| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.7172314864622557, | |
| "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": 5.21875, | |
| "learning_rate": 3.0000000000000004e-08, | |
| "loss": 1.4288, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 9.25, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 1.5985, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 6.90625, | |
| "learning_rate": 9e-08, | |
| "loss": 1.3356, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 1.3789, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 6.28125, | |
| "learning_rate": 1.5000000000000002e-07, | |
| "loss": 1.362, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 7.8125, | |
| "learning_rate": 1.8e-07, | |
| "loss": 1.3439, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.65625, | |
| "learning_rate": 2.1000000000000003e-07, | |
| "loss": 1.3946, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 5.5, | |
| "learning_rate": 2.4000000000000003e-07, | |
| "loss": 1.3411, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 2.7e-07, | |
| "loss": 1.3421, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 1.4355, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 6.59375, | |
| "learning_rate": 3.3e-07, | |
| "loss": 1.5107, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 7.09375, | |
| "learning_rate": 3.6e-07, | |
| "loss": 1.2753, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 3.9e-07, | |
| "loss": 1.3473, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 1.4292, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 7.875, | |
| "learning_rate": 4.5e-07, | |
| "loss": 1.4616, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.8125, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 1.3935, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 5.625, | |
| "learning_rate": 5.100000000000001e-07, | |
| "loss": 1.4381, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 5.4e-07, | |
| "loss": 1.3679, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 7.375, | |
| "learning_rate": 5.7e-07, | |
| "loss": 1.4046, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 7.0, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 1.3214, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.59375, | |
| "learning_rate": 6.3e-07, | |
| "loss": 1.328, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.25, | |
| "learning_rate": 6.6e-07, | |
| "loss": 1.3994, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 7.375, | |
| "learning_rate": 6.900000000000001e-07, | |
| "loss": 1.3372, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 7.2e-07, | |
| "loss": 1.4327, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 7.5e-07, | |
| "loss": 1.3006, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 8.5625, | |
| "learning_rate": 7.8e-07, | |
| "loss": 1.4955, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 8.100000000000001e-07, | |
| "loss": 1.2202, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.09375, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 1.3608, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 6.75, | |
| "learning_rate": 8.699999999999999e-07, | |
| "loss": 1.3799, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 9e-07, | |
| "loss": 1.3552, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 5.90625, | |
| "learning_rate": 9.3e-07, | |
| "loss": 1.4482, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 4.53125, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 1.4208, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 6.09375, | |
| "learning_rate": 9.9e-07, | |
| "loss": 1.373, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 6.40625, | |
| "learning_rate": 1.0200000000000002e-06, | |
| "loss": 1.3638, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.05e-06, | |
| "loss": 1.4984, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 4.25, | |
| "learning_rate": 1.08e-06, | |
| "loss": 1.4193, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 7.5625, | |
| "learning_rate": 1.11e-06, | |
| "loss": 1.4644, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 3.625, | |
| "learning_rate": 1.14e-06, | |
| "loss": 1.2974, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 7.0, | |
| "learning_rate": 1.17e-06, | |
| "loss": 1.4013, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 1.5707, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 6.5, | |
| "learning_rate": 1.2299999999999999e-06, | |
| "loss": 1.4111, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 1.26e-06, | |
| "loss": 1.3019, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 7.625, | |
| "learning_rate": 1.29e-06, | |
| "loss": 1.4128, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 1.32e-06, | |
| "loss": 1.2438, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 5.25, | |
| "learning_rate": 1.35e-06, | |
| "loss": 1.3703, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.3800000000000001e-06, | |
| "loss": 1.3877, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 1.41e-06, | |
| "loss": 1.3418, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 7.09375, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.3734, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 3.34375, | |
| "learning_rate": 1.4700000000000001e-06, | |
| "loss": 1.3055, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.5e-06, | |
| "loss": 1.4607, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 3.90625, | |
| "learning_rate": 1.498421052631579e-06, | |
| "loss": 1.3326, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 6.75, | |
| "learning_rate": 1.4968421052631579e-06, | |
| "loss": 1.3793, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 7.0625, | |
| "learning_rate": 1.4952631578947368e-06, | |
