| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 1776, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.966216216216216e-05, | |
| "loss": 0.426, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.9324324324324324e-05, | |
| "loss": 0.484, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.8986486486486485e-05, | |
| "loss": 0.4457, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5.8648648648648653e-05, | |
| "loss": 0.4083, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5.83445945945946e-05, | |
| "loss": 0.4695, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 5.800675675675676e-05, | |
| "loss": 0.5002, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5.766891891891892e-05, | |
| "loss": 0.4436, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 5.733108108108108e-05, | |
| "loss": 0.3807, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.699324324324325e-05, | |
| "loss": 0.4198, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 5.665540540540541e-05, | |
| "loss": 0.4475, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 5.631756756756757e-05, | |
| "loss": 0.434, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.5979729729729734e-05, | |
| "loss": 0.4036, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.5641891891891896e-05, | |
| "loss": 0.4386, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.530405405405406e-05, | |
| "loss": 0.399, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 5.496621621621622e-05, | |
| "loss": 0.5032, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5.4628378378378386e-05, | |
| "loss": 0.4435, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.429054054054054e-05, | |
| "loss": 0.4947, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 5.39527027027027e-05, | |
| "loss": 0.4708, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 5.361486486486486e-05, | |
| "loss": 0.4644, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 5.3277027027027024e-05, | |
| "loss": 0.4833, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 5.2939189189189186e-05, | |
| "loss": 0.4115, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 5.260135135135135e-05, | |
| "loss": 0.5, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 5.2263513513513515e-05, | |
| "loss": 0.4415, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.1925675675675676e-05, | |
| "loss": 0.4143, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.158783783783784e-05, | |
| "loss": 0.4295, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.125e-05, | |
| "loss": 0.4441, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.091216216216216e-05, | |
| "loss": 0.4273, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.057432432432432e-05, | |
| "loss": 0.4175, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.023648648648648e-05, | |
| "loss": 0.4219, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.989864864864865e-05, | |
| "loss": 0.4526, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.956081081081081e-05, | |
| "loss": 0.4396, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.922297297297297e-05, | |
| "loss": 0.444, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.8885135135135135e-05, | |
| "loss": 0.4236, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.8547297297297296e-05, | |
| "loss": 0.4228, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.820945945945946e-05, | |
| "loss": 0.4055, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.787162162162162e-05, | |
| "loss": 0.3954, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.7533783783783786e-05, | |
| "loss": 0.3883, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.719594594594595e-05, | |
| "loss": 0.4181, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.685810810810811e-05, | |
| "loss": 0.4249, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.652027027027027e-05, | |
| "loss": 0.4619, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.618243243243243e-05, | |
| "loss": 0.4298, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.584459459459459e-05, | |
| "loss": 0.4184, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.550675675675676e-05, | |
| "loss": 0.4062, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.516891891891892e-05, | |
| "loss": 0.3768, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.4831081081081083e-05, | |
| "loss": 0.4605, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.4493243243243245e-05, | |
| "loss": 0.485, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.4155405405405406e-05, | |
| "loss": 0.4169, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.381756756756757e-05, | |
| "loss": 0.4136, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.347972972972973e-05, | |
| "loss": 0.4337, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.31418918918919e-05, | |
| "loss": 0.444, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.280405405405406e-05, | |
| "loss": 0.4044, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.246621621621622e-05, | |
| "loss": 0.3761, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.212837837837838e-05, | |
| "loss": 0.3932, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.179054054054054e-05, | |
| "loss": 0.382, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.14527027027027e-05, | |
| "loss": 0.3858, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.1114864864864864e-05, | |
| "loss": 0.4047, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.077702702702703e-05, | |
| "loss": 0.4456, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.0439189189189194e-05, | |
| "loss": 0.351, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.0101351351351355e-05, | |
| "loss": 0.3687, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8888712871287129, | |
| "eval_loss": 0.4043530225753784, | |
| "eval_runtime": 35.1515, | |
| "eval_samples_per_second": 718.32, | |
| "eval_steps_per_second": 5.633, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.9763513513513516e-05, | |
| "loss": 0.3674, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.942567567567568e-05, | |
| "loss": 0.3655, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.908783783783784e-05, | |
| "loss": 0.4221, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.875e-05, | |
| "loss": 0.3846, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.841216216216217e-05, | |
| "loss": 0.4122, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.807432432432433e-05, | |
| "loss": 0.4057, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.773648648648649e-05, | |
| "loss": 0.4538, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.739864864864865e-05, | |
| "loss": 0.3738, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.706081081081081e-05, | |
| "loss": 0.4286, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.6722972972972974e-05, | |
| "loss": 0.3568, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.638513513513514e-05, | |
| "loss": 0.3664, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.6047297297297304e-05, | |
| "loss": 0.3662, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.5709459459459465e-05, | |
| "loss": 0.4049, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.537162162162162e-05, | |
| "loss": 0.3396, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.503378378378378e-05, | |
| "loss": 0.3986, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.469594594594594e-05, | |
| "loss": 0.4362, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.43581081081081e-05, | |
| "loss": 0.3804, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.402027027027027e-05, | |
| "loss": 0.3586, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.368243243243243e-05, | |
| "loss": 0.3433, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.3344594594594594e-05, | |
| "loss": 0.4387, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.3006756756756755e-05, | |
| "loss": 0.3773, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 3.2668918918918917e-05, | |
| "loss": 0.3687, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 3.233108108108108e-05, | |
| "loss": 0.3975, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 3.199324324324324e-05, | |
| "loss": 0.3489, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 3.165540540540541e-05, | |
| "loss": 0.3892, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.131756756756757e-05, | |
| "loss": 0.3245, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.097972972972973e-05, | |
| "loss": 0.3773, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.064189189189189e-05, | |
| "loss": 0.377, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.0304054054054052e-05, | |
| "loss": 0.3985, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.9966216216216217e-05, | |
| "loss": 0.3336, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.9628378378378378e-05, | |
| "loss": 0.3807, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.929054054054054e-05, | |
| "loss": 0.3827, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.8952702702702704e-05, | |
| "loss": 0.3531, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.8614864864864865e-05, | |
| "loss": 0.3814, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.8277027027027027e-05, | |
| "loss": 0.3634, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.7939189189189188e-05, | |
| "loss": 0.3309, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.7601351351351353e-05, | |
| "loss": 0.3763, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.7263513513513514e-05, | |
| "loss": 0.3521, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.6925675675675675e-05, | |
| "loss": 0.364, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.658783783783784e-05, | |
| "loss": 0.4004, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.625e-05, | |
| "loss": 0.3656, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.5912162162162162e-05, | |
| "loss": 0.3789, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.5574324324324324e-05, | |
| "loss": 0.3441, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.5236486486486488e-05, | |
| "loss": 0.3743, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.489864864864865e-05, | |
| "loss": 0.3844, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.456081081081081e-05, | |
| "loss": 0.4033, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.4222972972972976e-05, | |
| "loss": 0.3749, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.3885135135135137e-05, | |
| "loss": 0.3956, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.3547297297297298e-05, | |
| "loss": 0.3968, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 2.3209459459459463e-05, | |
| "loss": 0.3801, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.2871621621621624e-05, | |
| "loss": 0.321, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.2533783783783785e-05, | |
| "loss": 0.3457, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.2195945945945947e-05, | |
| "loss": 0.3789, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.185810810810811e-05, | |
| "loss": 0.3738, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 2.152027027027027e-05, | |
| "loss": 0.3785, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.118243243243243e-05, | |
| "loss": 0.329, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.0844594594594595e-05, | |
| "loss": 0.3649, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.0506756756756756e-05, | |
| "loss": 0.4189, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 2.0168918918918918e-05, | |
| "loss": 0.3422, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8953267326732673, | |
| "eval_loss": 0.3911304175853729, | |
| "eval_runtime": 34.9199, | |
| "eval_samples_per_second": 723.084, | |
| "eval_steps_per_second": 5.67, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.983108108108108e-05, | |
| "loss": 0.3691, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.9493243243243244e-05, | |
| "loss": 0.3039, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.9155405405405405e-05, | |
| "loss": 0.307, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.8817567567567566e-05, | |
| "loss": 0.3207, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.847972972972973e-05, | |
| "loss": 0.3473, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.8141891891891892e-05, | |
| "loss": 0.3155, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.7804054054054053e-05, | |
| "loss": 0.3217, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.7466216216216218e-05, | |
| "loss": 0.3473, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.712837837837838e-05, | |
| "loss": 0.2966, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.679054054054054e-05, | |
| "loss": 0.3411, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.6452702702702702e-05, | |
| "loss": 0.3876, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.6114864864864866e-05, | |
| "loss": 0.2818, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.5777027027027028e-05, | |
| "loss": 0.351, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.543918918918919e-05, | |
| "loss": 0.3947, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.5101351351351352e-05, | |
| "loss": 0.3157, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.4763513513513515e-05, | |
| "loss": 0.4128, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.4425675675675675e-05, | |
| "loss": 0.3189, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.4087837837837838e-05, | |
| "loss": 0.3791, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.375e-05, | |
| "loss": 0.3299, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.3412162162162162e-05, | |
| "loss": 0.3351, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.3074324324324325e-05, | |
| "loss": 0.3197, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.2736486486486486e-05, | |
| "loss": 0.3191, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.2398648648648649e-05, | |
| "loss": 0.3765, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.2060810810810812e-05, | |
| "loss": 0.3019, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.1722972972972973e-05, | |
| "loss": 0.3662, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 1.1385135135135136e-05, | |
| "loss": 0.3232, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.1047297297297297e-05, | |
| "loss": 0.3149, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 1.070945945945946e-05, | |
| "loss": 0.3516, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1.0371621621621622e-05, | |
| "loss": 0.3339, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1.0033783783783785e-05, | |
| "loss": 0.3046, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 9.695945945945946e-06, | |
| "loss": 0.3794, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 9.358108108108107e-06, | |
| "loss": 0.4017, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 9.02027027027027e-06, | |
| "loss": 0.3851, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 8.682432432432431e-06, | |
| "loss": 0.3449, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 8.344594594594594e-06, | |
| "loss": 0.3558, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 8.006756756756757e-06, | |
| "loss": 0.3176, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 7.668918918918919e-06, | |
| "loss": 0.3305, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 7.331081081081082e-06, | |
| "loss": 0.3171, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 6.993243243243244e-06, | |
| "loss": 0.332, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 6.655405405405406e-06, | |
| "loss": 0.2867, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 6.317567567567567e-06, | |
| "loss": 0.3478, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.979729729729729e-06, | |
| "loss": 0.3609, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 5.6418918918918914e-06, | |
| "loss": 0.3591, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 5.304054054054054e-06, | |
| "loss": 0.3131, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.9662162162162165e-06, | |
| "loss": 0.3224, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.628378378378379e-06, | |
| "loss": 0.3849, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.290540540540541e-06, | |
| "loss": 0.4212, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.952702702702702e-06, | |
| "loss": 0.3889, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.614864864864865e-06, | |
| "loss": 0.3444, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.277027027027027e-06, | |
| "loss": 0.3445, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.9391891891891893e-06, | |
| "loss": 0.3322, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.6013513513513514e-06, | |
| "loss": 0.3587, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 2.263513513513514e-06, | |
| "loss": 0.3127, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.9256756756756756e-06, | |
| "loss": 0.3552, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.587837837837838e-06, | |
| "loss": 0.3377, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.2499999999999999e-06, | |
| "loss": 0.3269, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 9.121621621621622e-07, | |
| "loss": 0.3098, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.743243243243243e-07, | |
| "loss": 0.3506, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.3648648648648647e-07, | |
| "loss": 0.3808, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8971089108910891, | |
| "eval_loss": 0.38559553027153015, | |
| "eval_runtime": 33.8726, | |
| "eval_samples_per_second": 745.441, | |
| "eval_steps_per_second": 5.845, | |
| "step": 1776 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 1776, | |
| "total_flos": 1.76256801415296e+19, | |
| "train_loss": 0.3829323179549999, | |
| "train_runtime": 916.6333, | |
| "train_samples_per_second": 247.918, | |
| "train_steps_per_second": 1.938 | |
| } | |
| ], | |
| "max_steps": 1776, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.76256801415296e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |