| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.1, |
| "eval_steps": 2000, |
| "global_step": 2000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "grad_norm": 6.03125, |
| "learning_rate": 3.0000000000000004e-07, |
| "loss": 0.5688, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.0, |
| "grad_norm": 5.5, |
| "learning_rate": 6.000000000000001e-07, |
| "loss": 0.4803, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.0, |
| "grad_norm": 4.6875, |
| "learning_rate": 9e-07, |
| "loss": 0.6168, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.0, |
| "grad_norm": 6.40625, |
| "learning_rate": 1.2000000000000002e-06, |
| "loss": 0.5197, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.0, |
| "grad_norm": 7.0, |
| "learning_rate": 1.5e-06, |
| "loss": 0.5441, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.0, |
| "grad_norm": 4.6875, |
| "learning_rate": 1.8e-06, |
| "loss": 0.6031, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.0, |
| "grad_norm": 5.4375, |
| "learning_rate": 2.1000000000000002e-06, |
| "loss": 0.4926, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.0, |
| "grad_norm": 6.4375, |
| "learning_rate": 2.4000000000000003e-06, |
| "loss": 0.5795, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.0, |
| "grad_norm": 4.75, |
| "learning_rate": 2.7e-06, |
| "loss": 0.5599, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 5.125, |
| "learning_rate": 3e-06, |
| "loss": 0.5279, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 4.96875, |
| "learning_rate": 3.3e-06, |
| "loss": 0.5455, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 7.09375, |
| "learning_rate": 3.6e-06, |
| "loss": 0.5655, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 6.25, |
| "learning_rate": 3.9e-06, |
| "loss": 0.4635, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 4.8125, |
| "learning_rate": 4.2000000000000004e-06, |
| "loss": 0.64, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 3.9375, |
| "learning_rate": 4.5e-06, |
| "loss": 0.4565, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 4.09375, |
| "learning_rate": 4.800000000000001e-06, |
| "loss": 0.5357, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 5.59375, |
| "learning_rate": 5.1e-06, |
| "loss": 0.4885, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 4.9375, |
| "learning_rate": 5.4e-06, |
| "loss": 0.5125, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 5.21875, |
| "learning_rate": 5.7000000000000005e-06, |
| "loss": 0.4882, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 3.625, |
| "learning_rate": 6e-06, |
| "loss": 0.4504, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 6.15625, |
| "learning_rate": 6.3e-06, |
| "loss": 0.404, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 5.1875, |
| "learning_rate": 6.6e-06, |
| "loss": 0.4566, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 5.125, |
| "learning_rate": 6.900000000000001e-06, |
| "loss": 0.4597, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 6.0625, |
| "learning_rate": 7.2e-06, |
| "loss": 0.3678, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 5.09375, |
| "learning_rate": 7.5e-06, |
| "loss": 0.5107, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 4.15625, |
| "learning_rate": 7.8e-06, |
| "loss": 0.524, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 4.78125, |
| "learning_rate": 8.1e-06, |
| "loss": 0.4228, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 4.71875, |
| "learning_rate": 8.400000000000001e-06, |
| "loss": 0.4605, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 6.4375, |
| "learning_rate": 8.7e-06, |
| "loss": 0.4211, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.01, |
| "grad_norm": 4.84375, |
| "learning_rate": 9e-06, |
| "loss": 0.438, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 4.375, |
| "learning_rate": 9.3e-06, |
| "loss": 0.4466, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 3.265625, |
| "learning_rate": 9.600000000000001e-06, |
| "loss": 0.3544, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 3.703125, |
| "learning_rate": 9.9e-06, |
| "loss": 0.3532, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 4.25, |
| "learning_rate": 1.02e-05, |
| "loss": 0.355, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 4.25, |
| "learning_rate": 1.05e-05, |
| "loss": 0.3628, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 3.9375, |
| "learning_rate": 1.08e-05, |
| "loss": 0.3914, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 6.3125, |
| "learning_rate": 1.11e-05, |
| "loss": 0.4033, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 6.625, |
| "learning_rate": 1.1400000000000001e-05, |
| "loss": 0.329, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 4.75, |
| "learning_rate": 1.1700000000000001e-05, |
| "loss": 0.3501, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 6.21875, |
| "learning_rate": 1.2e-05, |
| "loss": 0.3402, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 3.703125, |
| "learning_rate": 1.2299999999999999e-05, |
| "loss": 0.3807, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 4.4375, |
| "learning_rate": 1.26e-05, |
| "loss": 0.38, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 5.1875, |
| "learning_rate": 1.29e-05, |
| "loss": 0.3419, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 5.96875, |
| "learning_rate": 1.32e-05, |
| "loss": 0.4231, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 4.4375, |
| "learning_rate": 1.3500000000000001e-05, |
| "loss": 0.3869, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 7.375, |
| "learning_rate": 1.3800000000000002e-05, |
| "loss": 0.3628, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 3.59375, |
| "learning_rate": 1.4099999999999999e-05, |
| "loss": 0.3268, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 4.15625, |
| "learning_rate": 1.44e-05, |
| "loss": 0.4242, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.02, |
| "grad_norm": 5.40625, |
| "learning_rate": 1.47e-05, |
| "loss": 0.3388, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 4.46875, |
| "learning_rate": 1.5e-05, |
| "loss": 0.3188, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 5.03125, |
| "learning_rate": 1.4984210526315789e-05, |
| "loss": 0.3138, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 3.9375, |
| "learning_rate": 1.496842105263158e-05, |
| "loss": 0.3381, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 7.0625, |
| "learning_rate": 1.4952631578947368e-05, |
| "loss": 0.3579, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 5.40625, |
| "learning_rate": 1.4936842105263158e-05, |
| "loss": 0.308, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 5.125, |
| "learning_rate": 1.4921052631578947e-05, |
| "loss": 0.4211, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 4.375, |
| "learning_rate": 1.4905263157894737e-05, |
| "loss": 0.3596, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 7.4375, |
| "learning_rate": 1.4889473684210526e-05, |
| "loss": 0.3154, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 3.28125, |
| "learning_rate": 1.4873684210526315e-05, |
| "loss": 0.3132, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 5.9375, |
| "learning_rate": 1.4857894736842107e-05, |
| "loss": 0.3773, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 3.484375, |
| "learning_rate": 1.4842105263157895e-05, |
| "loss": 0.3634, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 4.875, |
| "learning_rate": 1.4826315789473686e-05, |
| "loss": 0.2933, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 3.515625, |
| "learning_rate": 1.4810526315789474e-05, |
| "loss": 0.3159, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 4.3125, |
| "learning_rate": 1.4794736842105265e-05, |
| "loss": 0.3357, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 4.875, |
| "learning_rate": 1.4778947368421053e-05, |
| "loss": 0.3478, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 5.0625, |
| "learning_rate": 1.4763157894736842e-05, |
| "loss": 0.3873, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 6.03125, |
| "learning_rate": 1.4747368421052632e-05, |
| "loss": 0.2899, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 4.78125, |
| "learning_rate": 1.4731578947368421e-05, |
| "loss": 0.3353, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 3.0625, |
| "learning_rate": 1.4715789473684211e-05, |
| "loss": 0.3488, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.03, |
| "grad_norm": 8.625, |
| "learning_rate": 1.47e-05, |
| "loss": 0.3456, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 6.125, |
| "learning_rate": 1.468421052631579e-05, |
| "loss": 0.3656, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.0625, |
| "learning_rate": 1.4668421052631579e-05, |
| "loss": 0.3494, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 5.28125, |
| "learning_rate": 1.4652631578947367e-05, |
| "loss": 0.2928, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 7.625, |
| "learning_rate": 1.4636842105263158e-05, |
| "loss": 0.391, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.21875, |
| "learning_rate": 1.4621052631578946e-05, |
| "loss": 0.3042, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 3.625, |
| "learning_rate": 1.4605263157894737e-05, |
| "loss": 0.2822, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.0625, |
| "learning_rate": 1.4589473684210527e-05, |
| "loss": 0.3103, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 6.9375, |
| "learning_rate": 1.4573684210526317e-05, |
| "loss": 0.2982, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.28125, |
| "learning_rate": 1.4557894736842106e-05, |
| "loss": 0.2873, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 11.375, |
| "learning_rate": 1.4542105263157895e-05, |
| "loss": 0.346, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.71875, |
| "learning_rate": 1.4526315789473685e-05, |
| "loss": 0.3055, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.96875, |
| "learning_rate": 1.4510526315789474e-05, |
| "loss": 0.2721, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.28125, |
| "learning_rate": 1.4494736842105264e-05, |
| "loss": 0.2343, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.90625, |
| "learning_rate": 1.4478947368421053e-05, |
| "loss": 0.2855, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 2.875, |
| "learning_rate": 1.4463157894736843e-05, |
| "loss": 0.2685, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.28125, |
| "learning_rate": 1.4447368421052632e-05, |
| "loss": 0.2675, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.84375, |
| "learning_rate": 1.443157894736842e-05, |
| "loss": 0.3448, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 5.5625, |
| "learning_rate": 1.441578947368421e-05, |
| "loss": 0.2843, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 5.4375, |
| "learning_rate": 1.44e-05, |
| "loss": 0.3069, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 4.75, |
| "learning_rate": 1.438421052631579e-05, |
| "loss": 0.2324, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.04, |
| "grad_norm": 5.8125, |
| "learning_rate": 1.4368421052631578e-05, |
| "loss": 0.3389, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 5.96875, |
| "learning_rate": 1.4352631578947369e-05, |
| "loss": 0.3082, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 4.0625, |
| "learning_rate": 1.4336842105263159e-05, |
| "loss": 0.2889, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 3.109375, |
| "learning_rate": 1.4321052631578948e-05, |
| "loss": 0.2842, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 3.96875, |
| "learning_rate": 1.4305263157894738e-05, |
| "loss": 0.2695, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 3.875, |
| "learning_rate": 1.4289473684210527e-05, |
| "loss": 0.2021, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 6.03125, |
| "learning_rate": 1.4273684210526317e-05, |
| "loss": 0.3382, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 5.09375, |
| "learning_rate": 1.4257894736842106e-05, |
| "loss": 0.3721, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 2.921875, |
| "learning_rate": 1.4242105263157896e-05, |
| "loss": 0.2715, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 2.71875, |
| "learning_rate": 1.4226315789473685e-05, |
| "loss": 0.2261, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 6.40625, |
| "learning_rate": 1.4210526315789473e-05, |
| "loss": 0.2718, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 5.09375, |
| "learning_rate": 1.4194736842105264e-05, |
| "loss": 0.355, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 6.0625, |
| "learning_rate": 1.4178947368421052e-05, |
| "loss": 0.3133, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 3.734375, |
| "learning_rate": 1.4163157894736843e-05, |
| "loss": 0.3427, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 5.5625, |
| "learning_rate": 1.4147368421052631e-05, |
| "loss": 0.2373, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 4.53125, |
| "learning_rate": 1.4131578947368422e-05, |
| "loss": 0.3505, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 4.21875, |
| "learning_rate": 1.411578947368421e-05, |
| "loss": 0.3211, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 2.984375, |
| "learning_rate": 1.4099999999999999e-05, |
| "loss": 0.3006, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 3.765625, |
| "learning_rate": 1.408421052631579e-05, |
| "loss": 0.31, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.05, |
| "grad_norm": 5.6875, |
| "learning_rate": 1.406842105263158e-05, |
| "loss": 0.2641, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 9.375, |
| "learning_rate": 1.405263157894737e-05, |
| "loss": 0.3475, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 6.09375, |
| "learning_rate": 1.4036842105263158e-05, |
| "loss": 0.2827, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 5.46875, |
| "learning_rate": 1.4021052631578949e-05, |
| "loss": 0.2408, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 3.84375, |
| "learning_rate": 1.4005263157894737e-05, |
| "loss": 0.2428, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 3.375, |
| "learning_rate": 1.3989473684210526e-05, |
| "loss": 0.2533, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 4.84375, |
| "learning_rate": 1.3973684210526316e-05, |
| "loss": 0.3098, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 3.875, |
| "learning_rate": 1.3957894736842105e-05, |
| "loss": 0.3235, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 4.6875, |
| "learning_rate": 1.3942105263157895e-05, |
| "loss": 0.2959, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 4.125, |
| "learning_rate": 1.3926315789473684e-05, |
| "loss": 0.2873, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 3.515625, |
| "learning_rate": 1.3910526315789474e-05, |
| "loss": 0.3158, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 2.734375, |
| "learning_rate": 1.3894736842105263e-05, |
| "loss": 0.3004, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 4.34375, |
| "learning_rate": 1.3878947368421052e-05, |
| "loss": 0.3596, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 3.453125, |
| "learning_rate": 1.3863157894736842e-05, |
| "loss": 0.2345, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 4.6875, |
| "learning_rate": 1.384736842105263e-05, |
| "loss": 0.3332, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 3.921875, |
| "learning_rate": 1.3831578947368421e-05, |
| "loss": 0.2898, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 3.890625, |
| "learning_rate": 1.3815789473684211e-05, |
| "loss": 0.2854, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 4.03125, |
| "learning_rate": 1.3800000000000002e-05, |
| "loss": 0.2932, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 4.03125, |
| "learning_rate": 1.378421052631579e-05, |
| "loss": 0.3153, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 6.21875, |
| "learning_rate": 1.3768421052631579e-05, |
| "loss": 0.2947, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.06, |
| "grad_norm": 3.234375, |
| "learning_rate": 1.375263157894737e-05, |
| "loss": 0.2211, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 3.15625, |
| "learning_rate": 1.3736842105263158e-05, |
| "loss": 0.296, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 5.21875, |
| "learning_rate": 1.3721052631578948e-05, |
| "loss": 0.3276, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 2.46875, |
| "learning_rate": 1.3705263157894737e-05, |
| "loss": 0.2716, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 4.53125, |
| "learning_rate": 1.3689473684210527e-05, |
| "loss": 0.2471, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 6.5, |
| "learning_rate": 1.3673684210526316e-05, |
| "loss": 0.3197, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 3.953125, |
| "learning_rate": 1.3657894736842106e-05, |
| "loss": 0.275, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 2.90625, |
| "learning_rate": 1.3642105263157895e-05, |
| "loss": 0.2567, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 3.890625, |
| "learning_rate": 1.3626315789473684e-05, |
| "loss": 0.2856, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 8.5, |
| "learning_rate": 1.3610526315789474e-05, |
| "loss": 0.3118, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 2.8125, |
| "learning_rate": 1.3594736842105263e-05, |
| "loss": 0.2669, |
| "step": 1390 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 3.375, |
| "learning_rate": 1.3578947368421053e-05, |
| "loss": 0.2718, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 6.90625, |
| "learning_rate": 1.3563157894736842e-05, |
| "loss": 0.3, |
| "step": 1410 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 7.59375, |
| "learning_rate": 1.3547368421052634e-05, |
| "loss": 0.2698, |
| "step": 1420 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 3.578125, |
| "learning_rate": 1.3531578947368422e-05, |
| "loss": 0.2467, |
| "step": 1430 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 5.09375, |
| "learning_rate": 1.3515789473684211e-05, |
| "loss": 0.2349, |
| "step": 1440 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 7.15625, |
| "learning_rate": 1.3500000000000001e-05, |
| "loss": 0.2565, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 3.375, |
| "learning_rate": 1.348421052631579e-05, |
| "loss": 0.2881, |
| "step": 1460 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 4.9375, |
| "learning_rate": 1.346842105263158e-05, |
| "loss": 0.2777, |
| "step": 1470 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 3.828125, |
| "learning_rate": 1.3452631578947369e-05, |
| "loss": 0.3411, |
| "step": 1480 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 4.28125, |
| "learning_rate": 1.343684210526316e-05, |
| "loss": 0.3102, |
| "step": 1490 |
| }, |
| { |
| "epoch": 0.07, |
| "grad_norm": 3.859375, |
| "learning_rate": 1.3421052631578948e-05, |
| "loss": 0.2269, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 3.71875, |
| "learning_rate": 1.3405263157894736e-05, |
| "loss": 0.2404, |
| "step": 1510 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 5.21875, |
| "learning_rate": 1.3389473684210527e-05, |
| "loss": 0.2708, |
| "step": 1520 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 3.171875, |
| "learning_rate": 1.3373684210526315e-05, |
| "loss": 0.2771, |
| "step": 1530 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 5.34375, |
| "learning_rate": 1.3357894736842106e-05, |
| "loss": 0.3431, |
| "step": 1540 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 4.21875, |
| "learning_rate": 1.3342105263157894e-05, |
| "loss": 0.2593, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 6.375, |
| "learning_rate": 1.3326315789473685e-05, |
| "loss": 0.3764, |
| "step": 1560 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 4.5625, |
| "learning_rate": 1.3310526315789473e-05, |
| "loss": 0.2739, |
| "step": 1570 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 4.46875, |
| "learning_rate": 1.3294736842105262e-05, |
| "loss": 0.3306, |
| "step": 1580 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 3.546875, |
| "learning_rate": 1.3278947368421054e-05, |
| "loss": 0.2687, |
| "step": 1590 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 5.5, |
| "learning_rate": 1.3263157894736843e-05, |
| "loss": 0.288, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 3.390625, |
| "learning_rate": 1.3247368421052633e-05, |
| "loss": 0.2626, |
| "step": 1610 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 6.0, |
| "learning_rate": 1.3231578947368422e-05, |
| "loss": 0.2635, |
| "step": 1620 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 4.1875, |
| "learning_rate": 1.3215789473684212e-05, |
| "loss": 0.2567, |
| "step": 1630 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 4.3125, |
| "learning_rate": 1.32e-05, |
| "loss": 0.2756, |
| "step": 1640 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 6.21875, |
| "learning_rate": 1.318421052631579e-05, |
| "loss": 0.32, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 4.65625, |
| "learning_rate": 1.316842105263158e-05, |
| "loss": 0.2499, |
| "step": 1660 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 6.125, |
| "learning_rate": 1.3152631578947368e-05, |
| "loss": 0.2084, |
| "step": 1670 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 5.0625, |
| "learning_rate": 1.3136842105263159e-05, |
| "loss": 0.2263, |
| "step": 1680 |
| }, |
| { |
| "epoch": 0.08, |
| "grad_norm": 4.71875, |
| "learning_rate": 1.3121052631578947e-05, |
| "loss": 0.3104, |
| "step": 1690 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 7.21875, |
| "learning_rate": 1.3105263157894738e-05, |
| "loss": 0.3948, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 2.921875, |
| "learning_rate": 1.3089473684210526e-05, |
| "loss": 0.2561, |
| "step": 1710 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 4.5, |
| "learning_rate": 1.3073684210526315e-05, |
| "loss": 0.2774, |
| "step": 1720 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 3.328125, |
| "learning_rate": 1.3057894736842105e-05, |
| "loss": 0.2165, |
| "step": 1730 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 3.046875, |
| "learning_rate": 1.3042105263157894e-05, |
| "loss": 0.2819, |
| "step": 1740 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 4.28125, |
| "learning_rate": 1.3026315789473684e-05, |
| "loss": 0.2111, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 2.21875, |
| "learning_rate": 1.3010526315789475e-05, |
| "loss": 0.2293, |
| "step": 1760 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 3.53125, |
| "learning_rate": 1.2994736842105265e-05, |
| "loss": 0.2363, |
| "step": 1770 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 3.296875, |
| "learning_rate": 1.2978947368421054e-05, |
| "loss": 0.3236, |
| "step": 1780 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 4.53125, |
| "learning_rate": 1.2963157894736842e-05, |
| "loss": 0.2785, |
| "step": 1790 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 2.265625, |
| "learning_rate": 1.2947368421052633e-05, |
| "loss": 0.238, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 3.828125, |
| "learning_rate": 1.2931578947368421e-05, |
| "loss": 0.2736, |
| "step": 1810 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 4.65625, |
| "learning_rate": 1.2915789473684212e-05, |
| "loss": 0.3267, |
| "step": 1820 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 4.03125, |
| "learning_rate": 1.29e-05, |
| "loss": 0.2644, |
| "step": 1830 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 4.875, |
| "learning_rate": 1.288421052631579e-05, |
| "loss": 0.2847, |
| "step": 1840 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 4.5625, |
| "learning_rate": 1.2868421052631579e-05, |
| "loss": 0.3246, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 3.328125, |
| "learning_rate": 1.2852631578947368e-05, |
| "loss": 0.2835, |
| "step": 1860 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 2.9375, |
| "learning_rate": 1.2836842105263158e-05, |
| "loss": 0.2773, |
| "step": 1870 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 4.46875, |
| "learning_rate": 1.2821052631578947e-05, |
| "loss": 0.2967, |
| "step": 1880 |
| }, |
| { |
| "epoch": 0.09, |
| "grad_norm": 3.5625, |
| "learning_rate": 1.2805263157894737e-05, |
| "loss": 0.3291, |
| "step": 1890 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 3.875, |
| "learning_rate": 1.2789473684210526e-05, |
| "loss": 0.2619, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 6.15625, |
| "learning_rate": 1.2773684210526316e-05, |
| "loss": 0.2467, |
| "step": 1910 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 5.25, |
| "learning_rate": 1.2757894736842106e-05, |
| "loss": 0.224, |
| "step": 1920 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 7.09375, |
| "learning_rate": 1.2742105263157895e-05, |
| "loss": 0.2697, |
| "step": 1930 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 2.765625, |
| "learning_rate": 1.2726315789473685e-05, |
| "loss": 0.245, |
| "step": 1940 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 4.65625, |
| "learning_rate": 1.2710526315789474e-05, |
| "loss": 0.22, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 2.828125, |
| "learning_rate": 1.2694736842105264e-05, |
| "loss": 0.3149, |
| "step": 1960 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 3.15625, |
| "learning_rate": 1.2678947368421053e-05, |
| "loss": 0.2345, |
| "step": 1970 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 5.0, |
| "learning_rate": 1.2663157894736843e-05, |
| "loss": 0.2053, |
| "step": 1980 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 5.84375, |
| "learning_rate": 1.2647368421052632e-05, |
| "loss": 0.2364, |
| "step": 1990 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 4.0, |
| "learning_rate": 1.263157894736842e-05, |
| "loss": 0.3397, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.1, |
| "eval_loss": 0.28045549988746643, |
| "eval_runtime": 64.869, |
| "eval_samples_per_second": 15.416, |
| "eval_steps_per_second": 15.416, |
| "step": 2000 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 10000, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 2000, |
| "total_flos": 3.227844083712e+16, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