| "loss": 1.3474, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 1.4936842105263159e-06, | |
| "loss": 1.3504, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 3.953125, | |
| "learning_rate": 1.4921052631578948e-06, | |
| "loss": 1.29, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 5.625, | |
| "learning_rate": 1.4905263157894737e-06, | |
| "loss": 1.3552, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 5.28125, | |
| "learning_rate": 1.4889473684210526e-06, | |
| "loss": 1.2646, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 1.4873684210526315e-06, | |
| "loss": 1.2683, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 7.03125, | |
| "learning_rate": 1.4857894736842106e-06, | |
| "loss": 1.3948, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 7.0, | |
| "learning_rate": 1.4842105263157895e-06, | |
| "loss": 1.2919, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 1.4826315789473684e-06, | |
| "loss": 1.3664, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 5.75, | |
| "learning_rate": 1.4810526315789473e-06, | |
| "loss": 1.3198, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 6.0625, | |
| "learning_rate": 1.4794736842105265e-06, | |
| "loss": 1.3457, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 6.28125, | |
| "learning_rate": 1.4778947368421054e-06, | |
| "loss": 1.2258, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.4763157894736843e-06, | |
| "loss": 1.2654, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 4.5, | |
| "learning_rate": 1.4747368421052632e-06, | |
| "loss": 1.2623, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 3.8125, | |
| "learning_rate": 1.473157894736842e-06, | |
| "loss": 1.3726, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.4715789473684212e-06, | |
| "loss": 1.2427, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.4700000000000001e-06, | |
| "loss": 1.2959, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 6.09375, | |
| "learning_rate": 1.468421052631579e-06, | |
| "loss": 1.2209, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.466842105263158e-06, | |
| "loss": 1.297, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 5.625, | |
| "learning_rate": 1.4652631578947368e-06, | |
| "loss": 1.3002, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 7.96875, | |
| "learning_rate": 1.463684210526316e-06, | |
| "loss": 1.3578, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 6.4375, | |
| "learning_rate": 1.4621052631578946e-06, | |
| "loss": 1.3381, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 6.25, | |
| "learning_rate": 1.4605263157894738e-06, | |
| "loss": 1.2755, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 5.28125, | |
| "learning_rate": 1.4589473684210527e-06, | |
| "loss": 1.3443, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 5.6875, | |
| "learning_rate": 1.4573684210526316e-06, | |
| "loss": 1.3165, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 5.125, | |
| "learning_rate": 1.4557894736842105e-06, | |
| "loss": 1.3626, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 6.625, | |
| "learning_rate": 1.4542105263157894e-06, | |
| "loss": 1.3729, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 5.125, | |
| "learning_rate": 1.4526315789473685e-06, | |
| "loss": 1.3209, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 1.4510526315789474e-06, | |
| "loss": 1.2652, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 3.796875, | |
| "learning_rate": 1.4494736842105263e-06, | |
| "loss": 1.2958, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 6.84375, | |
| "learning_rate": 1.4478947368421052e-06, | |
| "loss": 1.271, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 3.5, | |
| "learning_rate": 1.4463157894736844e-06, | |
| "loss": 1.2985, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 1.4447368421052633e-06, | |
| "loss": 1.3285, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 5.3125, | |
| "learning_rate": 1.4431578947368422e-06, | |
| "loss": 1.2987, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.441578947368421e-06, | |
| "loss": 1.3378, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 5.875, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.2942, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.438421052631579e-06, | |
| "loss": 1.259, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 1.436842105263158e-06, | |
| "loss": 1.2991, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.435263157894737e-06, | |
| "loss": 1.2565, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 1.4336842105263158e-06, | |
| "loss": 1.1737, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.4321052631578947e-06, | |
| "loss": 1.2779, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 5.375, | |
| "learning_rate": 1.4305263157894738e-06, | |
| "loss": 1.3104, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 5.53125, | |
| "learning_rate": 1.4289473684210525e-06, | |
| "loss": 1.2818, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 1.4273684210526317e-06, | |
| "loss": 1.3147, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 1.4257894736842106e-06, | |
| "loss": 1.2595, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 6.125, | |
| "learning_rate": 1.4242105263157895e-06, | |
| "loss": 1.3695, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 1.4226315789473684e-06, | |
| "loss": 1.1634, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 1.4210526315789473e-06, | |
| "loss": 1.2587, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 7.625, | |
| "learning_rate": 1.4194736842105264e-06, | |
| "loss": 1.2994, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.4178947368421053e-06, | |
| "loss": 1.3351, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 1.4163157894736842e-06, | |
| "loss": 1.2404, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 6.0, | |
| "learning_rate": 1.4147368421052631e-06, | |
| "loss": 1.3211, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 4.09375, | |
| "learning_rate": 1.4131578947368422e-06, | |
| "loss": 1.2811, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 5.6875, | |
| "learning_rate": 1.4115789473684211e-06, | |
| "loss": 1.2131, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 4.03125, | |
| "learning_rate": 1.41e-06, | |
| "loss": 1.1713, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.408421052631579e-06, | |
| "loss": 1.2058, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 3.6875, | |
| "learning_rate": 1.4068421052631579e-06, | |
| "loss": 1.3375, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 6.21875, | |
| "learning_rate": 1.405263157894737e-06, | |
| "loss": 1.3032, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 3.359375, | |
| "learning_rate": 1.4036842105263159e-06, | |
| "loss": 1.3644, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.4021052631578948e-06, | |
| "loss": 1.2786, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 5.8125, | |
| "learning_rate": 1.4005263157894737e-06, | |
| "loss": 1.2067, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 3.90625, | |
| "learning_rate": 1.3989473684210526e-06, | |
| "loss": 1.3127, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 4.0, | |
| "learning_rate": 1.3973684210526317e-06, | |
| "loss": 1.2442, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 5.875, | |
| "learning_rate": 1.3957894736842104e-06, | |
| "loss": 1.2307, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.3942105263157895e-06, | |
| "loss": 1.2828, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 4.28125, | |
| "learning_rate": 1.3926315789473685e-06, | |
| "loss": 1.2676, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 1.3910526315789476e-06, | |
| "loss": 1.1911, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 1.3894736842105263e-06, | |
| "loss": 1.3177, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 1.3878947368421052e-06, | |
| "loss": 1.2891, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 1.3863157894736843e-06, | |
| "loss": 1.303, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 4.125, | |
| "learning_rate": 1.3847368421052632e-06, | |
| "loss": 1.2586, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 6.8125, | |
| "learning_rate": 1.383157894736842e-06, | |
| "loss": 1.3141, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 4.3125, | |
| "learning_rate": 1.381578947368421e-06, | |
| "loss": 1.2174, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.3800000000000001e-06, | |
| "loss": 1.2609, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 5.75, | |
| "learning_rate": 1.378421052631579e-06, | |
| "loss": 1.1323, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.376842105263158e-06, | |
| "loss": 1.2328, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.3752631578947368e-06, | |
| "loss": 1.3087, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 1.3736842105263158e-06, | |
| "loss": 1.2428, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 6.46875, | |
| "learning_rate": 1.3721052631578949e-06, | |
| "loss": 1.235, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 1.3705263157894738e-06, | |
| "loss": 1.2108, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.3689473684210527e-06, | |
| "loss": 1.1704, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.3673684210526316e-06, | |
| "loss": 1.3411, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 1.3657894736842107e-06, | |
| "loss": 1.239, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 1.3642105263157896e-06, | |
| "loss": 1.1661, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.3626315789473683e-06, | |
| "loss": 1.338, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 5.0625, | |
| "learning_rate": 1.3610526315789474e-06, | |
| "loss": 1.2926, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 2.875, | |
| "learning_rate": 1.3594736842105263e-06, | |
| "loss": 1.2866, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 4.96875, | |
| "learning_rate": 1.3578947368421055e-06, | |
| "loss": 1.3333, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 6.53125, | |
| "learning_rate": 1.3563157894736842e-06, | |
| "loss": 1.2472, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 3.390625, | |
| "learning_rate": 1.3547368421052633e-06, | |
| "loss": 1.1796, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 6.125, | |
| "learning_rate": 1.3531578947368422e-06, | |
| "loss": 1.321, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.351578947368421e-06, | |
| "loss": 1.2529, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 1.35e-06, | |
| "loss": 1.2682, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.348421052631579e-06, | |
| "loss": 1.358, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 1.346842105263158e-06, | |
| "loss": 1.3318, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.345263157894737e-06, | |
| "loss": 1.2531, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 1.3436842105263158e-06, | |
| "loss": 1.2796, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 3.203125, | |
| "learning_rate": 1.3421052631578947e-06, | |
| "loss": 1.2315, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 4.71875, | |
| "learning_rate": 1.3405263157894736e-06, | |
| "loss": 1.2501, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 1.3389473684210528e-06, | |
| "loss": 1.3444, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 1.3373684210526317e-06, | |
| "loss": 1.2065, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 4.59375, | |
| "learning_rate": 1.3357894736842106e-06, | |
| "loss": 1.3091, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 4.625, | |
| "learning_rate": 1.3342105263157895e-06, | |
| "loss": 1.1932, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 3.859375, | |
| "learning_rate": 1.3326315789473686e-06, | |
| "loss": 1.1837, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 1.3310526315789475e-06, | |
| "loss": 1.1756, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 1.3294736842105262e-06, | |
| "loss": 1.3163, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 3.921875, | |
| "learning_rate": 1.3278947368421053e-06, | |
| "loss": 1.1937, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 1.3263157894736842e-06, | |
| "loss": 1.3058, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 3.609375, | |
| "learning_rate": 1.3247368421052633e-06, | |
| "loss": 1.2718, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.323157894736842e-06, | |
| "loss": 1.2027, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.3215789473684212e-06, | |
| "loss": 1.2548, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 5.875, | |
| "learning_rate": 1.32e-06, | |
| "loss": 1.3534, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 4.125, | |
| "learning_rate": 1.318421052631579e-06, | |
| "loss": 1.2223, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.3168421052631579e-06, | |
| "loss": 1.2684, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.3152631578947368e-06, | |
| "loss": 1.2611, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 4.71875, | |
| "learning_rate": 1.313684210526316e-06, | |
| "loss": 1.3116, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 5.5, | |
| "learning_rate": 1.3121052631578948e-06, | |
| "loss": 1.303, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.3105263157894737e-06, | |
| "loss": 1.2399, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 4.03125, | |
| "learning_rate": 1.3089473684210526e-06, | |
| "loss": 1.2954, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 5.125, | |
| "learning_rate": 1.3073684210526315e-06, | |
| "loss": 1.2689, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 5.625, | |
| "learning_rate": 1.3057894736842107e-06, | |
| "loss": 1.3605, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 4.125, | |
| "learning_rate": 1.3042105263157896e-06, | |
| "loss": 1.2873, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 5.5, | |
| "learning_rate": 1.3026315789473685e-06, | |
| "loss": 1.2686, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.3010526315789474e-06, | |
| "loss": 1.244, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 4.3125, | |
| "learning_rate": 1.2994736842105265e-06, | |
| "loss": 1.2044, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 3.5625, | |
| "learning_rate": 1.2978947368421054e-06, | |
| "loss": 1.2026, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 1.296315789473684e-06, | |
| "loss": 1.306, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 4.125, | |
| "learning_rate": 1.2947368421052632e-06, | |
| "loss": 1.2968, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 6.75, | |
| "learning_rate": 1.2931578947368421e-06, | |
| "loss": 1.2308, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 5.9375, | |
| "learning_rate": 1.2915789473684212e-06, | |
| "loss": 1.3022, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 6.34375, | |
| "learning_rate": 1.29e-06, | |
| "loss": 1.2298, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 4.8125, | |
| "learning_rate": 1.288421052631579e-06, | |
| "loss": 1.1421, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 5.0, | |
| "learning_rate": 1.286842105263158e-06, | |
| "loss": 1.2018, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 5.84375, | |
| "learning_rate": 1.2852631578947369e-06, | |
| "loss": 1.2329, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 5.375, | |
| "learning_rate": 1.2836842105263158e-06, | |
| "loss": 1.2509, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 1.2821052631578947e-06, | |
| "loss": 1.2556, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 1.2805263157894738e-06, | |
| "loss": 1.3504, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 1.2789473684210527e-06, | |
| "loss": 1.1525, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 1.2773684210526316e-06, | |
| "loss": 1.2227, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 3.65625, | |
| "learning_rate": 1.2757894736842105e-06, | |
| "loss": 1.2962, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 3.765625, | |
| "learning_rate": 1.2742105263157894e-06, | |
| "loss": 1.1717, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 6.65625, | |
| "learning_rate": 1.2726315789473685e-06, | |
| "loss": 1.2543, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 3.984375, | |
| "learning_rate": 1.2710526315789474e-06, | |
| "loss": 1.2092, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 1.2694736842105264e-06, | |
| "loss": 1.1855, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 4.3125, | |
| "learning_rate": 1.2678947368421053e-06, | |
| "loss": 1.2698, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 5.625, | |
| "learning_rate": 1.2663157894736844e-06, | |
| "loss": 1.2615, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 4.59375, | |
| "learning_rate": 1.2647368421052633e-06, | |
| "loss": 1.2156, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 6.875, | |
| "learning_rate": 1.263157894736842e-06, | |
| "loss": 1.2328, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "eval_loss": 1.3270684480667114, | |
| "eval_runtime": 64.5706, | |
| "eval_samples_per_second": 15.487, | |
| "eval_steps_per_second": 15.487, | |
| "step": 2000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 2000, | |
| "total_flos": 3.227844083712e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |