| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005, | |
| "grad_norm": 292.0, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 2.5141, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 284.0, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 2.5031, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.015, | |
| "grad_norm": 264.0, | |
| "learning_rate": 2.9e-06, | |
| "loss": 2.4906, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 464.0, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 2.4297, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "grad_norm": 294.0, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 2.3531, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 198.0, | |
| "learning_rate": 5.9e-06, | |
| "loss": 2.2484, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.035, | |
| "grad_norm": 466.0, | |
| "learning_rate": 6.9e-06, | |
| "loss": 2.0539, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 234.0, | |
| "learning_rate": 7.9e-06, | |
| "loss": 1.9094, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.045, | |
| "grad_norm": 134.0, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 1.6258, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 108.0, | |
| "learning_rate": 9.9e-06, | |
| "loss": 1.4383, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.055, | |
| "grad_norm": 113.5, | |
| "learning_rate": 1.0900000000000002e-05, | |
| "loss": 1.2742, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 87.0, | |
| "learning_rate": 1.1900000000000001e-05, | |
| "loss": 1.1359, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.065, | |
| "grad_norm": 88.5, | |
| "learning_rate": 1.2900000000000002e-05, | |
| "loss": 1.0875, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 149.0, | |
| "learning_rate": 1.39e-05, | |
| "loss": 0.9602, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "grad_norm": 44.25, | |
| "learning_rate": 1.4900000000000001e-05, | |
| "loss": 0.9809, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 65.0, | |
| "learning_rate": 1.5900000000000004e-05, | |
| "loss": 0.8633, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.085, | |
| "grad_norm": 24.0, | |
| "learning_rate": 1.69e-05, | |
| "loss": 0.8012, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 27.125, | |
| "learning_rate": 1.79e-05, | |
| "loss": 0.7484, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.095, | |
| "grad_norm": 14.875, | |
| "learning_rate": 1.8900000000000002e-05, | |
| "loss": 0.682, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 16.625, | |
| "learning_rate": 1.9900000000000003e-05, | |
| "loss": 0.6445, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.105, | |
| "grad_norm": 11.4375, | |
| "learning_rate": 1.9998766324816606e-05, | |
| "loss": 0.5703, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 11.875, | |
| "learning_rate": 1.9994502159417576e-05, | |
| "loss": 0.5605, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.115, | |
| "grad_norm": 12.9375, | |
| "learning_rate": 1.9987193571841865e-05, | |
| "loss": 0.5115, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 14.9375, | |
| "learning_rate": 1.9976842788356054e-05, | |
| "loss": 0.5072, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 11.875, | |
| "learning_rate": 1.9963452961909065e-05, | |
| "loss": 0.5357, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 12.6875, | |
| "learning_rate": 1.9947028171171742e-05, | |
| "loss": 0.4924, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.135, | |
| "grad_norm": 12.0, | |
| "learning_rate": 1.9927573419294456e-05, | |
| "loss": 0.4867, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 8.0, | |
| "learning_rate": 1.990509463238309e-05, | |
| "loss": 0.4533, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.145, | |
| "grad_norm": 6.5625, | |
| "learning_rate": 1.9879598657693894e-05, | |
| "loss": 0.4295, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 11.1875, | |
| "learning_rate": 1.985109326154774e-05, | |
| "loss": 0.4166, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.155, | |
| "grad_norm": 8.5, | |
| "learning_rate": 1.981958712696444e-05, | |
| "loss": 0.4002, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 11.0, | |
| "learning_rate": 1.9785089851017788e-05, | |
| "loss": 0.4008, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.165, | |
| "grad_norm": 6.71875, | |
| "learning_rate": 1.974761194191222e-05, | |
| "loss": 0.3975, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 8.125, | |
| "learning_rate": 1.970716481578191e-05, | |
| "loss": 0.4086, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "grad_norm": 11.875, | |
| "learning_rate": 1.9663760793213297e-05, | |
| "loss": 0.3832, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 6.78125, | |
| "learning_rate": 1.9617413095492114e-05, | |
| "loss": 0.4455, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.185, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 1.956813584057608e-05, | |
| "loss": 0.3797, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 10.5625, | |
| "learning_rate": 1.9515944038794384e-05, | |
| "loss": 0.3982, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.195, | |
| "grad_norm": 8.75, | |
| "learning_rate": 1.9460853588275454e-05, | |
| "loss": 0.4045, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 7.09375, | |
| "learning_rate": 1.940288127010419e-05, | |
| "loss": 0.3973, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.205, | |
| "grad_norm": 5.71875, | |
| "learning_rate": 1.9342044743210295e-05, | |
| "loss": 0.3715, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 6.8125, | |
| "learning_rate": 1.92783625389892e-05, | |
| "loss": 0.3781, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.215, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 1.9211854055657216e-05, | |
| "loss": 0.3662, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 1.9142539552342638e-05, | |
| "loss": 0.3656, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "grad_norm": 6.8125, | |
| "learning_rate": 1.907044014291465e-05, | |
| "loss": 0.3611, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 5.59375, | |
| "learning_rate": 1.8995577789551806e-05, | |
| "loss": 0.3867, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.235, | |
| "grad_norm": 16.25, | |
| "learning_rate": 1.8917975296052143e-05, | |
| "loss": 0.3602, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 6.71875, | |
| "learning_rate": 1.8837656300886937e-05, | |
| "loss": 0.3686, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.245, | |
| "grad_norm": 5.75, | |
| "learning_rate": 1.875464527000018e-05, | |
| "loss": 0.3604, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 4.15625, | |
| "learning_rate": 1.866896748935603e-05, | |
| "loss": 0.3652, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.255, | |
| "grad_norm": 10.4375, | |
| "learning_rate": 1.858064905723645e-05, | |
| "loss": 0.357, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.8489716876291417e-05, | |
| "loss": 0.3617, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.265, | |
| "grad_norm": 5.28125, | |
| "learning_rate": 1.8396198645344133e-05, | |
| "loss": 0.3447, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 7.59375, | |
| "learning_rate": 1.8300122850953678e-05, | |
| "loss": 0.3469, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "grad_norm": 7.03125, | |
| "learning_rate": 1.8201518758737726e-05, | |
| "loss": 0.3668, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 3.8125, | |
| "learning_rate": 1.8100416404457962e-05, | |
| "loss": 0.3518, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.285, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 1.799684658487091e-05, | |
| "loss": 0.3416, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 7.4375, | |
| "learning_rate": 1.789084084834691e-05, | |
| "loss": 0.3391, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.295, | |
| "grad_norm": 3.78125, | |
| "learning_rate": 1.778243148526021e-05, | |
| "loss": 0.3471, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 6.03125, | |
| "learning_rate": 1.7671651518153e-05, | |
| "loss": 0.358, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.305, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 1.7558534691676396e-05, | |
| "loss": 0.332, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 7.0, | |
| "learning_rate": 1.744311546231154e-05, | |
| "loss": 0.3463, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.315, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.732542898787379e-05, | |
| "loss": 0.3408, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 6.5625, | |
| "learning_rate": 1.7205511116803306e-05, | |
| "loss": 0.3174, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 1.708339837724529e-05, | |
| "loss": 0.3496, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 3.890625, | |
| "learning_rate": 1.6959127965923144e-05, | |
| "loss": 0.3287, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.335, | |
| "grad_norm": 8.875, | |
| "learning_rate": 1.6832737736807994e-05, | |
| "loss": 0.3297, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 8.0625, | |
| "learning_rate": 1.6704266189587992e-05, | |
| "loss": 0.3408, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.345, | |
| "grad_norm": 6.53125, | |
| "learning_rate": 1.657375245794096e-05, | |
| "loss": 0.3262, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 5.0, | |
| "learning_rate": 1.644123629761387e-05, | |
| "loss": 0.3377, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.355, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.6306758074312866e-05, | |
| "loss": 0.3328, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 7.6875, | |
| "learning_rate": 1.617035875140749e-05, | |
| "loss": 0.3387, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.365, | |
| "grad_norm": 7.03125, | |
| "learning_rate": 1.6032079877452825e-05, | |
| "loss": 0.3395, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 1.5891963573533424e-05, | |
| "loss": 0.3254, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 3.671875, | |
| "learning_rate": 1.575005252043279e-05, | |
| "loss": 0.3414, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 5.375, | |
| "learning_rate": 1.560638994563242e-05, | |
| "loss": 0.3307, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.385, | |
| "grad_norm": 7.96875, | |
| "learning_rate": 1.5461019610144292e-05, | |
| "loss": 0.3297, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 1.531398579518083e-05, | |
| "loss": 0.3285, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.395, | |
| "grad_norm": 22.75, | |
| "learning_rate": 1.516533328866642e-05, | |
| "loss": 0.3287, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 5.28125, | |
| "learning_rate": 1.5015107371594576e-05, | |
| "loss": 0.323, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.405, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 1.4863353804234906e-05, | |
| "loss": 0.3195, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 4.71875, | |
| "learning_rate": 1.47101188121941e-05, | |
| "loss": 0.3342, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.415, | |
| "grad_norm": 12.9375, | |
| "learning_rate": 1.4555449072335157e-05, | |
| "loss": 0.3166, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 1.4399391698559153e-05, | |
| "loss": 0.3207, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.425, | |
| "grad_norm": 6.40625, | |
| "learning_rate": 1.4241994227453902e-05, | |
| "loss": 0.3168, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 6.78125, | |
| "learning_rate": 1.408330460381385e-05, | |
| "loss": 0.3215, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.435, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 1.3923371166035615e-05, | |
| "loss": 0.3273, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 8.0, | |
| "learning_rate": 1.3762242631393656e-05, | |
| "loss": 0.3217, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.445, | |
| "grad_norm": 6.5625, | |
| "learning_rate": 1.3599968081200515e-05, | |
| "loss": 0.3086, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 1.3436596945856164e-05, | |
| "loss": 0.3053, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.455, | |
| "grad_norm": 5.5, | |
| "learning_rate": 1.327217898979104e-05, | |
| "loss": 0.334, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 6.90625, | |
| "learning_rate": 1.310676429630732e-05, | |
| "loss": 0.3307, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.465, | |
| "grad_norm": 9.625, | |
| "learning_rate": 1.294040325232304e-05, | |
| "loss": 0.3172, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 3.265625, | |
| "learning_rate": 1.2773146533023782e-05, | |
| "loss": 0.3057, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.475, | |
| "grad_norm": 6.625, | |
| "learning_rate": 1.2605045086426487e-05, | |
| "loss": 0.31, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 8.9375, | |
| "learning_rate": 1.2436150117860226e-05, | |
| "loss": 0.3145, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.485, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 1.2266513074368552e-05, | |
| "loss": 0.3033, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 1.2096185629038219e-05, | |
| "loss": 0.3057, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.495, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.1925219665259076e-05, | |
| "loss": 0.3045, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 6.65625, | |
| "learning_rate": 1.1753667260919872e-05, | |
| "loss": 0.2984, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.505, | |
| "grad_norm": 4.09375, | |
| "learning_rate": 1.1581580672544839e-05, | |
| "loss": 0.3219, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 1.1409012319375828e-05, | |
| "loss": 0.2994, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.515, | |
| "grad_norm": 4.5625, | |
| "learning_rate": 1.1236014767404929e-05, | |
| "loss": 0.2963, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 6.125, | |
| "learning_rate": 1.1062640713362333e-05, | |
| "loss": 0.2944, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.525, | |
| "grad_norm": 3.71875, | |
| "learning_rate": 1.0888942968664417e-05, | |
| "loss": 0.3053, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 1.071497444332686e-05, | |
| "loss": 0.2799, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.535, | |
| "grad_norm": 5.125, | |
| "learning_rate": 1.0540788129847757e-05, | |
| "loss": 0.3006, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 1.0366437087065564e-05, | |
| "loss": 0.2879, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.545, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 1.01919744239969e-05, | |
| "loss": 0.301, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 5.8125, | |
| "learning_rate": 1.0017453283658984e-05, | |
| "loss": 0.2893, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.555, | |
| "grad_norm": 6.5, | |
| "learning_rate": 9.842926826881796e-06, | |
| "loss": 0.2927, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 9.668448216114739e-06, | |
| "loss": 0.3023, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.565, | |
| "grad_norm": 3.65625, | |
| "learning_rate": 9.494070599232868e-06, | |
| "loss": 0.2893, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 5.6875, | |
| "learning_rate": 9.319847093347522e-06, | |
| "loss": 0.2852, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.575, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 9.145830768626326e-06, | |
| "loss": 0.2867, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 8.972074632127533e-06, | |
| "loss": 0.2893, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.585, | |
| "grad_norm": 7.71875, | |
| "learning_rate": 8.79863161165353e-06, | |
| "loss": 0.2994, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 8.625554539628536e-06, | |
| "loss": 0.3064, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.595, | |
| "grad_norm": 4.3125, | |
| "learning_rate": 8.452896137005322e-06, | |
| "loss": 0.2879, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 8.280708997205904e-06, | |
| "loss": 0.2965, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.605, | |
| "grad_norm": 5.625, | |
| "learning_rate": 8.109045570101086e-06, | |
| "loss": 0.2859, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 4.6875, | |
| "learning_rate": 7.937958146033706e-06, | |
| "loss": 0.2947, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.615, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 7.767498839890489e-06, | |
| "loss": 0.2764, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 3.84375, | |
| "learning_rate": 7.597719575227364e-06, | |
| "loss": 0.2838, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 4.40625, | |
| "learning_rate": 7.428672068453041e-06, | |
| "loss": 0.2879, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 7.260407813075676e-06, | |
| "loss": 0.2854, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.635, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 7.092978064017475e-06, | |
| "loss": 0.2908, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 3.859375, | |
| "learning_rate": 6.92643382200193e-06, | |
| "loss": 0.2771, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.645, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 6.7608258180185085e-06, | |
| "loss": 0.271, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 6.596204497869501e-06, | |
| "loss": 0.2959, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.655, | |
| "grad_norm": 4.53125, | |
| "learning_rate": 6.432620006803747e-06, | |
| "loss": 0.2788, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 6.2701221742419106e-06, | |
| "loss": 0.2773, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.665, | |
| "grad_norm": 5.375, | |
| "learning_rate": 6.108760498597939e-06, | |
| "loss": 0.2783, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 5.948584132201376e-06, | |
| "loss": 0.2724, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.675, | |
| "grad_norm": 5.15625, | |
| "learning_rate": 5.789641866325091e-06, | |
| "loss": 0.2826, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 5.631982116322981e-06, | |
| "loss": 0.2874, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.685, | |
| "grad_norm": 4.21875, | |
| "learning_rate": 5.475652906882173e-06, | |
| "loss": 0.274, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 5.625, | |
| "learning_rate": 5.3207018573942684e-06, | |
| "loss": 0.2892, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.695, | |
| "grad_norm": 4.0, | |
| "learning_rate": 5.167176167449977e-06, | |
| "loss": 0.2796, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 3.359375, | |
| "learning_rate": 5.015122602461698e-06, | |
| "loss": 0.2771, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.705, | |
| "grad_norm": 4.59375, | |
| "learning_rate": 4.864587479418302e-06, | |
| "loss": 0.2857, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 4.71561665277653e-06, | |
| "loss": 0.2787, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.715, | |
| "grad_norm": 6.28125, | |
| "learning_rate": 4.568255500493292e-06, | |
| "loss": 0.2822, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 3.796875, | |
| "learning_rate": 4.422548910203099e-06, | |
| "loss": 0.2934, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.725, | |
| "grad_norm": 4.09375, | |
| "learning_rate": 4.27854126554484e-06, | |
| "loss": 0.277, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 4.136276432642107e-06, | |
| "loss": 0.2833, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.735, | |
| "grad_norm": 4.625, | |
| "learning_rate": 3.9957977467411615e-06, | |
| "loss": 0.2797, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 3.765625, | |
| "learning_rate": 3.857147999010568e-06, | |
| "loss": 0.2705, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.745, | |
| "grad_norm": 7.34375, | |
| "learning_rate": 3.7203694235066224e-06, | |
| "loss": 0.2762, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 4.4375, | |
| "learning_rate": 3.5855036843084213e-06, | |
| "loss": 0.2863, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.755, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 3.452591862826603e-06, | |
| "loss": 0.2795, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 4.90625, | |
| "learning_rate": 3.3216744452895356e-06, | |
| "loss": 0.2887, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.765, | |
| "grad_norm": 3.34375, | |
| "learning_rate": 3.192791310410822e-06, | |
| "loss": 0.2762, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 3.0659817172418694e-06, | |
| "loss": 0.2851, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.775, | |
| "grad_norm": 5.09375, | |
| "learning_rate": 2.9412842932131904e-06, | |
| "loss": 0.2666, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 4.75, | |
| "learning_rate": 2.8187370223681134e-06, | |
| "loss": 0.2855, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.785, | |
| "grad_norm": 5.4375, | |
| "learning_rate": 2.698377233792476e-06, | |
| "loss": 0.2816, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 4.65625, | |
| "learning_rate": 2.5802415902438373e-06, | |
| "loss": 0.285, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.795, | |
| "grad_norm": 4.5, | |
| "learning_rate": 2.464366076983623e-06, | |
| "loss": 0.2855, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 4.0625, | |
| "learning_rate": 2.3507859908156828e-06, | |
| "loss": 0.2834, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.805, | |
| "grad_norm": 3.203125, | |
| "learning_rate": 2.2395359293345396e-06, | |
| "loss": 0.2703, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 3.453125, | |
| "learning_rate": 2.130649780386628e-06, | |
| "loss": 0.2764, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.815, | |
| "grad_norm": 4.34375, | |
| "learning_rate": 2.024160711747717e-06, | |
| "loss": 0.2791, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.9201011610196972e-06, | |
| "loss": 0.261, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.825, | |
| "grad_norm": 3.921875, | |
| "learning_rate": 1.818502825749764e-06, | |
| "loss": 0.2709, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 4.53125, | |
| "learning_rate": 1.7193966537750561e-06, | |
| "loss": 0.2628, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.835, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 1.6228128337956128e-06, | |
| "loss": 0.2697, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 1.5287807861786308e-06, | |
| "loss": 0.2707, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.845, | |
| "grad_norm": 6.9375, | |
| "learning_rate": 1.4373291539967182e-06, | |
| "loss": 0.2854, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 4.875, | |
| "learning_rate": 1.3484857943029572e-06, | |
| "loss": 0.293, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.855, | |
| "grad_norm": 3.859375, | |
| "learning_rate": 1.2622777696453482e-06, | |
| "loss": 0.2786, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 3.734375, | |
| "learning_rate": 1.1787313398233235e-06, | |
| "loss": 0.2725, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.865, | |
| "grad_norm": 3.734375, | |
| "learning_rate": 1.097871953888735e-06, | |
| "loss": 0.2753, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 8.375, | |
| "learning_rate": 1.0197242423938447e-06, | |
| "loss": 0.2815, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 3.515625, | |
| "learning_rate": 9.44312009888606e-07, | |
| "loss": 0.2753, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 4.46875, | |
| "learning_rate": 8.716582276695729e-07, | |
| "loss": 0.2733, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.885, | |
| "grad_norm": 2.265625, | |
| "learning_rate": 8.017850267826233e-07, | |
| "loss": 0.2724, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 3.828125, | |
| "learning_rate": 7.347136912816277e-07, | |
| "loss": 0.2797, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.895, | |
| "grad_norm": 5.75, | |
| "learning_rate": 6.704646517451108e-07, | |
| "loss": 0.2827, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 4.125, | |
| "learning_rate": 6.090574790529091e-07, | |
| "loss": 0.2844, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.905, | |
| "grad_norm": 4.0625, | |
| "learning_rate": 5.505108784246926e-07, | |
| "loss": 0.2653, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 2.796875, | |
| "learning_rate": 4.948426837221632e-07, | |
| "loss": 0.2786, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.915, | |
| "grad_norm": 5.375, | |
| "learning_rate": 4.420698520166988e-07, | |
| "loss": 0.2855, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 4.5, | |
| "learning_rate": 3.922084584240582e-07, | |
| "loss": 0.2798, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.925, | |
| "grad_norm": 5.46875, | |
| "learning_rate": 3.4527369120775036e-07, | |
| "loss": 0.2588, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 3.84375, | |
| "learning_rate": 3.0127984715253246e-07, | |
| "loss": 0.2796, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.935, | |
| "grad_norm": 3.953125, | |
| "learning_rate": 2.6024032720948446e-07, | |
| "loss": 0.2724, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 4.0625, | |
| "learning_rate": 2.221676324139377e-07, | |
| "loss": 0.2658, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.945, | |
| "grad_norm": 3.390625, | |
| "learning_rate": 1.8707336007754873e-07, | |
| "loss": 0.273, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 4.09375, | |
| "learning_rate": 1.549682002556341e-07, | |
| "loss": 0.2789, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.955, | |
| "grad_norm": 3.5625, | |
| "learning_rate": 1.2586193249088607e-07, | |
| "loss": 0.2721, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 3.109375, | |
| "learning_rate": 9.976342283442464e-08, | |
| "loss": 0.2777, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.965, | |
| "grad_norm": 4.15625, | |
| "learning_rate": 7.66806211451132e-08, | |
| "loss": 0.2751, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 5.662055866795357e-08, | |
| "loss": 0.2714, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.975, | |
| "grad_norm": 8.25, | |
| "learning_rate": 3.9589345892304673e-08, | |
| "loss": 0.2776, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 4.53125, | |
| "learning_rate": 2.5592170690560415e-08, | |
| "loss": 0.2812, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.985, | |
| "grad_norm": 5.1875, | |
| "learning_rate": 1.4633296737882607e-08, | |
| "loss": 0.2877, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 3.34375, | |
| "learning_rate": 6.716062213437679e-09, | |
| "loss": 0.2805, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.995, | |
| "grad_norm": 3.5625, | |
| "learning_rate": 1.8428787835578222e-09, | |
| "loss": 0.2799, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.9375, | |
| "learning_rate": 1.5230867123072757e-11, | |
| "loss": 0.2747, | |
| "step": 2000 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 2000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 2000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.2073827663872e+17, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |