| "eval_runtime": 8.0755, | |
| "eval_samples_per_second": 54.362, | |
| "eval_steps_per_second": 6.811, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.077806278246001e-05, | |
| "loss": 0.734, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.071699035055576e-05, | |
| "loss": 0.6451, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.065591791865152e-05, | |
| "loss": 0.6571, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.059484548674729e-05, | |
| "loss": 0.5855, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.053377305484305e-05, | |
| "loss": 0.5928, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.047270062293881e-05, | |
| "loss": 0.7069, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.041162819103458e-05, | |
| "loss": 0.5663, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.035055575913033e-05, | |
| "loss": 0.5971, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.02894833272261e-05, | |
| "loss": 0.6486, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.022841089532185e-05, | |
| "loss": 0.5543, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.016733846341762e-05, | |
| "loss": 0.544, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.010626603151338e-05, | |
| "loss": 0.6425, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.004519359960915e-05, | |
| "loss": 0.6209, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.998412116770491e-05, | |
| "loss": 0.5579, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.992304873580066e-05, | |
| "loss": 0.6749, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.986197630389642e-05, | |
| "loss": 0.589, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.980090387199219e-05, | |
| "loss": 0.5611, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.973983144008795e-05, | |
| "loss": 0.6156, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.967875900818371e-05, | |
| "loss": 0.6266, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.961768657627946e-05, | |
| "loss": 0.5836, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.955661414437523e-05, | |
| "loss": 0.5763, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.949554171247099e-05, | |
| "loss": 0.5341, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.943446928056676e-05, | |
| "loss": 0.5345, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.937339684866252e-05, | |
| "loss": 0.5139, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.931232441675827e-05, | |
| "loss": 0.6167, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.925125198485405e-05, | |
| "loss": 0.5927, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.91901795529498e-05, | |
| "loss": 0.6403, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.912910712104556e-05, | |
| "loss": 0.6145, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.906803468914132e-05, | |
| "loss": 0.6017, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.900696225723709e-05, | |
| "loss": 0.6891, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.894588982533285e-05, | |
| "loss": 0.579, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.888481739342862e-05, | |
| "loss": 0.5471, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.882374496152437e-05, | |
| "loss": 0.6414, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.876267252962014e-05, | |
| "loss": 0.5626, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.87016000977159e-05, | |
| "loss": 0.5196, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.864052766581166e-05, | |
| "loss": 0.6533, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.857945523390742e-05, | |
| "loss": 0.5923, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.851838280200317e-05, | |
| "loss": 0.625, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.845731037009895e-05, | |
| "loss": 0.6152, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.83962379381947e-05, | |
| "loss": 0.649, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.833516550629046e-05, | |
| "loss": 0.6156, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.827409307438623e-05, | |
| "loss": 0.4753, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.821302064248199e-05, | |
| "loss": 0.5705, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.815194821057775e-05, | |
| "loss": 0.61, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.80908757786735e-05, | |
| "loss": 0.6118, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.802980334676927e-05, | |
| "loss": 0.6332, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.796873091486503e-05, | |
| "loss": 0.6628, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.79076584829608e-05, | |
| "loss": 0.4956, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.784658605105656e-05, | |
| "loss": 0.6432, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.778551361915231e-05, | |
| "loss": 0.5211, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "eval_loss": 0.5171996355056763, | |
| "eval_runtime": 8.0856, | |
| "eval_samples_per_second": 54.294, | |
| "eval_steps_per_second": 6.802, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.772444118724809e-05, | |
| "loss": 0.592, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.766336875534385e-05, | |
| "loss": 0.5612, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.76022963234396e-05, | |
| "loss": 0.564, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 8.754122389153536e-05, | |
| "loss": 0.5323, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.748015145963113e-05, | |
| "loss": 0.5045, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.741907902772689e-05, | |
| "loss": 0.6624, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.735800659582266e-05, | |
| "loss": 0.687, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.72969341639184e-05, | |
| "loss": 0.6122, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.723586173201417e-05, | |
| "loss": 0.6526, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.717478930010993e-05, | |
| "loss": 0.6355, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.71137168682057e-05, | |
| "loss": 0.6384, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.705264443630146e-05, | |
| "loss": 0.5524, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.699157200439721e-05, | |
| "loss": 0.6293, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.693049957249299e-05, | |
| "loss": 0.7045, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.686942714058874e-05, | |
| "loss": 0.6872, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.68083547086845e-05, | |
| "loss": 0.5042, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.674728227678027e-05, | |
| "loss": 0.5441, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.668620984487603e-05, | |
| "loss": 0.6562, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.66251374129718e-05, | |
| "loss": 0.5497, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.656406498106754e-05, | |
| "loss": 0.6544, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 8.650299254916331e-05, | |
| "loss": 0.6161, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.644192011725909e-05, | |
| "loss": 0.5596, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.638084768535484e-05, | |
| "loss": 0.6018, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.63197752534506e-05, | |
| "loss": 0.5398, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.625870282154635e-05, | |
| "loss": 0.5483, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.619763038964211e-05, | |
| "loss": 0.6067, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.613655795773789e-05, | |
| "loss": 0.5194, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.607548552583364e-05, | |
| "loss": 0.6175, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.60144130939294e-05, | |
| "loss": 0.557, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.595334066202515e-05, | |
| "loss": 0.5102, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.589226823012093e-05, | |
| "loss": 0.5256, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.58311957982167e-05, | |
| "loss": 0.6525, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.577012336631245e-05, | |
| "loss": 0.7317, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.570905093440821e-05, | |
| "loss": 0.5625, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.564797850250397e-05, | |
| "loss": 0.5729, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.558690607059974e-05, | |
| "loss": 0.5448, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.55258336386955e-05, | |
| "loss": 0.5839, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.546476120679125e-05, | |
| "loss": 0.471, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.540368877488703e-05, | |
| "loss": 0.5516, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.534261634298278e-05, | |
| "loss": 0.6811, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.528154391107854e-05, | |
| "loss": 0.5281, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.52204714791743e-05, | |
| "loss": 0.6287, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.515939904727006e-05, | |
| "loss": 0.477, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.509832661536583e-05, | |
| "loss": 0.5652, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.503725418346158e-05, | |
| "loss": 0.6331, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.497618175155735e-05, | |
| "loss": 0.5294, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.491510931965311e-05, | |
| "loss": 0.5269, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.485403688774888e-05, | |
| "loss": 0.5379, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.479296445584464e-05, | |
| "loss": 0.5991, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.473189202394039e-05, | |
| "loss": 0.5668, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "eval_loss": 0.5164313316345215, | |
| "eval_runtime": 8.0662, | |
| "eval_samples_per_second": 54.425, | |
| "eval_steps_per_second": 6.819, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.467081959203615e-05, | |
| "loss": 0.556, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.460974716013193e-05, | |
| "loss": 0.5327, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.454867472822768e-05, | |
| "loss": 0.5802, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.448760229632344e-05, | |
| "loss": 0.6167, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.442652986441921e-05, | |
| "loss": 0.6166, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.436545743251497e-05, | |
| "loss": 0.4968, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.430438500061074e-05, | |
| "loss": 0.5387, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.424331256870649e-05, | |
| "loss": 0.5017, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.418224013680225e-05, | |
| "loss": 0.5476, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.412116770489801e-05, | |
| "loss": 0.5076, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.406009527299378e-05, | |
| "loss": 0.5741, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.399902284108954e-05, | |
| "loss": 0.6178, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.393795040918529e-05, | |
| "loss": 0.5648, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.387687797728105e-05, | |
| "loss": 0.5951, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.381580554537682e-05, | |
| "loss": 0.5612, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.375473311347258e-05, | |
| "loss": 0.6374, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.369366068156835e-05, | |
| "loss": 0.6246, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.36325882496641e-05, | |
| "loss": 0.5242, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.357151581775987e-05, | |
| "loss": 0.4975, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.351044338585562e-05, | |
| "loss": 0.5201, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.344937095395139e-05, | |
| "loss": 0.5522, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.338829852204715e-05, | |
| "loss": 0.6863, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.332722609014292e-05, | |
| "loss": 0.6245, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.326615365823868e-05, | |
| "loss": 0.528, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.320508122633444e-05, | |
| "loss": 0.6258, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.314400879443019e-05, | |
| "loss": 0.5364, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.308293636252597e-05, | |
| "loss": 0.5658, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.302186393062172e-05, | |
| "loss": 0.5565, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.296079149871748e-05, | |
| "loss": 0.6269, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.289971906681325e-05, | |
| "loss": 0.5115, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.2838646634909e-05, | |
| "loss": 0.5789, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.277757420300478e-05, | |
| "loss": 0.6566, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.271650177110053e-05, | |
| "loss": 0.6701, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.265542933919629e-05, | |
| "loss": 0.6259, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.259435690729205e-05, | |
| "loss": 0.5834, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.253328447538782e-05, | |
| "loss": 0.5659, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.247221204348358e-05, | |
| "loss": 0.5832, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.241113961157933e-05, | |
| "loss": 0.6762, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.23500671796751e-05, | |
| "loss": 0.6892, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.228899474777086e-05, | |
| "loss": 0.4719, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.222792231586662e-05, | |
| "loss": 0.5759, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.216684988396239e-05, | |
| "loss": 0.6255, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.210577745205814e-05, | |
| "loss": 0.4147, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.204470502015391e-05, | |
| "loss": 0.5653, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.198363258824966e-05, | |
| "loss": 0.4886, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.192256015634543e-05, | |
| "loss": 0.6251, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.186148772444119e-05, | |
| "loss": 0.581, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.180041529253695e-05, | |
| "loss": 0.5577, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.173934286063272e-05, | |
| "loss": 0.5349, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.167827042872848e-05, | |
| "loss": 0.5232, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_loss": 0.5029483437538147, | |
| "eval_runtime": 8.0668, | |
| "eval_samples_per_second": 54.421, | |
| "eval_steps_per_second": 6.818, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.161719799682423e-05, | |
| "loss": 0.5144, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.155612556492e-05, | |
| "loss": 0.6632, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.149505313301576e-05, | |
| "loss": 0.64, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.143398070111152e-05, | |
| "loss": 0.5647, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.137290826920729e-05, | |
| "loss": 0.6194, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.131183583730304e-05, | |
| "loss": 0.5993, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.125076340539882e-05, | |
| "loss": 0.6968, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.118969097349457e-05, | |
| "loss": 0.6182, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.112861854159033e-05, | |
| "loss": 0.6014, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.106754610968609e-05, | |
| "loss": 0.6745, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.100647367778186e-05, | |
| "loss": 0.5955, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.094540124587762e-05, | |
| "loss": 0.5833, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.088432881397337e-05, | |
| "loss": 0.5243, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.082325638206913e-05, | |
| "loss": 0.5785, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.07621839501649e-05, | |
| "loss": 0.5615, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.070111151826066e-05, | |
| "loss": 0.539, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.064003908635643e-05, | |
| "loss": 0.5422, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.057896665445218e-05, | |
| "loss": 0.5238, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.051789422254794e-05, | |
| "loss": 0.5395, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.045682179064372e-05, | |
| "loss": 0.5067, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.039574935873947e-05, | |
| "loss": 0.5789, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.033467692683523e-05, | |
| "loss": 0.6748, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.027360449493098e-05, | |
| "loss": 0.6444, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.021253206302676e-05, | |
| "loss": 0.546, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.015145963112252e-05, | |
| "loss": 0.5985, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.009038719921827e-05, | |
| "loss": 0.7293, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.002931476731404e-05, | |
| "loss": 0.5963, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.99682423354098e-05, | |
| "loss": 0.6152, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.990716990350556e-05, | |
| "loss": 0.6763, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.984609747160133e-05, | |
| "loss": 0.5406, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.978502503969708e-05, | |
| "loss": 0.5687, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.972395260779285e-05, | |
| "loss": 0.6623, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.96628801758886e-05, | |
| "loss": 0.5378, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.960180774398437e-05, | |
| "loss": 0.5346, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.954073531208013e-05, | |
| "loss": 0.6091, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.947966288017588e-05, | |
| "loss": 0.5683, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.941859044827166e-05, | |
| "loss": 0.5557, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.935751801636741e-05, | |
| "loss": 0.5003, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.929644558446317e-05, | |
| "loss": 0.5878, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.923537315255894e-05, | |
| "loss": 0.5633, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.91743007206547e-05, | |
| "loss": 0.5879, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.911322828875047e-05, | |
| "loss": 0.5606, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.905215585684622e-05, | |
| "loss": 0.7081, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.899108342494198e-05, | |
| "loss": 0.515, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.893001099303776e-05, | |
| "loss": 0.6274, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.886893856113351e-05, | |
| "loss": 0.6913, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.880786612922927e-05, | |
| "loss": 0.5931, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.874679369732502e-05, | |
| "loss": 0.539, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.86857212654208e-05, | |
| "loss": 0.5869, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.862464883351656e-05, | |
| "loss": 0.5203, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "eval_loss": 0.5056378841400146, | |
| "eval_runtime": 8.0659, | |
| "eval_samples_per_second": 54.426, | |
| "eval_steps_per_second": 6.819, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.856357640161231e-05, | |
| "loss": 0.5532, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.850250396970808e-05, | |
| "loss": 0.5349, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.844143153780384e-05, | |
| "loss": 0.5801, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.83803591058996e-05, | |
| "loss": 0.5449, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.831928667399537e-05, | |
| "loss": 0.5825, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.825821424209112e-05, | |
| "loss": 0.5731, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.819714181018688e-05, | |
| "loss": 0.5793, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.813606937828264e-05, | |
| "loss": 0.6599, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.807499694637841e-05, | |
| "loss": 0.6114, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.801392451447417e-05, | |
| "loss": 0.653, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.795285208256992e-05, | |
| "loss": 0.6344, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.78917796506657e-05, | |
| "loss": 0.5734, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.783070721876145e-05, | |
| "loss": 0.6511, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.776963478685721e-05, | |
| "loss": 0.5847, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.770856235495298e-05, | |
| "loss": 0.5053, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.764748992304874e-05, | |
| "loss": 0.6092, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.75864174911445e-05, | |
| "loss": 0.5548, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.752534505924026e-05, | |
| "loss": 0.5341, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.746427262733602e-05, | |
| "loss": 0.5768, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.74032001954318e-05, | |
| "loss": 0.4998, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.734212776352755e-05, | |
| "loss": 0.6269, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.728105533162331e-05, | |
| "loss": 0.5375, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.721998289971907e-05, | |
| "loss": 0.5692, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.715891046781482e-05, | |
| "loss": 0.6039, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.70978380359106e-05, | |
| "loss": 0.6248, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.703676560400635e-05, | |
| "loss": 0.6091, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.697569317210212e-05, | |
| "loss": 0.5721, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.691462074019788e-05, | |
| "loss": 0.6787, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.685354830829364e-05, | |
| "loss": 0.5557, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.679247587638941e-05, | |
| "loss": 0.6122, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.673140344448516e-05, | |
| "loss": 0.6396, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.667033101258092e-05, | |
| "loss": 0.4566, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.660925858067668e-05, | |
| "loss": 0.6009, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.654818614877245e-05, | |
| "loss": 0.5108, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.648711371686821e-05, | |
| "loss": 0.5418, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.642604128496396e-05, | |
| "loss": 0.5436, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.636496885305974e-05, | |
| "loss": 0.5486, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.630389642115549e-05, | |
| "loss": 0.513, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.624282398925125e-05, | |
| "loss": 0.5196, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.618175155734702e-05, | |
| "loss": 0.5243, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.612067912544278e-05, | |
| "loss": 0.5757, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.605960669353855e-05, | |
| "loss": 0.5942, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.599853426163431e-05, | |
| "loss": 0.6109, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.593746182973006e-05, | |
| "loss": 0.5322, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.587638939782582e-05, | |
| "loss": 0.6137, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.581531696592159e-05, | |
| "loss": 0.6209, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.575424453401735e-05, | |
| "loss": 0.5833, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.569317210211311e-05, | |
| "loss": 0.5345, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.563209967020886e-05, | |
| "loss": 0.5388, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.557102723830464e-05, | |
| "loss": 0.5281, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "eval_loss": 0.49683529138565063, | |
| "eval_runtime": 8.0702, | |
| "eval_samples_per_second": 54.398, | |
| "eval_steps_per_second": 6.815, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.550995480640039e-05, | |
| "loss": 0.542, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.544888237449616e-05, | |
| "loss": 0.5696, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.538780994259192e-05, | |
| "loss": 0.5516, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.532673751068768e-05, | |
| "loss": 0.617, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.526566507878345e-05, | |
| "loss": 0.6802, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.52045926468792e-05, | |
| "loss": 0.5774, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.514352021497496e-05, | |
| "loss": 0.449, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.508244778307072e-05, | |
| "loss": 0.5883, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.502137535116649e-05, | |
| "loss": 0.6559, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.496030291926225e-05, | |
| "loss": 0.5434, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.4899230487358e-05, | |
| "loss": 0.5756, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.483815805545377e-05, | |
| "loss": 0.5449, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.477708562354953e-05, | |
| "loss": 0.601, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.47160131916453e-05, | |
| "loss": 0.6555, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.465494075974106e-05, | |
| "loss": 0.5513, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.459386832783681e-05, | |
| "loss": 0.5649, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.453279589593258e-05, | |
| "loss": 0.5595, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.447172346402835e-05, | |
| "loss": 0.636, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.44106510321241e-05, | |
| "loss": 0.5962, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.434957860021986e-05, | |
| "loss": 0.5298, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.428850616831563e-05, | |
| "loss": 0.5074, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.422743373641139e-05, | |
| "loss": 0.4845, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.416636130450715e-05, | |
| "loss": 0.5993, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.41052888726029e-05, | |
| "loss": 0.6404, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.404421644069868e-05, | |
| "loss": 0.661, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.398314400879443e-05, | |
| "loss": 0.5754, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.39220715768902e-05, | |
| "loss": 0.7076, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.386099914498596e-05, | |
| "loss": 0.4918, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.379992671308171e-05, | |
| "loss": 0.4912, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.373885428117749e-05, | |
| "loss": 0.6144, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.367778184927324e-05, | |
| "loss": 0.5915, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.3616709417369e-05, | |
| "loss": 0.5374, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.355563698546476e-05, | |
| "loss": 0.5579, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.349456455356053e-05, | |
| "loss": 0.7185, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.343349212165629e-05, | |
| "loss": 0.5626, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.337241968975204e-05, | |
| "loss": 0.4352, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.33113472578478e-05, | |
| "loss": 0.5116, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.325027482594358e-05, | |
| "loss": 0.5894, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.318920239403933e-05, | |
| "loss": 0.5914, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.31281299621351e-05, | |
| "loss": 0.5836, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.306705753023085e-05, | |
| "loss": 0.6193, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.300598509832662e-05, | |
| "loss": 0.6008, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.294491266642239e-05, | |
| "loss": 0.5211, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.288384023451814e-05, | |
| "loss": 0.609, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.28227678026139e-05, | |
| "loss": 0.6192, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.276169537070967e-05, | |
| "loss": 0.665, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.270062293880543e-05, | |
| "loss": 0.5003, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.26395505069012e-05, | |
| "loss": 0.5028, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.257847807499694e-05, | |
| "loss": 0.525, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.251740564309271e-05, | |
| "loss": 0.5169, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_loss": 0.49463245272636414, | |
| "eval_runtime": 8.0742, | |
| "eval_samples_per_second": 54.37, | |
| "eval_steps_per_second": 6.812, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.245633321118847e-05, | |
| "loss": 0.5328, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.239526077928424e-05, | |
| "loss": 0.5724, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.233418834738e-05, | |
| "loss": 0.683, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.227311591547575e-05, | |
| "loss": 0.5697, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.221204348357153e-05, | |
| "loss": 0.5939, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.215097105166728e-05, | |
| "loss": 0.5325, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.208989861976304e-05, | |
| "loss": 0.5503, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.20288261878588e-05, | |
| "loss": 0.5618, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.196775375595457e-05, | |
| "loss": 0.6228, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.190668132405033e-05, | |
| "loss": 0.5255, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.184560889214608e-05, | |
| "loss": 0.484, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.178453646024185e-05, | |
| "loss": 0.5212, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.172346402833762e-05, | |
| "loss": 0.52, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.166239159643337e-05, | |
| "loss": 0.626, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.160131916452914e-05, | |
| "loss": 0.6418, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.154024673262489e-05, | |
| "loss": 0.4746, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.147917430072065e-05, | |
| "loss": 0.6784, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.141810186881643e-05, | |
| "loss": 0.5306, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.135702943691218e-05, | |
| "loss": 0.5448, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.129595700500794e-05, | |
| "loss": 0.5339, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.12348845731037e-05, | |
| "loss": 0.5129, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.117381214119947e-05, | |
| "loss": 0.5486, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.111273970929523e-05, | |
| "loss": 0.6468, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.105166727739098e-05, | |
| "loss": 0.4958, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.099059484548675e-05, | |
| "loss": 0.6085, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.092952241358251e-05, | |
| "loss": 0.5155, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.086844998167827e-05, | |
| "loss": 0.5969, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.080737754977404e-05, | |
| "loss": 0.5941, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.074630511786979e-05, | |
| "loss": 0.5871, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.068523268596557e-05, | |
| "loss": 0.5531, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.062416025406132e-05, | |
| "loss": 0.4792, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.056308782215708e-05, | |
| "loss": 0.4931, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.050201539025284e-05, | |
| "loss": 0.5715, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.044094295834861e-05, | |
| "loss": 0.5688, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.037987052644437e-05, | |
| "loss": 0.507, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.031879809454012e-05, | |
| "loss": 0.5876, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.025772566263589e-05, | |
| "loss": 0.5448, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.019665323073165e-05, | |
| "loss": 0.5301, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.013558079882741e-05, | |
| "loss": 0.6726, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.007450836692318e-05, | |
| "loss": 0.4992, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.001343593501894e-05, | |
| "loss": 0.5037, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.995236350311469e-05, | |
| "loss": 0.584, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.989129107121047e-05, | |
| "loss": 0.5564, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.983021863930622e-05, | |
| "loss": 0.6197, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.976914620740198e-05, | |
| "loss": 0.5715, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.970807377549775e-05, | |
| "loss": 0.4912, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.964700134359351e-05, | |
| "loss": 0.6008, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.958592891168927e-05, | |
| "loss": 0.6023, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.952485647978502e-05, | |
| "loss": 0.5724, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.946378404788079e-05, | |
| "loss": 0.6114, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "eval_loss": 0.48395612835884094, | |
| "eval_runtime": 8.0713, | |
| "eval_samples_per_second": 54.391, | |
| "eval_steps_per_second": 6.814, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.940271161597655e-05, | |
| "loss": 0.4923, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.934163918407231e-05, | |
| "loss": 0.5899, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.928056675216808e-05, | |
| "loss": 0.5101, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.921949432026383e-05, | |
| "loss": 0.5108, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.915842188835959e-05, | |
| "loss": 0.5938, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.909734945645536e-05, | |
| "loss": 0.6524, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.903627702455112e-05, | |
| "loss": 0.6049, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.897520459264688e-05, | |
| "loss": 0.602, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.891413216074263e-05, | |
| "loss": 0.5693, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.885305972883841e-05, | |
| "loss": 0.4968, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.879198729693418e-05, | |
| "loss": 0.6252, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.873091486502993e-05, | |
| "loss": 0.5054, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.866984243312569e-05, | |
| "loss": 0.5394, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.860877000122145e-05, | |
| "loss": 0.5695, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.854769756931722e-05, | |
| "loss": 0.6331, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.848662513741298e-05, | |
| "loss": 0.5709, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.842555270550873e-05, | |
| "loss": 0.6106, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.836448027360451e-05, | |
| "loss": 0.6309, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.830340784170026e-05, | |
| "loss": 0.5864, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.824233540979602e-05, | |
| "loss": 0.4849, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.818126297789179e-05, | |
| "loss": 0.4716, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.812019054598755e-05, | |
| "loss": 0.5002, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.805911811408331e-05, | |
| "loss": 0.6034, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.799804568217906e-05, | |
| "loss": 0.5131, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.793697325027483e-05, | |
| "loss": 0.554, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.787590081837059e-05, | |
| "loss": 0.6631, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.781482838646635e-05, | |
| "loss": 0.6101, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.775375595456212e-05, | |
| "loss": 0.5592, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.769268352265787e-05, | |
| "loss": 0.4374, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.763161109075363e-05, | |
| "loss": 0.536, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.75705386588494e-05, | |
| "loss": 0.6093, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.750946622694516e-05, | |
| "loss": 0.6082, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.744839379504092e-05, | |
| "loss": 0.47, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.738732136313667e-05, | |
| "loss": 0.5553, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.732624893123245e-05, | |
| "loss": 0.4803, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.726517649932821e-05, | |
| "loss": 0.6149, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.720410406742397e-05, | |
| "loss": 0.4449, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.714303163551973e-05, | |
| "loss": 0.5756, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.708195920361549e-05, | |
| "loss": 0.6226, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.702088677171126e-05, | |
| "loss": 0.5853, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.695981433980702e-05, | |
| "loss": 0.5568, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.689874190790277e-05, | |
| "loss": 0.5288, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.683766947599853e-05, | |
| "loss": 0.5664, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.67765970440943e-05, | |
| "loss": 0.4867, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.671552461219006e-05, | |
| "loss": 0.6374, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.665445218028583e-05, | |
| "loss": 0.5033, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.659337974838158e-05, | |
| "loss": 0.3433, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.653230731647735e-05, | |
| "loss": 0.547, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.64712348845731e-05, | |
| "loss": 0.5423, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.641016245266887e-05, | |
| "loss": 0.57, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "eval_loss": 0.48337310552597046, | |
| "eval_runtime": 8.0714, | |
| "eval_samples_per_second": 54.39, | |
| "eval_steps_per_second": 6.814, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.634909002076463e-05, | |
| "loss": 0.4685, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.62880175888604e-05, | |
| "loss": 0.4702, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 6.622694515695616e-05, | |
| "loss": 0.4421, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.616587272505191e-05, | |
| "loss": 0.3957, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.610480029314767e-05, | |
| "loss": 0.4499, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.604372786124345e-05, | |
| "loss": 0.387, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.59826554293392e-05, | |
| "loss": 0.4279, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.592158299743496e-05, | |
| "loss": 0.4825, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.586051056553071e-05, | |
| "loss": 0.4426, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.579943813362648e-05, | |
| "loss": 0.4507, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.573836570172225e-05, | |
| "loss": 0.4347, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.5677293269818e-05, | |
| "loss": 0.4578, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.561622083791377e-05, | |
| "loss": 0.48, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.555514840600952e-05, | |
| "loss": 0.4759, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.54940759741053e-05, | |
| "loss": 0.3647, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.543300354220106e-05, | |
| "loss": 0.3988, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.537193111029681e-05, | |
| "loss": 0.5145, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.531085867839257e-05, | |
| "loss": 0.4659, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.524978624648834e-05, | |
| "loss": 0.4831, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.51887138145841e-05, | |
| "loss": 0.5085, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.512764138267987e-05, | |
| "loss": 0.4461, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.506656895077562e-05, | |
| "loss": 0.4365, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.500549651887139e-05, | |
| "loss": 0.5382, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.494442408696714e-05, | |
| "loss": 0.4365, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.48833516550629e-05, | |
| "loss": 0.524, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.482227922315867e-05, | |
| "loss": 0.4897, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.476120679125443e-05, | |
| "loss": 0.3722, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.47001343593502e-05, | |
| "loss": 0.5144, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.463906192744595e-05, | |
| "loss": 0.4821, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.457798949554171e-05, | |
| "loss": 0.4912, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.451691706363748e-05, | |
| "loss": 0.4531, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.445584463173324e-05, | |
| "loss": 0.4936, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.4394772199829e-05, | |
| "loss": 0.4263, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.433369976792475e-05, | |
| "loss": 0.5121, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.427262733602052e-05, | |
| "loss": 0.4662, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.42115549041163e-05, | |
| "loss": 0.4866, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.415048247221204e-05, | |
| "loss": 0.5437, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.408941004030781e-05, | |
| "loss": 0.5342, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.402833760840357e-05, | |
| "loss": 0.4663, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.396726517649934e-05, | |
| "loss": 0.4183, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.39061927445951e-05, | |
| "loss": 0.466, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.384512031269085e-05, | |
| "loss": 0.3704, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.378404788078661e-05, | |
| "loss": 0.4633, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.372297544888238e-05, | |
| "loss": 0.489, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.366190301697814e-05, | |
| "loss": 0.4516, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.36008305850739e-05, | |
| "loss": 0.4578, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.353975815316966e-05, | |
| "loss": 0.5411, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.347868572126542e-05, | |
| "loss": 0.4943, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.341761328936118e-05, | |
| "loss": 0.601, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.335654085745695e-05, | |
| "loss": 0.4512, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "eval_loss": 0.4769096076488495, | |
| "eval_runtime": 8.0563, | |
| "eval_samples_per_second": 54.491, | |
| "eval_steps_per_second": 6.827, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.329546842555271e-05, | |
| "loss": 0.5506, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.323439599364846e-05, | |
| "loss": 0.5073, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.317332356174424e-05, | |
| "loss": 0.4743, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.311225112983999e-05, | |
| "loss": 0.5297, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.305117869793575e-05, | |
| "loss": 0.5185, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.299010626603152e-05, | |
| "loss": 0.5, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.292903383412728e-05, | |
| "loss": 0.4941, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.286796140222304e-05, | |
| "loss": 0.4573, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.28068889703188e-05, | |
| "loss": 0.4418, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.274581653841456e-05, | |
| "loss": 0.5291, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.268474410651033e-05, | |
| "loss": 0.4897, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.262367167460608e-05, | |
| "loss": 0.5277, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.256259924270185e-05, | |
| "loss": 0.4301, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 6.250152681079761e-05, | |
| "loss": 0.4446, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.244045437889338e-05, | |
| "loss": 0.556, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.237938194698914e-05, | |
| "loss": 0.4555, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.231830951508489e-05, | |
| "loss": 0.366, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.225723708318065e-05, | |
| "loss": 0.5686, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 6.219616465127642e-05, | |
| "loss": 0.4858, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.213509221937218e-05, | |
| "loss": 0.5321, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.207401978746794e-05, | |
| "loss": 0.4494, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.20129473555637e-05, | |
| "loss": 0.4595, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.195187492365946e-05, | |
| "loss": 0.613, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 6.189080249175522e-05, | |
| "loss": 0.5001, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.182973005985099e-05, | |
| "loss": 0.4698, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.176865762794675e-05, | |
| "loss": 0.4511, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.17075851960425e-05, | |
| "loss": 0.4767, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.164651276413828e-05, | |
| "loss": 0.4065, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.158544033223404e-05, | |
| "loss": 0.4887, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 6.152436790032979e-05, | |
| "loss": 0.4651, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.146329546842556e-05, | |
| "loss": 0.4558, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.140222303652132e-05, | |
| "loss": 0.4762, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.134115060461708e-05, | |
| "loss": 0.4765, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.128007817271285e-05, | |
| "loss": 0.5861, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 6.12190057408086e-05, | |
| "loss": 0.4383, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.115793330890436e-05, | |
| "loss": 0.5076, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.109686087700012e-05, | |
| "loss": 0.4751, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.103578844509589e-05, | |
| "loss": 0.4864, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.0974716013191645e-05, | |
| "loss": 0.4496, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.09136435812874e-05, | |
| "loss": 0.497, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 6.085257114938317e-05, | |
| "loss": 0.4895, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.0791498717478936e-05, | |
| "loss": 0.4204, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.073042628557469e-05, | |
| "loss": 0.5396, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.066935385367045e-05, | |
| "loss": 0.4727, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.060828142176622e-05, | |
| "loss": 0.5083, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 6.054720898986198e-05, | |
| "loss": 0.4621, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.048613655795774e-05, | |
| "loss": 0.4545, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.04250641260535e-05, | |
| "loss": 0.4201, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.036399169414927e-05, | |
| "loss": 0.4145, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.0302919262245026e-05, | |
| "loss": 0.4919, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "eval_loss": 0.4809638559818268, | |
| "eval_runtime": 8.0729, | |
| "eval_samples_per_second": 54.379, | |
| "eval_steps_per_second": 6.813, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.024184683034079e-05, | |
| "loss": 0.5099, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 6.018077439843655e-05, | |
| "loss": 0.4572, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.0119701966532304e-05, | |
| "loss": 0.4873, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 6.0058629534628074e-05, | |
| "loss": 0.4548, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 5.999755710272383e-05, | |
| "loss": 0.5781, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 5.9936484670819595e-05, | |
| "loss": 0.4808, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 5.987541223891535e-05, | |
| "loss": 0.4657, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.981433980701112e-05, | |
| "loss": 0.545, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.975326737510688e-05, | |
| "loss": 0.4645, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.9692194943202637e-05, | |
| "loss": 0.4517, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.96311225112984e-05, | |
| "loss": 0.5021, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.957005007939417e-05, | |
| "loss": 0.3697, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.950897764748993e-05, | |
| "loss": 0.4351, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.9447905215585685e-05, | |
| "loss": 0.4725, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.938683278368145e-05, | |
| "loss": 0.4526, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.932576035177721e-05, | |
| "loss": 0.539, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.9264687919872976e-05, | |
| "loss": 0.5074, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.920361548796873e-05, | |
| "loss": 0.4153, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.914254305606449e-05, | |
| "loss": 0.4113, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.908147062416026e-05, | |
| "loss": 0.5269, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.9020398192256024e-05, | |
| "loss": 0.5501, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.895932576035178e-05, | |
| "loss": 0.5132, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.889825332844754e-05, | |
| "loss": 0.6271, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.8837180896543295e-05, | |
| "loss": 0.4854, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 5.8776108464639066e-05, | |
| "loss": 0.4618, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 5.871503603273483e-05, | |
| "loss": 0.4802, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 5.8653963600830587e-05, | |
| "loss": 0.4158, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 5.8592891168926344e-05, | |
| "loss": 0.4787, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 5.8531818737022114e-05, | |
| "loss": 0.536, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.847074630511787e-05, | |
| "loss": 0.553, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.8409673873213635e-05, | |
| "loss": 0.4619, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.834860144130939e-05, | |
| "loss": 0.4421, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.828752900940516e-05, | |
| "loss": 0.5026, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.822645657750092e-05, | |
| "loss": 0.4881, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.816538414559668e-05, | |
| "loss": 0.5504, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.810431171369244e-05, | |
| "loss": 0.4825, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.804323928178821e-05, | |
| "loss": 0.4211, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.798216684988397e-05, | |
| "loss": 0.4565, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.7921094417979725e-05, | |
| "loss": 0.4079, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.786002198607549e-05, | |
| "loss": 0.4776, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.7798949554171245e-05, | |
| "loss": 0.4799, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.7737877122267016e-05, | |
| "loss": 0.4955, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.767680469036277e-05, | |
| "loss": 0.4251, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.761573225845853e-05, | |
| "loss": 0.5322, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.7554659826554294e-05, | |
| "loss": 0.515, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.7493587394650064e-05, | |
| "loss": 0.5156, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.743251496274582e-05, | |
| "loss": 0.4609, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.737144253084158e-05, | |
| "loss": 0.5192, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.7310370098937335e-05, | |
| "loss": 0.466, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.7249297667033106e-05, | |
| "loss": 0.4691, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "eval_loss": 0.4783366322517395, | |
| "eval_runtime": 8.0477, | |
| "eval_samples_per_second": 54.55, | |
| "eval_steps_per_second": 6.834, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.718822523512887e-05, | |
| "loss": 0.486, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.7127152803224626e-05, | |
| "loss": 0.4048, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.706608037132038e-05, | |
| "loss": 0.5411, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.7005007939416154e-05, | |
| "loss": 0.4338, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.694393550751191e-05, | |
| "loss": 0.4579, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.6882863075607675e-05, | |
| "loss": 0.4766, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.682179064370343e-05, | |
| "loss": 0.4878, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.67607182117992e-05, | |
| "loss": 0.443, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.669964577989496e-05, | |
| "loss": 0.4268, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.663857334799072e-05, | |
| "loss": 0.431, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.657750091608648e-05, | |
| "loss": 0.3938, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 5.651642848418224e-05, | |
| "loss": 0.4443, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.645535605227801e-05, | |
| "loss": 0.4496, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.6394283620373764e-05, | |
| "loss": 0.4752, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.633321118846953e-05, | |
| "loss": 0.4711, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.6272138756565285e-05, | |
| "loss": 0.4501, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.6211066324661056e-05, | |
| "loss": 0.4938, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.614999389275681e-05, | |
| "loss": 0.4515, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.608892146085257e-05, | |
| "loss": 0.4438, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.602784902894833e-05, | |
| "loss": 0.4232, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.5966776597044104e-05, | |
| "loss": 0.4976, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.590570416513986e-05, | |
| "loss": 0.5391, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.584463173323562e-05, | |
| "loss": 0.4602, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.578355930133138e-05, | |
| "loss": 0.5158, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.5722486869427145e-05, | |
| "loss": 0.5111, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.566141443752291e-05, | |
| "loss": 0.5023, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.5600342005618666e-05, | |
| "loss": 0.4453, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.553926957371442e-05, | |
| "loss": 0.5204, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.547819714181019e-05, | |
| "loss": 0.4364, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.541712470990596e-05, | |
| "loss": 0.4686, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.5356052278001714e-05, | |
| "loss": 0.4584, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.529497984609747e-05, | |
| "loss": 0.507, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.523390741419323e-05, | |
| "loss": 0.4225, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 5.5172834982289e-05, | |
| "loss": 0.4276, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.511176255038476e-05, | |
| "loss": 0.4363, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.505069011848052e-05, | |
| "loss": 0.4306, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.498961768657628e-05, | |
| "loss": 0.5049, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.492854525467205e-05, | |
| "loss": 0.428, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 5.4867472822767804e-05, | |
| "loss": 0.5013, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.480640039086357e-05, | |
| "loss": 0.4242, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.4745327958959325e-05, | |
| "loss": 0.6251, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.4684255527055095e-05, | |
| "loss": 0.4377, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.462318309515085e-05, | |
| "loss": 0.4161, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.456211066324661e-05, | |
| "loss": 0.526, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 5.450103823134237e-05, | |
| "loss": 0.4344, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.443996579943813e-05, | |
| "loss": 0.5485, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.43788933675339e-05, | |
| "loss": 0.4343, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.431782093562966e-05, | |
| "loss": 0.4332, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.425674850372542e-05, | |
| "loss": 0.4689, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 5.419567607182118e-05, | |
| "loss": 0.4444, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "eval_loss": 0.47835543751716614, | |
| "eval_runtime": 8.0988, | |
| "eval_samples_per_second": 54.206, | |
| "eval_steps_per_second": 6.791, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.413460363991695e-05, | |
| "loss": 0.4557, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.4073531208012706e-05, | |
| "loss": 0.4997, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.401245877610846e-05, | |
| "loss": 0.46, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.395138634420423e-05, | |
| "loss": 0.4962, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.38903139123e-05, | |
| "loss": 0.4572, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.3829241480395754e-05, | |
| "loss": 0.5327, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.376816904849151e-05, | |
| "loss": 0.4548, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.370709661658727e-05, | |
| "loss": 0.5156, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.364602418468304e-05, | |
| "loss": 0.4758, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.35849517527788e-05, | |
| "loss": 0.4207, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 5.352387932087456e-05, | |
| "loss": 0.5442, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.3462806888970316e-05, | |
| "loss": 0.5797, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.340173445706609e-05, | |
| "loss": 0.4804, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.3340662025161844e-05, | |
| "loss": 0.4421, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.327958959325761e-05, | |
| "loss": 0.6741, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 5.3218517161353365e-05, | |
| "loss": 0.5357, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.315744472944912e-05, | |
| "loss": 0.4625, | |
| "step": 7670 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.309637229754489e-05, | |
| "loss": 0.4209, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.3035299865640656e-05, | |
| "loss": 0.4564, | |
| "step": 7690 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.297422743373641e-05, | |
| "loss": 0.458, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.291315500183217e-05, | |
| "loss": 0.4381, | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 5.285208256992794e-05, | |
| "loss": 0.4098, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.27910101380237e-05, | |
| "loss": 0.4997, | |
| "step": 7730 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.272993770611946e-05, | |
| "loss": 0.4262, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.266886527421522e-05, | |
| "loss": 0.4791, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.260779284231099e-05, | |
| "loss": 0.5229, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.2546720410406746e-05, | |
| "loss": 0.4288, | |
| "step": 7770 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.24856479785025e-05, | |
| "loss": 0.4647, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.2424575546598266e-05, | |
| "loss": 0.5176, | |
| "step": 7790 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.236350311469404e-05, | |
| "loss": 0.5359, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.2302430682789794e-05, | |
| "loss": 0.5106, | |
| "step": 7810 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.224135825088555e-05, | |
| "loss": 0.4668, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 5.2180285818981315e-05, | |
| "loss": 0.4748, | |
| "step": 7830 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.211921338707707e-05, | |
| "loss": 0.4392, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.205814095517284e-05, | |
| "loss": 0.4578, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.19970685232686e-05, | |
| "loss": 0.5055, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.1935996091364356e-05, | |
| "loss": 0.4851, | |
| "step": 7870 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 5.187492365946012e-05, | |
| "loss": 0.4732, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.181385122755589e-05, | |
| "loss": 0.4532, | |
| "step": 7890 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.175277879565165e-05, | |
| "loss": 0.4841, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.1691706363747405e-05, | |
| "loss": 0.4117, | |
| "step": 7910 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.163063393184316e-05, | |
| "loss": 0.5108, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.156956149993893e-05, | |
| "loss": 0.5498, | |
| "step": 7930 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.1508489068034696e-05, | |
| "loss": 0.4584, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.144741663613045e-05, | |
| "loss": 0.4372, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.138634420422621e-05, | |
| "loss": 0.4935, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.132527177232198e-05, | |
| "loss": 0.4772, | |
| "step": 7970 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.126419934041774e-05, | |
| "loss": 0.461, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 5.12031269085135e-05, | |
| "loss": 0.4542, | |
| "step": 7990 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 5.114205447660926e-05, | |
| "loss": 0.5152, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "eval_loss": 0.471835196018219, | |
| "eval_runtime": 8.0443, | |
| "eval_samples_per_second": 54.573, | |
| "eval_steps_per_second": 6.837, | |
| "step": 8000 | |
| } | |
| ], | |
| "max_steps": 16374, | |
| "num_train_epochs": 3, | |
| "total_flos": 4871663124480000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |