| { | |
| "best_metric": 0.8666666666666667, | |
| "best_model_checkpoint": "CTMAE2_CS_V7_5/checkpoint-1755", | |
| "epoch": 49.01494845360825, | |
| "eval_steps": 500, | |
| "global_step": 9700, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010309278350515464, | |
| "grad_norm": 5.501185894012451, | |
| "learning_rate": 1.0309278350515465e-07, | |
| "loss": 0.6908, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.002061855670103093, | |
| "grad_norm": 2.460876941680908, | |
| "learning_rate": 2.061855670103093e-07, | |
| "loss": 0.6912, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.003092783505154639, | |
| "grad_norm": 2.6134722232818604, | |
| "learning_rate": 3.0927835051546394e-07, | |
| "loss": 0.6906, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.004123711340206186, | |
| "grad_norm": 3.098764181137085, | |
| "learning_rate": 4.123711340206186e-07, | |
| "loss": 0.6779, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.005154639175257732, | |
| "grad_norm": 2.6916747093200684, | |
| "learning_rate": 5.154639175257732e-07, | |
| "loss": 0.688, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.006185567010309278, | |
| "grad_norm": 2.6864609718322754, | |
| "learning_rate": 6.185567010309279e-07, | |
| "loss": 0.6744, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.007216494845360825, | |
| "grad_norm": 1.6162508726119995, | |
| "learning_rate": 7.216494845360824e-07, | |
| "loss": 0.6806, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.008247422680412371, | |
| "grad_norm": 6.75019645690918, | |
| "learning_rate": 8.247422680412372e-07, | |
| "loss": 0.6594, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.009278350515463918, | |
| "grad_norm": 3.412571668624878, | |
| "learning_rate": 9.278350515463919e-07, | |
| "loss": 0.6602, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.010309278350515464, | |
| "grad_norm": 3.0713565349578857, | |
| "learning_rate": 1.0309278350515464e-06, | |
| "loss": 0.6957, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01134020618556701, | |
| "grad_norm": 2.902609348297119, | |
| "learning_rate": 1.134020618556701e-06, | |
| "loss": 0.6819, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.012371134020618556, | |
| "grad_norm": 13.36193561553955, | |
| "learning_rate": 1.2371134020618557e-06, | |
| "loss": 0.6161, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.013402061855670102, | |
| "grad_norm": 5.576000213623047, | |
| "learning_rate": 1.3402061855670104e-06, | |
| "loss": 0.6261, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.01443298969072165, | |
| "grad_norm": 6.617053031921387, | |
| "learning_rate": 1.4432989690721649e-06, | |
| "loss": 0.529, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.015463917525773196, | |
| "grad_norm": 12.840880393981934, | |
| "learning_rate": 1.5463917525773197e-06, | |
| "loss": 0.7079, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.016494845360824743, | |
| "grad_norm": 10.174439430236816, | |
| "learning_rate": 1.6494845360824744e-06, | |
| "loss": 0.6972, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01752577319587629, | |
| "grad_norm": 19.259891510009766, | |
| "learning_rate": 1.7525773195876288e-06, | |
| "loss": 0.54, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.018556701030927835, | |
| "grad_norm": 6.065194129943848, | |
| "learning_rate": 1.8556701030927837e-06, | |
| "loss": 0.646, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.01958762886597938, | |
| "grad_norm": 9.375090599060059, | |
| "learning_rate": 1.9587628865979384e-06, | |
| "loss": 0.6793, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.020103092783505156, | |
| "eval_accuracy": 0.4666666666666667, | |
| "eval_loss": 0.7615872025489807, | |
| "eval_runtime": 15.3942, | |
| "eval_samples_per_second": 2.923, | |
| "eval_steps_per_second": 0.78, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.0005154639175258, | |
| "grad_norm": 13.356775283813477, | |
| "learning_rate": 2.061855670103093e-06, | |
| "loss": 0.5875, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.0015463917525773, | |
| "grad_norm": 4.08266544342041, | |
| "learning_rate": 2.1649484536082477e-06, | |
| "loss": 0.6459, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.0025773195876289, | |
| "grad_norm": 6.374290466308594, | |
| "learning_rate": 2.268041237113402e-06, | |
| "loss": 0.6691, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.0036082474226804, | |
| "grad_norm": 5.553038597106934, | |
| "learning_rate": 2.3711340206185566e-06, | |
| "loss": 0.6392, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.004639175257732, | |
| "grad_norm": 19.62584686279297, | |
| "learning_rate": 2.4742268041237115e-06, | |
| "loss": 0.634, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.0056701030927835, | |
| "grad_norm": 8.070799827575684, | |
| "learning_rate": 2.577319587628866e-06, | |
| "loss": 0.6665, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.006701030927835, | |
| "grad_norm": 8.732718467712402, | |
| "learning_rate": 2.680412371134021e-06, | |
| "loss": 0.6228, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.0077319587628866, | |
| "grad_norm": 4.9655890464782715, | |
| "learning_rate": 2.7835051546391757e-06, | |
| "loss": 0.6189, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.0087628865979381, | |
| "grad_norm": 9.290786743164062, | |
| "learning_rate": 2.8865979381443297e-06, | |
| "loss": 0.7474, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.0097938144329897, | |
| "grad_norm": 7.63596248626709, | |
| "learning_rate": 2.9896907216494846e-06, | |
| "loss": 0.5911, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.0108247422680412, | |
| "grad_norm": 10.840821266174316, | |
| "learning_rate": 3.0927835051546395e-06, | |
| "loss": 0.6093, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.0118556701030927, | |
| "grad_norm": 4.624059200286865, | |
| "learning_rate": 3.195876288659794e-06, | |
| "loss": 0.7803, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.0128865979381443, | |
| "grad_norm": 5.822396755218506, | |
| "learning_rate": 3.298969072164949e-06, | |
| "loss": 0.598, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.0139175257731958, | |
| "grad_norm": 3.33036732673645, | |
| "learning_rate": 3.4020618556701037e-06, | |
| "loss": 0.5865, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.0149484536082474, | |
| "grad_norm": 5.949455261230469, | |
| "learning_rate": 3.5051546391752577e-06, | |
| "loss": 0.6764, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.015979381443299, | |
| "grad_norm": 6.306659698486328, | |
| "learning_rate": 3.6082474226804126e-06, | |
| "loss": 0.645, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.0170103092783505, | |
| "grad_norm": 17.39265251159668, | |
| "learning_rate": 3.7113402061855674e-06, | |
| "loss": 0.64, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.018041237113402, | |
| "grad_norm": 7.141329288482666, | |
| "learning_rate": 3.814432989690722e-06, | |
| "loss": 0.5337, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.0190721649484535, | |
| "grad_norm": 6.430778980255127, | |
| "learning_rate": 3.917525773195877e-06, | |
| "loss": 0.5442, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.020103092783505, | |
| "grad_norm": 15.002116203308105, | |
| "learning_rate": 4.020618556701032e-06, | |
| "loss": 0.5591, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.020103092783505, | |
| "eval_accuracy": 0.4666666666666667, | |
| "eval_loss": 0.7040268182754517, | |
| "eval_runtime": 12.9016, | |
| "eval_samples_per_second": 3.488, | |
| "eval_steps_per_second": 0.93, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.0010309278350515, | |
| "grad_norm": 9.877669334411621, | |
| "learning_rate": 4.123711340206186e-06, | |
| "loss": 0.5959, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.002061855670103, | |
| "grad_norm": 7.665306568145752, | |
| "learning_rate": 4.2268041237113405e-06, | |
| "loss": 0.6647, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.0030927835051546, | |
| "grad_norm": 19.635908126831055, | |
| "learning_rate": 4.329896907216495e-06, | |
| "loss": 0.5142, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.004123711340206, | |
| "grad_norm": 13.251553535461426, | |
| "learning_rate": 4.4329896907216494e-06, | |
| "loss": 0.7387, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.0051546391752577, | |
| "grad_norm": 33.24567794799805, | |
| "learning_rate": 4.536082474226804e-06, | |
| "loss": 0.5326, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.0061855670103093, | |
| "grad_norm": 17.012880325317383, | |
| "learning_rate": 4.639175257731959e-06, | |
| "loss": 0.6149, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.007216494845361, | |
| "grad_norm": 12.129619598388672, | |
| "learning_rate": 4.742268041237113e-06, | |
| "loss": 0.4566, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.0082474226804123, | |
| "grad_norm": 32.084442138671875, | |
| "learning_rate": 4.845360824742268e-06, | |
| "loss": 0.8027, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.009278350515464, | |
| "grad_norm": 6.839757442474365, | |
| "learning_rate": 4.948453608247423e-06, | |
| "loss": 0.5719, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.0103092783505154, | |
| "grad_norm": 18.979419708251953, | |
| "learning_rate": 5.051546391752578e-06, | |
| "loss": 0.5684, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.011340206185567, | |
| "grad_norm": 11.15890121459961, | |
| "learning_rate": 5.154639175257732e-06, | |
| "loss": 0.4082, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0123711340206185, | |
| "grad_norm": 10.10776138305664, | |
| "learning_rate": 5.257731958762888e-06, | |
| "loss": 0.5722, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.01340206185567, | |
| "grad_norm": 14.203100204467773, | |
| "learning_rate": 5.360824742268042e-06, | |
| "loss": 0.4529, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.0144329896907216, | |
| "grad_norm": 13.093475341796875, | |
| "learning_rate": 5.463917525773196e-06, | |
| "loss": 0.5994, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.015463917525773, | |
| "grad_norm": 23.01972007751465, | |
| "learning_rate": 5.567010309278351e-06, | |
| "loss": 0.5696, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.0164948453608247, | |
| "grad_norm": 24.430442810058594, | |
| "learning_rate": 5.670103092783505e-06, | |
| "loss": 0.5652, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.0175257731958762, | |
| "grad_norm": 15.79038143157959, | |
| "learning_rate": 5.7731958762886594e-06, | |
| "loss": 0.5913, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.0185567010309278, | |
| "grad_norm": 21.04030990600586, | |
| "learning_rate": 5.876288659793815e-06, | |
| "loss": 0.484, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.0195876288659793, | |
| "grad_norm": 8.871047973632812, | |
| "learning_rate": 5.979381443298969e-06, | |
| "loss": 0.7211, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.020103092783505, | |
| "eval_accuracy": 0.8222222222222222, | |
| "eval_loss": 0.4916439354419708, | |
| "eval_runtime": 12.7637, | |
| "eval_samples_per_second": 3.526, | |
| "eval_steps_per_second": 0.94, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 3.0005154639175258, | |
| "grad_norm": 9.930680274963379, | |
| "learning_rate": 6.082474226804124e-06, | |
| "loss": 0.4872, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 3.0015463917525773, | |
| "grad_norm": 8.775280952453613, | |
| "learning_rate": 6.185567010309279e-06, | |
| "loss": 0.4803, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.002577319587629, | |
| "grad_norm": 16.243377685546875, | |
| "learning_rate": 6.288659793814433e-06, | |
| "loss": 0.4948, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 3.0036082474226804, | |
| "grad_norm": 43.30958557128906, | |
| "learning_rate": 6.391752577319588e-06, | |
| "loss": 0.6622, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 3.004639175257732, | |
| "grad_norm": 22.228527069091797, | |
| "learning_rate": 6.494845360824743e-06, | |
| "loss": 0.4748, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 3.0056701030927835, | |
| "grad_norm": 17.75151252746582, | |
| "learning_rate": 6.597938144329898e-06, | |
| "loss": 0.4999, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 3.006701030927835, | |
| "grad_norm": 11.695698738098145, | |
| "learning_rate": 6.701030927835052e-06, | |
| "loss": 0.259, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.0077319587628866, | |
| "grad_norm": 38.64352035522461, | |
| "learning_rate": 6.804123711340207e-06, | |
| "loss": 1.0941, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 3.008762886597938, | |
| "grad_norm": 13.178338050842285, | |
| "learning_rate": 6.907216494845361e-06, | |
| "loss": 0.9189, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 3.0097938144329897, | |
| "grad_norm": 18.1324462890625, | |
| "learning_rate": 7.010309278350515e-06, | |
| "loss": 0.6344, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.010824742268041, | |
| "grad_norm": 11.067273139953613, | |
| "learning_rate": 7.113402061855671e-06, | |
| "loss": 0.4916, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 3.0118556701030927, | |
| "grad_norm": 7.603753566741943, | |
| "learning_rate": 7.216494845360825e-06, | |
| "loss": 0.5001, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.0128865979381443, | |
| "grad_norm": 4.259734630584717, | |
| "learning_rate": 7.319587628865979e-06, | |
| "loss": 0.4538, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 3.013917525773196, | |
| "grad_norm": 19.640830993652344, | |
| "learning_rate": 7.422680412371135e-06, | |
| "loss": 0.5484, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 3.0149484536082474, | |
| "grad_norm": 20.696624755859375, | |
| "learning_rate": 7.525773195876289e-06, | |
| "loss": 0.5014, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 3.015979381443299, | |
| "grad_norm": 59.72232437133789, | |
| "learning_rate": 7.628865979381444e-06, | |
| "loss": 0.5271, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.0170103092783505, | |
| "grad_norm": 11.003219604492188, | |
| "learning_rate": 7.731958762886599e-06, | |
| "loss": 0.5286, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.018041237113402, | |
| "grad_norm": 13.869205474853516, | |
| "learning_rate": 7.835051546391754e-06, | |
| "loss": 0.4798, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.0190721649484535, | |
| "grad_norm": 12.36577320098877, | |
| "learning_rate": 7.938144329896907e-06, | |
| "loss": 0.6944, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 3.020103092783505, | |
| "grad_norm": 30.60397720336914, | |
| "learning_rate": 8.041237113402063e-06, | |
| "loss": 0.5544, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.020103092783505, | |
| "eval_accuracy": 0.5555555555555556, | |
| "eval_loss": 0.7589893937110901, | |
| "eval_runtime": 12.6816, | |
| "eval_samples_per_second": 3.548, | |
| "eval_steps_per_second": 0.946, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 4.0010309278350515, | |
| "grad_norm": 20.62627601623535, | |
| "learning_rate": 8.144329896907216e-06, | |
| "loss": 0.5034, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 4.002061855670103, | |
| "grad_norm": 4.391873836517334, | |
| "learning_rate": 8.247422680412371e-06, | |
| "loss": 0.7131, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.003092783505155, | |
| "grad_norm": 14.114368438720703, | |
| "learning_rate": 8.350515463917526e-06, | |
| "loss": 0.5573, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 4.004123711340206, | |
| "grad_norm": 3.9918265342712402, | |
| "learning_rate": 8.453608247422681e-06, | |
| "loss": 0.5594, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 4.005154639175258, | |
| "grad_norm": 29.062408447265625, | |
| "learning_rate": 8.556701030927836e-06, | |
| "loss": 0.5178, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 4.006185567010309, | |
| "grad_norm": 13.267045021057129, | |
| "learning_rate": 8.65979381443299e-06, | |
| "loss": 0.6529, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 4.007216494845361, | |
| "grad_norm": 31.6133975982666, | |
| "learning_rate": 8.762886597938146e-06, | |
| "loss": 0.4495, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 4.008247422680412, | |
| "grad_norm": 7.6835150718688965, | |
| "learning_rate": 8.865979381443299e-06, | |
| "loss": 0.5844, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 4.009278350515464, | |
| "grad_norm": 3.838587522506714, | |
| "learning_rate": 8.969072164948455e-06, | |
| "loss": 0.6229, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 4.010309278350515, | |
| "grad_norm": 12.973073959350586, | |
| "learning_rate": 9.072164948453609e-06, | |
| "loss": 0.5575, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 4.011340206185567, | |
| "grad_norm": 9.979464530944824, | |
| "learning_rate": 9.175257731958764e-06, | |
| "loss": 0.7956, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 4.0123711340206185, | |
| "grad_norm": 10.023995399475098, | |
| "learning_rate": 9.278350515463918e-06, | |
| "loss": 0.5851, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 4.01340206185567, | |
| "grad_norm": 25.86085319519043, | |
| "learning_rate": 9.381443298969073e-06, | |
| "loss": 0.4864, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 4.014432989690722, | |
| "grad_norm": 5.599601745605469, | |
| "learning_rate": 9.484536082474226e-06, | |
| "loss": 0.5482, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 4.015463917525773, | |
| "grad_norm": 1.694999098777771, | |
| "learning_rate": 9.587628865979383e-06, | |
| "loss": 0.4848, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 4.016494845360825, | |
| "grad_norm": 49.36362075805664, | |
| "learning_rate": 9.690721649484536e-06, | |
| "loss": 0.3432, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 4.017525773195876, | |
| "grad_norm": 62.547489166259766, | |
| "learning_rate": 9.793814432989691e-06, | |
| "loss": 0.9801, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.018556701030928, | |
| "grad_norm": 6.7530517578125, | |
| "learning_rate": 9.896907216494846e-06, | |
| "loss": 0.7231, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 4.019587628865979, | |
| "grad_norm": 18.402511596679688, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6032, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 4.020103092783505, | |
| "eval_accuracy": 0.6666666666666666, | |
| "eval_loss": 0.5508340001106262, | |
| "eval_runtime": 13.2243, | |
| "eval_samples_per_second": 3.403, | |
| "eval_steps_per_second": 0.907, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 5.000515463917526, | |
| "grad_norm": 9.008790969848633, | |
| "learning_rate": 9.988545246277205e-06, | |
| "loss": 0.5444, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 5.001546391752577, | |
| "grad_norm": 24.455364227294922, | |
| "learning_rate": 9.977090492554411e-06, | |
| "loss": 0.6265, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 5.002577319587629, | |
| "grad_norm": 2.534010410308838, | |
| "learning_rate": 9.965635738831616e-06, | |
| "loss": 0.4103, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.00360824742268, | |
| "grad_norm": 2.197962999343872, | |
| "learning_rate": 9.95418098510882e-06, | |
| "loss": 0.6683, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 5.004639175257732, | |
| "grad_norm": 21.681041717529297, | |
| "learning_rate": 9.942726231386026e-06, | |
| "loss": 0.5277, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 5.0056701030927835, | |
| "grad_norm": 6.0828752517700195, | |
| "learning_rate": 9.931271477663231e-06, | |
| "loss": 0.8429, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 5.006701030927835, | |
| "grad_norm": 9.508393287658691, | |
| "learning_rate": 9.919816723940437e-06, | |
| "loss": 0.4463, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 5.007731958762887, | |
| "grad_norm": 8.229544639587402, | |
| "learning_rate": 9.908361970217641e-06, | |
| "loss": 0.4216, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 5.008762886597938, | |
| "grad_norm": 45.83427429199219, | |
| "learning_rate": 9.896907216494846e-06, | |
| "loss": 0.5766, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 5.00979381443299, | |
| "grad_norm": 6.862059116363525, | |
| "learning_rate": 9.885452462772052e-06, | |
| "loss": 0.587, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 5.010824742268041, | |
| "grad_norm": 8.74276065826416, | |
| "learning_rate": 9.873997709049257e-06, | |
| "loss": 0.4442, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 5.011855670103093, | |
| "grad_norm": 22.86684226989746, | |
| "learning_rate": 9.862542955326461e-06, | |
| "loss": 0.6656, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 5.012886597938144, | |
| "grad_norm": 6.2574849128723145, | |
| "learning_rate": 9.851088201603667e-06, | |
| "loss": 0.421, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 5.013917525773196, | |
| "grad_norm": 19.12122344970703, | |
| "learning_rate": 9.839633447880872e-06, | |
| "loss": 0.4395, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 5.014948453608247, | |
| "grad_norm": 47.755435943603516, | |
| "learning_rate": 9.828178694158076e-06, | |
| "loss": 0.6335, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 5.015979381443299, | |
| "grad_norm": 11.343196868896484, | |
| "learning_rate": 9.81672394043528e-06, | |
| "loss": 0.4231, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 5.0170103092783505, | |
| "grad_norm": 15.7100248336792, | |
| "learning_rate": 9.805269186712487e-06, | |
| "loss": 0.4681, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 5.018041237113402, | |
| "grad_norm": 29.985126495361328, | |
| "learning_rate": 9.793814432989691e-06, | |
| "loss": 0.7692, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 5.0190721649484535, | |
| "grad_norm": 4.043577194213867, | |
| "learning_rate": 9.782359679266896e-06, | |
| "loss": 0.5417, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 5.020103092783505, | |
| "grad_norm": 6.033001899719238, | |
| "learning_rate": 9.770904925544102e-06, | |
| "loss": 0.518, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 5.020103092783505, | |
| "eval_accuracy": 0.4666666666666667, | |
| "eval_loss": 0.8928155899047852, | |
| "eval_runtime": 13.9107, | |
| "eval_samples_per_second": 3.235, | |
| "eval_steps_per_second": 0.863, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 6.0010309278350515, | |
| "grad_norm": 6.170740604400635, | |
| "learning_rate": 9.759450171821306e-06, | |
| "loss": 0.5075, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 6.002061855670103, | |
| "grad_norm": 6.046166896820068, | |
| "learning_rate": 9.747995418098512e-06, | |
| "loss": 0.6887, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 6.003092783505155, | |
| "grad_norm": 8.674880981445312, | |
| "learning_rate": 9.736540664375717e-06, | |
| "loss": 0.5482, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 6.004123711340206, | |
| "grad_norm": 4.15226411819458, | |
| "learning_rate": 9.725085910652921e-06, | |
| "loss": 0.5811, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 6.005154639175258, | |
| "grad_norm": 9.141851425170898, | |
| "learning_rate": 9.713631156930127e-06, | |
| "loss": 0.507, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 6.006185567010309, | |
| "grad_norm": 10.06373119354248, | |
| "learning_rate": 9.702176403207332e-06, | |
| "loss": 0.6606, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 6.007216494845361, | |
| "grad_norm": 1.9274550676345825, | |
| "learning_rate": 9.690721649484536e-06, | |
| "loss": 0.4037, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 6.008247422680412, | |
| "grad_norm": 15.267620086669922, | |
| "learning_rate": 9.679266895761742e-06, | |
| "loss": 0.439, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 6.009278350515464, | |
| "grad_norm": 87.103759765625, | |
| "learning_rate": 9.667812142038947e-06, | |
| "loss": 0.6971, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 6.010309278350515, | |
| "grad_norm": 17.93890380859375, | |
| "learning_rate": 9.656357388316153e-06, | |
| "loss": 0.6488, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 6.011340206185567, | |
| "grad_norm": 13.628206253051758, | |
| "learning_rate": 9.644902634593357e-06, | |
| "loss": 0.4316, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 6.0123711340206185, | |
| "grad_norm": 1.0886458158493042, | |
| "learning_rate": 9.633447880870562e-06, | |
| "loss": 0.4252, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 6.01340206185567, | |
| "grad_norm": 0.41759127378463745, | |
| "learning_rate": 9.621993127147768e-06, | |
| "loss": 0.4604, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 6.014432989690722, | |
| "grad_norm": 7.780788421630859, | |
| "learning_rate": 9.610538373424972e-06, | |
| "loss": 0.7416, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 6.015463917525773, | |
| "grad_norm": 1.952467918395996, | |
| "learning_rate": 9.599083619702177e-06, | |
| "loss": 0.3673, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 6.016494845360825, | |
| "grad_norm": 2.227653741836548, | |
| "learning_rate": 9.587628865979383e-06, | |
| "loss": 0.5467, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 6.017525773195876, | |
| "grad_norm": 45.32661819458008, | |
| "learning_rate": 9.576174112256587e-06, | |
| "loss": 0.6066, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 6.018556701030928, | |
| "grad_norm": 6.04111385345459, | |
| "learning_rate": 9.564719358533792e-06, | |
| "loss": 0.9221, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 6.019587628865979, | |
| "grad_norm": 19.719083786010742, | |
| "learning_rate": 9.553264604810998e-06, | |
| "loss": 0.4857, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 6.020103092783505, | |
| "eval_accuracy": 0.6222222222222222, | |
| "eval_loss": 0.5889319777488708, | |
| "eval_runtime": 12.7268, | |
| "eval_samples_per_second": 3.536, | |
| "eval_steps_per_second": 0.943, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 7.000515463917526, | |
| "grad_norm": 4.581728935241699, | |
| "learning_rate": 9.541809851088203e-06, | |
| "loss": 0.488, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 7.001546391752577, | |
| "grad_norm": 27.277809143066406, | |
| "learning_rate": 9.530355097365407e-06, | |
| "loss": 0.5966, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 7.002577319587629, | |
| "grad_norm": 0.2609100341796875, | |
| "learning_rate": 9.518900343642611e-06, | |
| "loss": 0.3383, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 7.00360824742268, | |
| "grad_norm": 22.71802520751953, | |
| "learning_rate": 9.507445589919818e-06, | |
| "loss": 0.7841, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 7.004639175257732, | |
| "grad_norm": 16.892574310302734, | |
| "learning_rate": 9.495990836197022e-06, | |
| "loss": 0.4176, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 7.0056701030927835, | |
| "grad_norm": 39.4376335144043, | |
| "learning_rate": 9.484536082474226e-06, | |
| "loss": 0.3999, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 7.006701030927835, | |
| "grad_norm": 14.214164733886719, | |
| "learning_rate": 9.473081328751433e-06, | |
| "loss": 0.6387, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 7.007731958762887, | |
| "grad_norm": 11.498187065124512, | |
| "learning_rate": 9.461626575028637e-06, | |
| "loss": 0.5193, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 7.008762886597938, | |
| "grad_norm": 8.90798568725586, | |
| "learning_rate": 9.450171821305843e-06, | |
| "loss": 0.4921, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 7.00979381443299, | |
| "grad_norm": 4.212175369262695, | |
| "learning_rate": 9.438717067583048e-06, | |
| "loss": 0.4248, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 7.010824742268041, | |
| "grad_norm": 15.754873275756836, | |
| "learning_rate": 9.427262313860252e-06, | |
| "loss": 0.5522, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 7.011855670103093, | |
| "grad_norm": 20.611268997192383, | |
| "learning_rate": 9.415807560137458e-06, | |
| "loss": 0.5084, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 7.012886597938144, | |
| "grad_norm": 9.555137634277344, | |
| "learning_rate": 9.404352806414663e-06, | |
| "loss": 0.2632, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 7.013917525773196, | |
| "grad_norm": 2.367072343826294, | |
| "learning_rate": 9.392898052691867e-06, | |
| "loss": 0.4345, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 7.014948453608247, | |
| "grad_norm": 3.8319485187530518, | |
| "learning_rate": 9.381443298969073e-06, | |
| "loss": 0.5227, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 7.015979381443299, | |
| "grad_norm": 7.583319187164307, | |
| "learning_rate": 9.369988545246278e-06, | |
| "loss": 0.3003, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 7.0170103092783505, | |
| "grad_norm": 7.431451797485352, | |
| "learning_rate": 9.358533791523484e-06, | |
| "loss": 0.6448, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 7.018041237113402, | |
| "grad_norm": 32.28150939941406, | |
| "learning_rate": 9.347079037800688e-06, | |
| "loss": 0.4669, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 7.0190721649484535, | |
| "grad_norm": 24.32545280456543, | |
| "learning_rate": 9.335624284077893e-06, | |
| "loss": 0.5635, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 7.020103092783505, | |
| "grad_norm": 1.7349843978881836, | |
| "learning_rate": 9.324169530355099e-06, | |
| "loss": 0.3634, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 7.020103092783505, | |
| "eval_accuracy": 0.6444444444444445, | |
| "eval_loss": 0.8522602319717407, | |
| "eval_runtime": 12.7523, | |
| "eval_samples_per_second": 3.529, | |
| "eval_steps_per_second": 0.941, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 8.001030927835052, | |
| "grad_norm": 27.929956436157227, | |
| "learning_rate": 9.312714776632303e-06, | |
| "loss": 0.8071, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 8.002061855670103, | |
| "grad_norm": 19.774404525756836, | |
| "learning_rate": 9.301260022909508e-06, | |
| "loss": 0.714, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 8.003092783505155, | |
| "grad_norm": 5.483359336853027, | |
| "learning_rate": 9.289805269186714e-06, | |
| "loss": 0.4367, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 8.004123711340206, | |
| "grad_norm": 13.389904022216797, | |
| "learning_rate": 9.278350515463918e-06, | |
| "loss": 0.5099, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 8.005154639175258, | |
| "grad_norm": 7.315684795379639, | |
| "learning_rate": 9.266895761741125e-06, | |
| "loss": 0.4736, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 8.00618556701031, | |
| "grad_norm": 14.2125825881958, | |
| "learning_rate": 9.255441008018329e-06, | |
| "loss": 0.392, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 8.00721649484536, | |
| "grad_norm": 8.564682006835938, | |
| "learning_rate": 9.243986254295533e-06, | |
| "loss": 0.4436, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 8.008247422680412, | |
| "grad_norm": 27.293010711669922, | |
| "learning_rate": 9.23253150057274e-06, | |
| "loss": 0.4409, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 8.009278350515464, | |
| "grad_norm": 15.106826782226562, | |
| "learning_rate": 9.221076746849944e-06, | |
| "loss": 0.3801, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 8.010309278350515, | |
| "grad_norm": 22.984642028808594, | |
| "learning_rate": 9.209621993127148e-06, | |
| "loss": 0.5309, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 8.011340206185567, | |
| "grad_norm": 40.4260139465332, | |
| "learning_rate": 9.198167239404353e-06, | |
| "loss": 0.6704, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 8.012371134020619, | |
| "grad_norm": 11.083319664001465, | |
| "learning_rate": 9.186712485681557e-06, | |
| "loss": 0.651, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 8.01340206185567, | |
| "grad_norm": 6.827057838439941, | |
| "learning_rate": 9.175257731958764e-06, | |
| "loss": 0.7953, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 8.014432989690722, | |
| "grad_norm": 6.367009162902832, | |
| "learning_rate": 9.163802978235968e-06, | |
| "loss": 0.454, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 8.015463917525773, | |
| "grad_norm": 23.583059310913086, | |
| "learning_rate": 9.152348224513174e-06, | |
| "loss": 0.609, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 8.016494845360825, | |
| "grad_norm": 3.92618989944458, | |
| "learning_rate": 9.140893470790379e-06, | |
| "loss": 0.4297, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 8.017525773195876, | |
| "grad_norm": 4.04769229888916, | |
| "learning_rate": 9.129438717067583e-06, | |
| "loss": 0.3581, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 8.018556701030928, | |
| "grad_norm": 16.875659942626953, | |
| "learning_rate": 9.117983963344789e-06, | |
| "loss": 0.5501, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 8.01958762886598, | |
| "grad_norm": 2.816603183746338, | |
| "learning_rate": 9.106529209621994e-06, | |
| "loss": 0.4082, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 8.020103092783506, | |
| "eval_accuracy": 0.8666666666666667, | |
| "eval_loss": 0.36465370655059814, | |
| "eval_runtime": 12.703, | |
| "eval_samples_per_second": 3.542, | |
| "eval_steps_per_second": 0.945, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 9.000515463917527, | |
| "grad_norm": 19.659175872802734, | |
| "learning_rate": 9.095074455899198e-06, | |
| "loss": 0.2411, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 9.001546391752578, | |
| "grad_norm": 9.694845199584961, | |
| "learning_rate": 9.083619702176404e-06, | |
| "loss": 0.5444, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 9.00257731958763, | |
| "grad_norm": 51.400840759277344, | |
| "learning_rate": 9.072164948453609e-06, | |
| "loss": 0.5556, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 9.003608247422681, | |
| "grad_norm": 20.920190811157227, | |
| "learning_rate": 9.060710194730815e-06, | |
| "loss": 0.4964, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 9.004639175257733, | |
| "grad_norm": 25.781005859375, | |
| "learning_rate": 9.04925544100802e-06, | |
| "loss": 0.4506, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 9.005670103092784, | |
| "grad_norm": 21.402894973754883, | |
| "learning_rate": 9.037800687285224e-06, | |
| "loss": 0.4645, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 9.006701030927836, | |
| "grad_norm": 6.41744327545166, | |
| "learning_rate": 9.02634593356243e-06, | |
| "loss": 0.2999, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 9.007731958762887, | |
| "grad_norm": 8.043073654174805, | |
| "learning_rate": 9.014891179839634e-06, | |
| "loss": 0.8446, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 9.008762886597939, | |
| "grad_norm": 10.563735008239746, | |
| "learning_rate": 9.003436426116839e-06, | |
| "loss": 0.3282, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 9.00979381443299, | |
| "grad_norm": 5.581331253051758, | |
| "learning_rate": 8.991981672394045e-06, | |
| "loss": 0.398, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 9.010824742268042, | |
| "grad_norm": 1.6693127155303955, | |
| "learning_rate": 8.98052691867125e-06, | |
| "loss": 0.2961, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 9.011855670103094, | |
| "grad_norm": 9.04586410522461, | |
| "learning_rate": 8.969072164948455e-06, | |
| "loss": 0.2784, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 9.012886597938145, | |
| "grad_norm": 15.87678337097168, | |
| "learning_rate": 8.95761741122566e-06, | |
| "loss": 0.4126, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 9.013917525773197, | |
| "grad_norm": 53.66142272949219, | |
| "learning_rate": 8.946162657502864e-06, | |
| "loss": 0.4191, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 9.014948453608248, | |
| "grad_norm": 31.80818748474121, | |
| "learning_rate": 8.93470790378007e-06, | |
| "loss": 0.4929, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 9.0159793814433, | |
| "grad_norm": 12.100385665893555, | |
| "learning_rate": 8.923253150057275e-06, | |
| "loss": 0.7639, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 9.017010309278351, | |
| "grad_norm": 3.1334633827209473, | |
| "learning_rate": 8.91179839633448e-06, | |
| "loss": 0.5291, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 9.018041237113403, | |
| "grad_norm": 6.849935531616211, | |
| "learning_rate": 8.900343642611684e-06, | |
| "loss": 0.6711, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 9.019072164948454, | |
| "grad_norm": 18.349355697631836, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.7262, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 9.020103092783506, | |
| "grad_norm": 41.263458251953125, | |
| "learning_rate": 8.877434135166094e-06, | |
| "loss": 0.4897, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 9.020103092783506, | |
| "eval_accuracy": 0.7777777777777778, | |
| "eval_loss": 0.5648357272148132, | |
| "eval_runtime": 12.0862, | |
| "eval_samples_per_second": 3.723, | |
| "eval_steps_per_second": 0.993, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 10.001030927835052, | |
| "grad_norm": 22.894468307495117, | |
| "learning_rate": 8.865979381443299e-06, | |
| "loss": 0.4276, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 10.002061855670103, | |
| "grad_norm": 23.770139694213867, | |
| "learning_rate": 8.854524627720505e-06, | |
| "loss": 0.5575, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 10.003092783505155, | |
| "grad_norm": 7.501220703125, | |
| "learning_rate": 8.84306987399771e-06, | |
| "loss": 0.2925, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 10.004123711340206, | |
| "grad_norm": 16.40544891357422, | |
| "learning_rate": 8.831615120274914e-06, | |
| "loss": 0.4975, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 10.005154639175258, | |
| "grad_norm": 18.579830169677734, | |
| "learning_rate": 8.82016036655212e-06, | |
| "loss": 0.5642, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 10.00618556701031, | |
| "grad_norm": 10.547638893127441, | |
| "learning_rate": 8.808705612829325e-06, | |
| "loss": 0.3936, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 10.00721649484536, | |
| "grad_norm": 26.152042388916016, | |
| "learning_rate": 8.797250859106529e-06, | |
| "loss": 0.4257, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 10.008247422680412, | |
| "grad_norm": 29.884458541870117, | |
| "learning_rate": 8.785796105383735e-06, | |
| "loss": 0.6164, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 10.009278350515464, | |
| "grad_norm": 6.939556121826172, | |
| "learning_rate": 8.77434135166094e-06, | |
| "loss": 0.57, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 10.010309278350515, | |
| "grad_norm": 53.5316162109375, | |
| "learning_rate": 8.762886597938146e-06, | |
| "loss": 0.4892, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 10.011340206185567, | |
| "grad_norm": 14.673813819885254, | |
| "learning_rate": 8.75143184421535e-06, | |
| "loss": 0.3881, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 10.012371134020619, | |
| "grad_norm": 13.877957344055176, | |
| "learning_rate": 8.739977090492555e-06, | |
| "loss": 0.3445, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 10.01340206185567, | |
| "grad_norm": 0.6452627778053284, | |
| "learning_rate": 8.72852233676976e-06, | |
| "loss": 0.3769, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 10.014432989690722, | |
| "grad_norm": 27.587717056274414, | |
| "learning_rate": 8.717067583046965e-06, | |
| "loss": 0.4407, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 10.015463917525773, | |
| "grad_norm": 19.309782028198242, | |
| "learning_rate": 8.70561282932417e-06, | |
| "loss": 0.3936, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 10.016494845360825, | |
| "grad_norm": 13.032252311706543, | |
| "learning_rate": 8.694158075601376e-06, | |
| "loss": 0.3719, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 10.017525773195876, | |
| "grad_norm": 2.494875431060791, | |
| "learning_rate": 8.68270332187858e-06, | |
| "loss": 0.411, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 10.018556701030928, | |
| "grad_norm": 46.90223693847656, | |
| "learning_rate": 8.671248568155786e-06, | |
| "loss": 0.554, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 10.01958762886598, | |
| "grad_norm": 11.451055526733398, | |
| "learning_rate": 8.65979381443299e-06, | |
| "loss": 0.389, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 10.020103092783506, | |
| "eval_accuracy": 0.7777777777777778, | |
| "eval_loss": 0.5735762715339661, | |
| "eval_runtime": 12.8299, | |
| "eval_samples_per_second": 3.507, | |
| "eval_steps_per_second": 0.935, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 11.000515463917527, | |
| "grad_norm": 11.298035621643066, | |
| "learning_rate": 8.648339060710195e-06, | |
| "loss": 0.5879, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 11.001546391752578, | |
| "grad_norm": 19.15152931213379, | |
| "learning_rate": 8.636884306987401e-06, | |
| "loss": 0.4341, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 11.00257731958763, | |
| "grad_norm": 1.2121660709381104, | |
| "learning_rate": 8.625429553264606e-06, | |
| "loss": 0.2884, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 11.003608247422681, | |
| "grad_norm": 36.596675872802734, | |
| "learning_rate": 8.61397479954181e-06, | |
| "loss": 0.4636, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 11.004639175257733, | |
| "grad_norm": 14.555014610290527, | |
| "learning_rate": 8.602520045819016e-06, | |
| "loss": 0.3522, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 11.005670103092784, | |
| "grad_norm": 77.03166961669922, | |
| "learning_rate": 8.591065292096221e-06, | |
| "loss": 0.3229, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 11.006701030927836, | |
| "grad_norm": 0.5664463639259338, | |
| "learning_rate": 8.579610538373425e-06, | |
| "loss": 0.5331, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 11.007731958762887, | |
| "grad_norm": 1.5499430894851685, | |
| "learning_rate": 8.56815578465063e-06, | |
| "loss": 0.4122, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 11.008762886597939, | |
| "grad_norm": 25.140165328979492, | |
| "learning_rate": 8.556701030927836e-06, | |
| "loss": 0.9055, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 11.00979381443299, | |
| "grad_norm": 42.95240020751953, | |
| "learning_rate": 8.54524627720504e-06, | |
| "loss": 0.4527, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 11.010824742268042, | |
| "grad_norm": 35.971248626708984, | |
| "learning_rate": 8.533791523482245e-06, | |
| "loss": 0.5452, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 11.011855670103094, | |
| "grad_norm": 21.85768699645996, | |
| "learning_rate": 8.522336769759451e-06, | |
| "loss": 0.5303, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 11.012886597938145, | |
| "grad_norm": 6.7367119789123535, | |
| "learning_rate": 8.510882016036655e-06, | |
| "loss": 0.5654, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 11.013917525773197, | |
| "grad_norm": 62.03700637817383, | |
| "learning_rate": 8.49942726231386e-06, | |
| "loss": 0.3585, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 11.014948453608248, | |
| "grad_norm": 79.50271606445312, | |
| "learning_rate": 8.487972508591066e-06, | |
| "loss": 0.5117, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 11.0159793814433, | |
| "grad_norm": 12.448205947875977, | |
| "learning_rate": 8.47651775486827e-06, | |
| "loss": 0.352, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 11.017010309278351, | |
| "grad_norm": 5.7825541496276855, | |
| "learning_rate": 8.465063001145477e-06, | |
| "loss": 0.4759, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 11.018041237113403, | |
| "grad_norm": 64.92884826660156, | |
| "learning_rate": 8.453608247422681e-06, | |
| "loss": 0.4795, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 11.019072164948454, | |
| "grad_norm": 8.412237167358398, | |
| "learning_rate": 8.442153493699886e-06, | |
| "loss": 0.4374, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 11.020103092783506, | |
| "grad_norm": 0.19983382523059845, | |
| "learning_rate": 8.430698739977092e-06, | |
| "loss": 0.3753, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 11.020103092783506, | |
| "eval_accuracy": 0.5777777777777777, | |
| "eval_loss": 1.0849213600158691, | |
| "eval_runtime": 12.7494, | |
| "eval_samples_per_second": 3.53, | |
| "eval_steps_per_second": 0.941, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 12.001030927835052, | |
| "grad_norm": 0.6972218155860901, | |
| "learning_rate": 8.419243986254296e-06, | |
| "loss": 0.2594, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 12.002061855670103, | |
| "grad_norm": 27.135608673095703, | |
| "learning_rate": 8.4077892325315e-06, | |
| "loss": 0.3156, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 12.003092783505155, | |
| "grad_norm": 126.41283416748047, | |
| "learning_rate": 8.396334478808707e-06, | |
| "loss": 0.3705, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 12.004123711340206, | |
| "grad_norm": 96.77040100097656, | |
| "learning_rate": 8.384879725085911e-06, | |
| "loss": 0.503, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 12.005154639175258, | |
| "grad_norm": 0.21895720064640045, | |
| "learning_rate": 8.373424971363117e-06, | |
| "loss": 0.5193, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 12.00618556701031, | |
| "grad_norm": 0.8776753544807434, | |
| "learning_rate": 8.361970217640322e-06, | |
| "loss": 0.3678, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 12.00721649484536, | |
| "grad_norm": 0.986826479434967, | |
| "learning_rate": 8.350515463917526e-06, | |
| "loss": 0.9035, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 12.008247422680412, | |
| "grad_norm": 33.315673828125, | |
| "learning_rate": 8.339060710194732e-06, | |
| "loss": 0.5579, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 12.009278350515464, | |
| "grad_norm": 25.64463996887207, | |
| "learning_rate": 8.327605956471937e-06, | |
| "loss": 0.4433, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 12.010309278350515, | |
| "grad_norm": 61.11088180541992, | |
| "learning_rate": 8.316151202749141e-06, | |
| "loss": 0.4935, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 12.011340206185567, | |
| "grad_norm": 0.02137608826160431, | |
| "learning_rate": 8.304696449026347e-06, | |
| "loss": 0.1616, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 12.012371134020619, | |
| "grad_norm": 54.19511413574219, | |
| "learning_rate": 8.293241695303552e-06, | |
| "loss": 1.0093, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 12.01340206185567, | |
| "grad_norm": 8.395423889160156, | |
| "learning_rate": 8.281786941580758e-06, | |
| "loss": 0.1185, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 12.014432989690722, | |
| "grad_norm": 42.02037811279297, | |
| "learning_rate": 8.27033218785796e-06, | |
| "loss": 0.8235, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 12.015463917525773, | |
| "grad_norm": 22.178823471069336, | |
| "learning_rate": 8.258877434135167e-06, | |
| "loss": 0.5585, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 12.016494845360825, | |
| "grad_norm": 2.1525092124938965, | |
| "learning_rate": 8.247422680412371e-06, | |
| "loss": 0.7552, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 12.017525773195876, | |
| "grad_norm": 42.38743209838867, | |
| "learning_rate": 8.235967926689576e-06, | |
| "loss": 0.3942, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 12.018556701030928, | |
| "grad_norm": 6.113926410675049, | |
| "learning_rate": 8.224513172966782e-06, | |
| "loss": 0.4449, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 12.01958762886598, | |
| "grad_norm": 1.72330904006958, | |
| "learning_rate": 8.213058419243986e-06, | |
| "loss": 0.3118, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 12.020103092783506, | |
| "eval_accuracy": 0.6222222222222222, | |
| "eval_loss": 1.0597846508026123, | |
| "eval_runtime": 12.6829, | |
| "eval_samples_per_second": 3.548, | |
| "eval_steps_per_second": 0.946, | |
| "step": 2535 | |
| }, | |
| { | |
| "epoch": 13.000515463917527, | |
| "grad_norm": 8.192316055297852, | |
| "learning_rate": 8.201603665521193e-06, | |
| "loss": 0.4995, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 13.001546391752578, | |
| "grad_norm": 45.1771125793457, | |
| "learning_rate": 8.190148911798397e-06, | |
| "loss": 0.3395, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 13.00257731958763, | |
| "grad_norm": 42.817996978759766, | |
| "learning_rate": 8.178694158075601e-06, | |
| "loss": 0.7111, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 13.003608247422681, | |
| "grad_norm": 63.306854248046875, | |
| "learning_rate": 8.167239404352808e-06, | |
| "loss": 0.3741, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 13.004639175257733, | |
| "grad_norm": 14.730040550231934, | |
| "learning_rate": 8.155784650630012e-06, | |
| "loss": 0.4185, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 13.005670103092784, | |
| "grad_norm": 7.792994499206543, | |
| "learning_rate": 8.144329896907216e-06, | |
| "loss": 0.7574, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 13.006701030927836, | |
| "grad_norm": 3.4603207111358643, | |
| "learning_rate": 8.132875143184423e-06, | |
| "loss": 0.4004, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 13.007731958762887, | |
| "grad_norm": 34.50138854980469, | |
| "learning_rate": 8.121420389461627e-06, | |
| "loss": 0.7797, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 13.008762886597939, | |
| "grad_norm": 30.966228485107422, | |
| "learning_rate": 8.109965635738832e-06, | |
| "loss": 0.308, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 13.00979381443299, | |
| "grad_norm": 14.006564140319824, | |
| "learning_rate": 8.098510882016038e-06, | |
| "loss": 0.1939, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 13.010824742268042, | |
| "grad_norm": 31.138307571411133, | |
| "learning_rate": 8.087056128293242e-06, | |
| "loss": 0.308, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 13.011855670103094, | |
| "grad_norm": 17.796743392944336, | |
| "learning_rate": 8.075601374570448e-06, | |
| "loss": 0.3016, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 13.012886597938145, | |
| "grad_norm": 34.422054290771484, | |
| "learning_rate": 8.064146620847653e-06, | |
| "loss": 0.4823, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 13.013917525773197, | |
| "grad_norm": 0.29599234461784363, | |
| "learning_rate": 8.052691867124857e-06, | |
| "loss": 0.1329, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 13.014948453608248, | |
| "grad_norm": 46.61289978027344, | |
| "learning_rate": 8.041237113402063e-06, | |
| "loss": 0.5176, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 13.0159793814433, | |
| "grad_norm": 20.459152221679688, | |
| "learning_rate": 8.029782359679268e-06, | |
| "loss": 0.5521, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 13.017010309278351, | |
| "grad_norm": 48.8476676940918, | |
| "learning_rate": 8.018327605956472e-06, | |
| "loss": 0.3581, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 13.018041237113403, | |
| "grad_norm": 36.25406265258789, | |
| "learning_rate": 8.006872852233678e-06, | |
| "loss": 0.5387, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 13.019072164948454, | |
| "grad_norm": 15.75755786895752, | |
| "learning_rate": 7.995418098510883e-06, | |
| "loss": 0.3677, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 13.020103092783506, | |
| "grad_norm": 0.06314048171043396, | |
| "learning_rate": 7.983963344788089e-06, | |
| "loss": 0.5823, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 13.020103092783506, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 0.7085598111152649, | |
| "eval_runtime": 14.437, | |
| "eval_samples_per_second": 3.117, | |
| "eval_steps_per_second": 0.831, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 14.001030927835052, | |
| "grad_norm": 15.936594009399414, | |
| "learning_rate": 7.972508591065293e-06, | |
| "loss": 0.2965, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 14.002061855670103, | |
| "grad_norm": 34.82296371459961, | |
| "learning_rate": 7.961053837342498e-06, | |
| "loss": 0.3932, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 14.003092783505155, | |
| "grad_norm": 24.281137466430664, | |
| "learning_rate": 7.949599083619702e-06, | |
| "loss": 0.2176, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 14.004123711340206, | |
| "grad_norm": 3.835561990737915, | |
| "learning_rate": 7.938144329896907e-06, | |
| "loss": 0.3466, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 14.005154639175258, | |
| "grad_norm": 59.81629943847656, | |
| "learning_rate": 7.926689576174113e-06, | |
| "loss": 0.4342, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 14.00618556701031, | |
| "grad_norm": 21.131526947021484, | |
| "learning_rate": 7.915234822451317e-06, | |
| "loss": 0.3176, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 14.00721649484536, | |
| "grad_norm": 28.3918399810791, | |
| "learning_rate": 7.903780068728523e-06, | |
| "loss": 0.508, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 14.008247422680412, | |
| "grad_norm": 68.05310821533203, | |
| "learning_rate": 7.892325315005728e-06, | |
| "loss": 0.2753, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 14.009278350515464, | |
| "grad_norm": 15.491647720336914, | |
| "learning_rate": 7.880870561282932e-06, | |
| "loss": 0.233, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 14.010309278350515, | |
| "grad_norm": 46.456077575683594, | |
| "learning_rate": 7.869415807560138e-06, | |
| "loss": 0.2713, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 14.011340206185567, | |
| "grad_norm": 11.542130470275879, | |
| "learning_rate": 7.857961053837343e-06, | |
| "loss": 0.488, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 14.012371134020619, | |
| "grad_norm": 21.926034927368164, | |
| "learning_rate": 7.846506300114547e-06, | |
| "loss": 1.0477, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 14.01340206185567, | |
| "grad_norm": 0.11074031889438629, | |
| "learning_rate": 7.835051546391754e-06, | |
| "loss": 0.3838, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 14.014432989690722, | |
| "grad_norm": 21.328231811523438, | |
| "learning_rate": 7.823596792668958e-06, | |
| "loss": 0.6549, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 14.015463917525773, | |
| "grad_norm": 45.43595886230469, | |
| "learning_rate": 7.812142038946164e-06, | |
| "loss": 0.5422, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 14.016494845360825, | |
| "grad_norm": 43.23012161254883, | |
| "learning_rate": 7.800687285223369e-06, | |
| "loss": 0.348, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 14.017525773195876, | |
| "grad_norm": 97.96592712402344, | |
| "learning_rate": 7.789232531500573e-06, | |
| "loss": 0.3177, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 14.018556701030928, | |
| "grad_norm": 24.8525390625, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.46, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 14.01958762886598, | |
| "grad_norm": 0.3702309727668762, | |
| "learning_rate": 7.766323024054984e-06, | |
| "loss": 0.2604, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 14.020103092783506, | |
| "eval_accuracy": 0.6222222222222222, | |
| "eval_loss": 1.4168084859848022, | |
| "eval_runtime": 13.2629, | |
| "eval_samples_per_second": 3.393, | |
| "eval_steps_per_second": 0.905, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 15.000515463917527, | |
| "grad_norm": 21.372608184814453, | |
| "learning_rate": 7.754868270332188e-06, | |
| "loss": 0.5669, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 15.001546391752578, | |
| "grad_norm": 2.491281270980835, | |
| "learning_rate": 7.743413516609394e-06, | |
| "loss": 0.5194, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 15.00257731958763, | |
| "grad_norm": 30.641578674316406, | |
| "learning_rate": 7.731958762886599e-06, | |
| "loss": 0.3343, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 15.003608247422681, | |
| "grad_norm": 61.93765640258789, | |
| "learning_rate": 7.720504009163803e-06, | |
| "loss": 0.399, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 15.004639175257733, | |
| "grad_norm": 41.56774139404297, | |
| "learning_rate": 7.70904925544101e-06, | |
| "loss": 0.2883, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 15.005670103092784, | |
| "grad_norm": 31.026432037353516, | |
| "learning_rate": 7.697594501718214e-06, | |
| "loss": 0.5769, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 15.006701030927836, | |
| "grad_norm": 32.793243408203125, | |
| "learning_rate": 7.68613974799542e-06, | |
| "loss": 0.5666, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 15.007731958762887, | |
| "grad_norm": 68.84432220458984, | |
| "learning_rate": 7.674684994272624e-06, | |
| "loss": 0.5853, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 15.008762886597939, | |
| "grad_norm": 79.7370376586914, | |
| "learning_rate": 7.663230240549829e-06, | |
| "loss": 0.3274, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 15.00979381443299, | |
| "grad_norm": 0.1271493136882782, | |
| "learning_rate": 7.651775486827033e-06, | |
| "loss": 0.286, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 15.010824742268042, | |
| "grad_norm": 5.414377212524414, | |
| "learning_rate": 7.640320733104238e-06, | |
| "loss": 0.2541, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 15.011855670103094, | |
| "grad_norm": 7.974775791168213, | |
| "learning_rate": 7.628865979381444e-06, | |
| "loss": 0.6325, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 15.012886597938145, | |
| "grad_norm": 2.6058754920959473, | |
| "learning_rate": 7.617411225658648e-06, | |
| "loss": 0.6288, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 15.013917525773197, | |
| "grad_norm": 5.675416469573975, | |
| "learning_rate": 7.6059564719358535e-06, | |
| "loss": 0.2737, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 15.014948453608248, | |
| "grad_norm": 21.808191299438477, | |
| "learning_rate": 7.594501718213059e-06, | |
| "loss": 0.4687, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 15.0159793814433, | |
| "grad_norm": 4.567986011505127, | |
| "learning_rate": 7.583046964490264e-06, | |
| "loss": 0.2163, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 15.017010309278351, | |
| "grad_norm": 30.94685173034668, | |
| "learning_rate": 7.5715922107674686e-06, | |
| "loss": 0.3175, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 15.018041237113403, | |
| "grad_norm": 8.627877235412598, | |
| "learning_rate": 7.560137457044674e-06, | |
| "loss": 0.1734, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 15.019072164948454, | |
| "grad_norm": 91.24742889404297, | |
| "learning_rate": 7.548682703321879e-06, | |
| "loss": 0.3208, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 15.020103092783506, | |
| "grad_norm": 0.007905744016170502, | |
| "learning_rate": 7.5372279495990845e-06, | |
| "loss": 0.5767, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 15.020103092783506, | |
| "eval_accuracy": 0.8, | |
| "eval_loss": 0.7965757846832275, | |
| "eval_runtime": 13.3334, | |
| "eval_samples_per_second": 3.375, | |
| "eval_steps_per_second": 0.9, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 16.001030927835053, | |
| "grad_norm": 4.9155964851379395, | |
| "learning_rate": 7.525773195876289e-06, | |
| "loss": 0.3086, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 16.002061855670103, | |
| "grad_norm": 41.20609664916992, | |
| "learning_rate": 7.514318442153494e-06, | |
| "loss": 0.7778, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 16.003092783505156, | |
| "grad_norm": 31.576202392578125, | |
| "learning_rate": 7.5028636884306995e-06, | |
| "loss": 0.0978, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 16.004123711340206, | |
| "grad_norm": 0.07261794060468674, | |
| "learning_rate": 7.491408934707905e-06, | |
| "loss": 0.2957, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 16.00515463917526, | |
| "grad_norm": 34.24305725097656, | |
| "learning_rate": 7.479954180985109e-06, | |
| "loss": 0.6108, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 16.00618556701031, | |
| "grad_norm": 64.02971649169922, | |
| "learning_rate": 7.4684994272623145e-06, | |
| "loss": 0.256, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 16.007216494845363, | |
| "grad_norm": 31.043434143066406, | |
| "learning_rate": 7.45704467353952e-06, | |
| "loss": 0.8459, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 16.008247422680412, | |
| "grad_norm": 100.2379379272461, | |
| "learning_rate": 7.445589919816725e-06, | |
| "loss": 0.391, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 16.009278350515466, | |
| "grad_norm": 34.66457748413086, | |
| "learning_rate": 7.43413516609393e-06, | |
| "loss": 0.7614, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 16.010309278350515, | |
| "grad_norm": 78.97329711914062, | |
| "learning_rate": 7.422680412371135e-06, | |
| "loss": 0.5367, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 16.01134020618557, | |
| "grad_norm": 55.71527862548828, | |
| "learning_rate": 7.41122565864834e-06, | |
| "loss": 0.5274, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 16.01237113402062, | |
| "grad_norm": 10.22823715209961, | |
| "learning_rate": 7.3997709049255455e-06, | |
| "loss": 0.3361, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 16.013402061855672, | |
| "grad_norm": 2.243086814880371, | |
| "learning_rate": 7.38831615120275e-06, | |
| "loss": 0.7985, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 16.01443298969072, | |
| "grad_norm": 4.293857574462891, | |
| "learning_rate": 7.376861397479955e-06, | |
| "loss": 0.2727, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 16.015463917525775, | |
| "grad_norm": 0.2114896923303604, | |
| "learning_rate": 7.3654066437571605e-06, | |
| "loss": 0.2435, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 16.016494845360825, | |
| "grad_norm": 36.44148635864258, | |
| "learning_rate": 7.353951890034365e-06, | |
| "loss": 0.5246, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 16.017525773195878, | |
| "grad_norm": 51.74038314819336, | |
| "learning_rate": 7.342497136311569e-06, | |
| "loss": 0.0543, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 16.018556701030928, | |
| "grad_norm": 50.51845169067383, | |
| "learning_rate": 7.331042382588775e-06, | |
| "loss": 0.7459, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 16.01958762886598, | |
| "grad_norm": 37.770626068115234, | |
| "learning_rate": 7.319587628865979e-06, | |
| "loss": 0.4844, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 16.020103092783504, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 1.0488171577453613, | |
| "eval_runtime": 12.4047, | |
| "eval_samples_per_second": 3.628, | |
| "eval_steps_per_second": 0.967, | |
| "step": 3315 | |
| }, | |
| { | |
| "epoch": 17.000515463917527, | |
| "grad_norm": 97.32837677001953, | |
| "learning_rate": 7.3081328751431845e-06, | |
| "loss": 0.1819, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 17.001546391752576, | |
| "grad_norm": 4.759186744689941, | |
| "learning_rate": 7.29667812142039e-06, | |
| "loss": 0.2479, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 17.00257731958763, | |
| "grad_norm": 1.728135347366333, | |
| "learning_rate": 7.285223367697595e-06, | |
| "loss": 0.5371, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 17.00360824742268, | |
| "grad_norm": 0.7207375168800354, | |
| "learning_rate": 7.2737686139747995e-06, | |
| "loss": 0.2002, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 17.004639175257733, | |
| "grad_norm": 32.98098373413086, | |
| "learning_rate": 7.262313860252005e-06, | |
| "loss": 0.4171, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 17.005670103092783, | |
| "grad_norm": 1.2624742984771729, | |
| "learning_rate": 7.25085910652921e-06, | |
| "loss": 0.2694, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 17.006701030927836, | |
| "grad_norm": 40.10099411010742, | |
| "learning_rate": 7.239404352806415e-06, | |
| "loss": 0.4878, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 17.007731958762886, | |
| "grad_norm": 0.11596504598855972, | |
| "learning_rate": 7.22794959908362e-06, | |
| "loss": 0.5169, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 17.00876288659794, | |
| "grad_norm": 52.03158187866211, | |
| "learning_rate": 7.216494845360825e-06, | |
| "loss": 0.6374, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 17.00979381443299, | |
| "grad_norm": 20.594680786132812, | |
| "learning_rate": 7.2050400916380304e-06, | |
| "loss": 0.8605, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 17.010824742268042, | |
| "grad_norm": 61.95948791503906, | |
| "learning_rate": 7.193585337915236e-06, | |
| "loss": 0.357, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 17.011855670103092, | |
| "grad_norm": 56.67446517944336, | |
| "learning_rate": 7.18213058419244e-06, | |
| "loss": 0.3436, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 17.012886597938145, | |
| "grad_norm": 0.013172958046197891, | |
| "learning_rate": 7.1706758304696455e-06, | |
| "loss": 0.1324, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 17.013917525773195, | |
| "grad_norm": 31.45111846923828, | |
| "learning_rate": 7.159221076746851e-06, | |
| "loss": 0.1581, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 17.01494845360825, | |
| "grad_norm": 49.000885009765625, | |
| "learning_rate": 7.147766323024056e-06, | |
| "loss": 0.2419, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 17.015979381443298, | |
| "grad_norm": 36.576114654541016, | |
| "learning_rate": 7.1363115693012605e-06, | |
| "loss": 0.4095, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 17.01701030927835, | |
| "grad_norm": 2.826425552368164, | |
| "learning_rate": 7.124856815578466e-06, | |
| "loss": 0.2097, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 17.0180412371134, | |
| "grad_norm": 43.49801254272461, | |
| "learning_rate": 7.113402061855671e-06, | |
| "loss": 0.2707, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 17.019072164948454, | |
| "grad_norm": 0.675110399723053, | |
| "learning_rate": 7.101947308132876e-06, | |
| "loss": 0.3188, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 17.020103092783504, | |
| "grad_norm": 0.0034136411268264055, | |
| "learning_rate": 7.090492554410081e-06, | |
| "loss": 0.0729, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 17.020103092783504, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 1.0075429677963257, | |
| "eval_runtime": 12.9969, | |
| "eval_samples_per_second": 3.462, | |
| "eval_steps_per_second": 0.923, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 18.001030927835053, | |
| "grad_norm": 19.52350425720215, | |
| "learning_rate": 7.079037800687286e-06, | |
| "loss": 0.4801, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 18.002061855670103, | |
| "grad_norm": 0.09289558976888657, | |
| "learning_rate": 7.0675830469644914e-06, | |
| "loss": 0.2281, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 18.003092783505156, | |
| "grad_norm": 38.042137145996094, | |
| "learning_rate": 7.056128293241697e-06, | |
| "loss": 0.1157, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 18.004123711340206, | |
| "grad_norm": 2.72111439704895, | |
| "learning_rate": 7.044673539518901e-06, | |
| "loss": 0.3428, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 18.00515463917526, | |
| "grad_norm": 1.5013760328292847, | |
| "learning_rate": 7.0332187857961065e-06, | |
| "loss": 0.5245, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 18.00618556701031, | |
| "grad_norm": 123.11444854736328, | |
| "learning_rate": 7.02176403207331e-06, | |
| "loss": 0.6386, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 18.007216494845363, | |
| "grad_norm": 5.601233959197998, | |
| "learning_rate": 7.010309278350515e-06, | |
| "loss": 0.3835, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 18.008247422680412, | |
| "grad_norm": 0.00440385052934289, | |
| "learning_rate": 6.998854524627721e-06, | |
| "loss": 0.6067, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 18.009278350515466, | |
| "grad_norm": 16.199575424194336, | |
| "learning_rate": 6.987399770904926e-06, | |
| "loss": 0.4997, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 18.010309278350515, | |
| "grad_norm": 65.31697082519531, | |
| "learning_rate": 6.9759450171821304e-06, | |
| "loss": 0.3521, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 18.01134020618557, | |
| "grad_norm": 9.682928085327148, | |
| "learning_rate": 6.964490263459336e-06, | |
| "loss": 0.3728, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 18.01237113402062, | |
| "grad_norm": 61.22377014160156, | |
| "learning_rate": 6.953035509736541e-06, | |
| "loss": 0.7717, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 18.013402061855672, | |
| "grad_norm": 55.61003494262695, | |
| "learning_rate": 6.941580756013746e-06, | |
| "loss": 0.4314, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 18.01443298969072, | |
| "grad_norm": 9.212549209594727, | |
| "learning_rate": 6.930126002290951e-06, | |
| "loss": 0.5837, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 18.015463917525775, | |
| "grad_norm": 9.7285737991333, | |
| "learning_rate": 6.918671248568156e-06, | |
| "loss": 0.3758, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 18.016494845360825, | |
| "grad_norm": 40.76766586303711, | |
| "learning_rate": 6.907216494845361e-06, | |
| "loss": 0.1162, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 18.017525773195878, | |
| "grad_norm": 50.804962158203125, | |
| "learning_rate": 6.895761741122567e-06, | |
| "loss": 0.0843, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 18.018556701030928, | |
| "grad_norm": 2.414264440536499, | |
| "learning_rate": 6.884306987399771e-06, | |
| "loss": 0.1919, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 18.01958762886598, | |
| "grad_norm": 1.428833246231079, | |
| "learning_rate": 6.872852233676976e-06, | |
| "loss": 0.4188, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 18.020103092783504, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 1.2723888158798218, | |
| "eval_runtime": 12.9654, | |
| "eval_samples_per_second": 3.471, | |
| "eval_steps_per_second": 0.926, | |
| "step": 3705 | |
| }, | |
| { | |
| "epoch": 19.000515463917527, | |
| "grad_norm": 44.43056869506836, | |
| "learning_rate": 6.861397479954182e-06, | |
| "loss": 0.2822, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 19.001546391752576, | |
| "grad_norm": 0.05244099348783493, | |
| "learning_rate": 6.849942726231387e-06, | |
| "loss": 0.1549, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 19.00257731958763, | |
| "grad_norm": 0.09560086578130722, | |
| "learning_rate": 6.8384879725085914e-06, | |
| "loss": 0.1708, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 19.00360824742268, | |
| "grad_norm": 0.2790253758430481, | |
| "learning_rate": 6.827033218785797e-06, | |
| "loss": 0.0709, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 19.004639175257733, | |
| "grad_norm": 4.10269021987915, | |
| "learning_rate": 6.815578465063002e-06, | |
| "loss": 0.46, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 19.005670103092783, | |
| "grad_norm": 22.924274444580078, | |
| "learning_rate": 6.804123711340207e-06, | |
| "loss": 0.4041, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 19.006701030927836, | |
| "grad_norm": 0.1639637053012848, | |
| "learning_rate": 6.792668957617412e-06, | |
| "loss": 0.3816, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 19.007731958762886, | |
| "grad_norm": 0.21744728088378906, | |
| "learning_rate": 6.781214203894617e-06, | |
| "loss": 0.1098, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 19.00876288659794, | |
| "grad_norm": 135.28933715820312, | |
| "learning_rate": 6.769759450171822e-06, | |
| "loss": 0.3937, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 19.00979381443299, | |
| "grad_norm": 104.67581176757812, | |
| "learning_rate": 6.758304696449028e-06, | |
| "loss": 0.4015, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 19.010824742268042, | |
| "grad_norm": 68.89933776855469, | |
| "learning_rate": 6.746849942726232e-06, | |
| "loss": 0.2513, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 19.011855670103092, | |
| "grad_norm": 0.19136613607406616, | |
| "learning_rate": 6.735395189003437e-06, | |
| "loss": 0.1594, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 19.012886597938145, | |
| "grad_norm": 106.2833023071289, | |
| "learning_rate": 6.723940435280643e-06, | |
| "loss": 0.6457, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 19.013917525773195, | |
| "grad_norm": 12.311858177185059, | |
| "learning_rate": 6.712485681557846e-06, | |
| "loss": 0.2894, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 19.01494845360825, | |
| "grad_norm": 10.986001014709473, | |
| "learning_rate": 6.701030927835052e-06, | |
| "loss": 0.6104, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 19.015979381443298, | |
| "grad_norm": 9.053174018859863, | |
| "learning_rate": 6.689576174112257e-06, | |
| "loss": 0.6795, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 19.01701030927835, | |
| "grad_norm": 0.6393508911132812, | |
| "learning_rate": 6.678121420389461e-06, | |
| "loss": 0.1746, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 19.0180412371134, | |
| "grad_norm": 41.1427001953125, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.1791, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 19.019072164948454, | |
| "grad_norm": 0.29769831895828247, | |
| "learning_rate": 6.655211912943872e-06, | |
| "loss": 0.1533, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 19.020103092783504, | |
| "grad_norm": 0.005904354155063629, | |
| "learning_rate": 6.643757159221077e-06, | |
| "loss": 0.247, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 19.020103092783504, | |
| "eval_accuracy": 0.7555555555555555, | |
| "eval_loss": 1.3884011507034302, | |
| "eval_runtime": 12.8798, | |
| "eval_samples_per_second": 3.494, | |
| "eval_steps_per_second": 0.932, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 20.001030927835053, | |
| "grad_norm": 121.39861297607422, | |
| "learning_rate": 6.632302405498282e-06, | |
| "loss": 0.4984, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 20.002061855670103, | |
| "grad_norm": 80.33428192138672, | |
| "learning_rate": 6.620847651775487e-06, | |
| "loss": 0.4636, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 20.003092783505156, | |
| "grad_norm": 0.3919704854488373, | |
| "learning_rate": 6.609392898052692e-06, | |
| "loss": 0.3251, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 20.004123711340206, | |
| "grad_norm": 0.08509159088134766, | |
| "learning_rate": 6.597938144329898e-06, | |
| "loss": 0.1908, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 20.00515463917526, | |
| "grad_norm": 0.09845045208930969, | |
| "learning_rate": 6.586483390607102e-06, | |
| "loss": 0.4081, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 20.00618556701031, | |
| "grad_norm": 0.17746928334236145, | |
| "learning_rate": 6.575028636884307e-06, | |
| "loss": 0.1395, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 20.007216494845363, | |
| "grad_norm": 133.173828125, | |
| "learning_rate": 6.563573883161513e-06, | |
| "loss": 0.4535, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 20.008247422680412, | |
| "grad_norm": 0.980322003364563, | |
| "learning_rate": 6.552119129438718e-06, | |
| "loss": 0.473, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 20.009278350515466, | |
| "grad_norm": 21.204669952392578, | |
| "learning_rate": 6.540664375715922e-06, | |
| "loss": 0.2637, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 20.010309278350515, | |
| "grad_norm": 88.95044708251953, | |
| "learning_rate": 6.529209621993128e-06, | |
| "loss": 0.2103, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 20.01134020618557, | |
| "grad_norm": 0.22797317802906036, | |
| "learning_rate": 6.517754868270333e-06, | |
| "loss": 0.3344, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 20.01237113402062, | |
| "grad_norm": 0.12223013490438461, | |
| "learning_rate": 6.506300114547538e-06, | |
| "loss": 0.0848, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 20.013402061855672, | |
| "grad_norm": 0.11136689782142639, | |
| "learning_rate": 6.494845360824743e-06, | |
| "loss": 0.3754, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 20.01443298969072, | |
| "grad_norm": 0.18626107275485992, | |
| "learning_rate": 6.483390607101948e-06, | |
| "loss": 0.1341, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 20.015463917525775, | |
| "grad_norm": 10.783482551574707, | |
| "learning_rate": 6.471935853379153e-06, | |
| "loss": 0.0183, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 20.016494845360825, | |
| "grad_norm": 0.06843023747205734, | |
| "learning_rate": 6.460481099656359e-06, | |
| "loss": 0.2099, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 20.017525773195878, | |
| "grad_norm": 0.16018100082874298, | |
| "learning_rate": 6.449026345933563e-06, | |
| "loss": 0.1934, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 20.018556701030928, | |
| "grad_norm": 0.03921249136328697, | |
| "learning_rate": 6.437571592210768e-06, | |
| "loss": 0.2822, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 20.01958762886598, | |
| "grad_norm": 73.11908721923828, | |
| "learning_rate": 6.426116838487974e-06, | |
| "loss": 0.8041, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 20.020103092783504, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 1.1552423238754272, | |
| "eval_runtime": 12.8488, | |
| "eval_samples_per_second": 3.502, | |
| "eval_steps_per_second": 0.934, | |
| "step": 4095 | |
| }, | |
| { | |
| "epoch": 21.000515463917527, | |
| "grad_norm": 0.8366032242774963, | |
| "learning_rate": 6.414662084765179e-06, | |
| "loss": 0.4461, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 21.001546391752576, | |
| "grad_norm": 0.06696546822786331, | |
| "learning_rate": 6.4032073310423825e-06, | |
| "loss": 0.3449, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 21.00257731958763, | |
| "grad_norm": 55.97826385498047, | |
| "learning_rate": 6.391752577319588e-06, | |
| "loss": 0.6324, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 21.00360824742268, | |
| "grad_norm": 30.455825805664062, | |
| "learning_rate": 6.380297823596793e-06, | |
| "loss": 0.4075, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 21.004639175257733, | |
| "grad_norm": 0.05050866678357124, | |
| "learning_rate": 6.3688430698739976e-06, | |
| "loss": 0.0636, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 21.005670103092783, | |
| "grad_norm": 114.8602294921875, | |
| "learning_rate": 6.357388316151203e-06, | |
| "loss": 0.3984, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 21.006701030927836, | |
| "grad_norm": 8.037965774536133, | |
| "learning_rate": 6.345933562428408e-06, | |
| "loss": 0.1513, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 21.007731958762886, | |
| "grad_norm": 0.20525996387004852, | |
| "learning_rate": 6.3344788087056135e-06, | |
| "loss": 0.1569, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 21.00876288659794, | |
| "grad_norm": 0.16769638657569885, | |
| "learning_rate": 6.323024054982818e-06, | |
| "loss": 0.2572, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 21.00979381443299, | |
| "grad_norm": 12.56114387512207, | |
| "learning_rate": 6.311569301260023e-06, | |
| "loss": 0.1308, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 21.010824742268042, | |
| "grad_norm": 0.05304228886961937, | |
| "learning_rate": 6.3001145475372285e-06, | |
| "loss": 0.2517, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 21.011855670103092, | |
| "grad_norm": 0.02033202536404133, | |
| "learning_rate": 6.288659793814433e-06, | |
| "loss": 0.1589, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 21.012886597938145, | |
| "grad_norm": 40.56743240356445, | |
| "learning_rate": 6.277205040091638e-06, | |
| "loss": 0.1517, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 21.013917525773195, | |
| "grad_norm": 0.18166792392730713, | |
| "learning_rate": 6.2657502863688435e-06, | |
| "loss": 0.1245, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 21.01494845360825, | |
| "grad_norm": 168.2699432373047, | |
| "learning_rate": 6.254295532646049e-06, | |
| "loss": 0.751, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 21.015979381443298, | |
| "grad_norm": 63.15861511230469, | |
| "learning_rate": 6.242840778923253e-06, | |
| "loss": 0.4252, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 21.01701030927835, | |
| "grad_norm": 0.03196326643228531, | |
| "learning_rate": 6.231386025200459e-06, | |
| "loss": 0.188, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 21.0180412371134, | |
| "grad_norm": 0.10200751572847366, | |
| "learning_rate": 6.219931271477664e-06, | |
| "loss": 0.44, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 21.019072164948454, | |
| "grad_norm": 0.9600178003311157, | |
| "learning_rate": 6.208476517754869e-06, | |
| "loss": 0.6086, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 21.020103092783504, | |
| "grad_norm": 0.15486416220664978, | |
| "learning_rate": 6.197021764032074e-06, | |
| "loss": 0.168, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 21.020103092783504, | |
| "eval_accuracy": 0.6888888888888889, | |
| "eval_loss": 1.7923972606658936, | |
| "eval_runtime": 13.1906, | |
| "eval_samples_per_second": 3.412, | |
| "eval_steps_per_second": 0.91, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 22.001030927835053, | |
| "grad_norm": 0.07751865684986115, | |
| "learning_rate": 6.185567010309279e-06, | |
| "loss": 0.187, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 22.002061855670103, | |
| "grad_norm": 10.4796142578125, | |
| "learning_rate": 6.174112256586484e-06, | |
| "loss": 0.6962, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 22.003092783505156, | |
| "grad_norm": 92.16351318359375, | |
| "learning_rate": 6.1626575028636895e-06, | |
| "loss": 0.2417, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 22.004123711340206, | |
| "grad_norm": 45.09387969970703, | |
| "learning_rate": 6.151202749140894e-06, | |
| "loss": 0.3242, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 22.00515463917526, | |
| "grad_norm": 78.62527465820312, | |
| "learning_rate": 6.139747995418099e-06, | |
| "loss": 0.1895, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 22.00618556701031, | |
| "grad_norm": 24.385234832763672, | |
| "learning_rate": 6.1282932416953046e-06, | |
| "loss": 0.487, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 22.007216494845363, | |
| "grad_norm": 0.06612977385520935, | |
| "learning_rate": 6.11683848797251e-06, | |
| "loss": 0.3641, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 22.008247422680412, | |
| "grad_norm": 37.19966125488281, | |
| "learning_rate": 6.105383734249714e-06, | |
| "loss": 0.4266, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 22.009278350515466, | |
| "grad_norm": 121.07097625732422, | |
| "learning_rate": 6.09392898052692e-06, | |
| "loss": 0.4789, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 22.010309278350515, | |
| "grad_norm": 40.00523376464844, | |
| "learning_rate": 6.082474226804124e-06, | |
| "loss": 0.4866, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 22.01134020618557, | |
| "grad_norm": 0.05597120150923729, | |
| "learning_rate": 6.0710194730813285e-06, | |
| "loss": 0.3815, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 22.01237113402062, | |
| "grad_norm": 0.8036267757415771, | |
| "learning_rate": 6.059564719358534e-06, | |
| "loss": 0.1899, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 22.013402061855672, | |
| "grad_norm": 144.4720001220703, | |
| "learning_rate": 6.048109965635739e-06, | |
| "loss": 0.4231, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 22.01443298969072, | |
| "grad_norm": 0.2990947961807251, | |
| "learning_rate": 6.036655211912944e-06, | |
| "loss": 0.2152, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 22.015463917525775, | |
| "grad_norm": 0.10453028976917267, | |
| "learning_rate": 6.025200458190149e-06, | |
| "loss": 0.404, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 22.016494845360825, | |
| "grad_norm": 112.81282043457031, | |
| "learning_rate": 6.013745704467354e-06, | |
| "loss": 0.1327, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 22.017525773195878, | |
| "grad_norm": 0.112935371696949, | |
| "learning_rate": 6.0022909507445594e-06, | |
| "loss": 0.117, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 22.018556701030928, | |
| "grad_norm": 4.366937160491943, | |
| "learning_rate": 5.990836197021765e-06, | |
| "loss": 0.4038, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 22.01958762886598, | |
| "grad_norm": 0.005484442692250013, | |
| "learning_rate": 5.979381443298969e-06, | |
| "loss": 0.2564, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 22.020103092783504, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 1.5681549310684204, | |
| "eval_runtime": 21.1102, | |
| "eval_samples_per_second": 2.132, | |
| "eval_steps_per_second": 0.568, | |
| "step": 4485 | |
| }, | |
| { | |
| "epoch": 23.000515463917527, | |
| "grad_norm": 14.104279518127441, | |
| "learning_rate": 5.9679266895761745e-06, | |
| "loss": 0.8201, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 23.001546391752576, | |
| "grad_norm": 1.4341076612472534, | |
| "learning_rate": 5.95647193585338e-06, | |
| "loss": 0.0827, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 23.00257731958763, | |
| "grad_norm": 13.500885963439941, | |
| "learning_rate": 5.945017182130585e-06, | |
| "loss": 0.1132, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 23.00360824742268, | |
| "grad_norm": 0.08413302898406982, | |
| "learning_rate": 5.9335624284077895e-06, | |
| "loss": 0.2336, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 23.004639175257733, | |
| "grad_norm": 18.12093734741211, | |
| "learning_rate": 5.922107674684995e-06, | |
| "loss": 0.4751, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 23.005670103092783, | |
| "grad_norm": 1.0863031148910522, | |
| "learning_rate": 5.9106529209622e-06, | |
| "loss": 0.1428, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 23.006701030927836, | |
| "grad_norm": 0.06023244187235832, | |
| "learning_rate": 5.8991981672394046e-06, | |
| "loss": 0.1603, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 23.007731958762886, | |
| "grad_norm": 2.2026829719543457, | |
| "learning_rate": 5.88774341351661e-06, | |
| "loss": 0.3565, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 23.00876288659794, | |
| "grad_norm": 100.63137817382812, | |
| "learning_rate": 5.876288659793815e-06, | |
| "loss": 0.1315, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 23.00979381443299, | |
| "grad_norm": 399.76123046875, | |
| "learning_rate": 5.8648339060710204e-06, | |
| "loss": 0.3714, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 23.010824742268042, | |
| "grad_norm": 154.24449157714844, | |
| "learning_rate": 5.853379152348225e-06, | |
| "loss": 0.3266, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 23.011855670103092, | |
| "grad_norm": 0.03562017157673836, | |
| "learning_rate": 5.84192439862543e-06, | |
| "loss": 0.3065, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 23.012886597938145, | |
| "grad_norm": 0.08134841173887253, | |
| "learning_rate": 5.8304696449026355e-06, | |
| "loss": 0.3782, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 23.013917525773195, | |
| "grad_norm": 181.0287628173828, | |
| "learning_rate": 5.819014891179841e-06, | |
| "loss": 0.4084, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 23.01494845360825, | |
| "grad_norm": 2.42154598236084, | |
| "learning_rate": 5.807560137457045e-06, | |
| "loss": 0.0032, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 23.015979381443298, | |
| "grad_norm": 123.42518615722656, | |
| "learning_rate": 5.7961053837342505e-06, | |
| "loss": 0.4781, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 23.01701030927835, | |
| "grad_norm": 3.3009181022644043, | |
| "learning_rate": 5.784650630011456e-06, | |
| "loss": 0.0389, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 23.0180412371134, | |
| "grad_norm": 0.07052547484636307, | |
| "learning_rate": 5.7731958762886594e-06, | |
| "loss": 0.3134, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 23.019072164948454, | |
| "grad_norm": 0.1382911652326584, | |
| "learning_rate": 5.761741122565865e-06, | |
| "loss": 0.5257, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 23.020103092783504, | |
| "grad_norm": 0.008108101785182953, | |
| "learning_rate": 5.75028636884307e-06, | |
| "loss": 0.2034, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 23.020103092783504, | |
| "eval_accuracy": 0.6888888888888889, | |
| "eval_loss": 1.6061440706253052, | |
| "eval_runtime": 12.9527, | |
| "eval_samples_per_second": 3.474, | |
| "eval_steps_per_second": 0.926, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 24.001030927835053, | |
| "grad_norm": 0.15025880932807922, | |
| "learning_rate": 5.738831615120275e-06, | |
| "loss": 0.0039, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 24.002061855670103, | |
| "grad_norm": 112.1488037109375, | |
| "learning_rate": 5.72737686139748e-06, | |
| "loss": 0.1306, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 24.003092783505156, | |
| "grad_norm": 0.02043057605624199, | |
| "learning_rate": 5.715922107674685e-06, | |
| "loss": 0.091, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 24.004123711340206, | |
| "grad_norm": 51.4376220703125, | |
| "learning_rate": 5.70446735395189e-06, | |
| "loss": 0.2416, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 24.00515463917526, | |
| "grad_norm": 47.24472427368164, | |
| "learning_rate": 5.693012600229096e-06, | |
| "loss": 0.4605, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 24.00618556701031, | |
| "grad_norm": 164.00759887695312, | |
| "learning_rate": 5.6815578465063e-06, | |
| "loss": 0.3363, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 24.007216494845363, | |
| "grad_norm": 0.009168693795800209, | |
| "learning_rate": 5.670103092783505e-06, | |
| "loss": 0.0162, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 24.008247422680412, | |
| "grad_norm": 38.39619827270508, | |
| "learning_rate": 5.658648339060711e-06, | |
| "loss": 0.4564, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 24.009278350515466, | |
| "grad_norm": 1.3691141605377197, | |
| "learning_rate": 5.647193585337916e-06, | |
| "loss": 0.2679, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 24.010309278350515, | |
| "grad_norm": 269.83489990234375, | |
| "learning_rate": 5.6357388316151204e-06, | |
| "loss": 0.4089, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 24.01134020618557, | |
| "grad_norm": 0.0756688266992569, | |
| "learning_rate": 5.624284077892326e-06, | |
| "loss": 0.1391, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 24.01237113402062, | |
| "grad_norm": 0.003854341572150588, | |
| "learning_rate": 5.612829324169531e-06, | |
| "loss": 0.548, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 24.013402061855672, | |
| "grad_norm": 93.3668212890625, | |
| "learning_rate": 5.601374570446736e-06, | |
| "loss": 0.2312, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 24.01443298969072, | |
| "grad_norm": 0.05672283470630646, | |
| "learning_rate": 5.589919816723941e-06, | |
| "loss": 0.1191, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 24.015463917525775, | |
| "grad_norm": 2.6616814136505127, | |
| "learning_rate": 5.578465063001146e-06, | |
| "loss": 0.0088, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 24.016494845360825, | |
| "grad_norm": 1.235160231590271, | |
| "learning_rate": 5.567010309278351e-06, | |
| "loss": 0.2314, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 24.017525773195878, | |
| "grad_norm": 14.321471214294434, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.35, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 24.018556701030928, | |
| "grad_norm": 0.02161994017660618, | |
| "learning_rate": 5.544100801832761e-06, | |
| "loss": 0.4611, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 24.01958762886598, | |
| "grad_norm": 0.03108271211385727, | |
| "learning_rate": 5.532646048109966e-06, | |
| "loss": 0.436, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 24.020103092783504, | |
| "eval_accuracy": 0.6888888888888889, | |
| "eval_loss": 1.5507913827896118, | |
| "eval_runtime": 12.8715, | |
| "eval_samples_per_second": 3.496, | |
| "eval_steps_per_second": 0.932, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 25.000515463917527, | |
| "grad_norm": 161.4268035888672, | |
| "learning_rate": 5.521191294387172e-06, | |
| "loss": 0.5077, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 25.001546391752576, | |
| "grad_norm": 0.3123011291027069, | |
| "learning_rate": 5.509736540664376e-06, | |
| "loss": 0.1336, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 25.00257731958763, | |
| "grad_norm": 138.39981079101562, | |
| "learning_rate": 5.4982817869415815e-06, | |
| "loss": 0.5236, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 25.00360824742268, | |
| "grad_norm": 0.1495334655046463, | |
| "learning_rate": 5.486827033218787e-06, | |
| "loss": 0.3321, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 25.004639175257733, | |
| "grad_norm": 122.72096252441406, | |
| "learning_rate": 5.475372279495992e-06, | |
| "loss": 0.7098, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 25.005670103092783, | |
| "grad_norm": 13.088062286376953, | |
| "learning_rate": 5.463917525773196e-06, | |
| "loss": 0.3201, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 25.006701030927836, | |
| "grad_norm": 0.04432636499404907, | |
| "learning_rate": 5.452462772050401e-06, | |
| "loss": 0.1142, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 25.007731958762886, | |
| "grad_norm": 0.874567985534668, | |
| "learning_rate": 5.441008018327606e-06, | |
| "loss": 0.1487, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 25.00876288659794, | |
| "grad_norm": 0.19093450903892517, | |
| "learning_rate": 5.429553264604811e-06, | |
| "loss": 0.1065, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 25.00979381443299, | |
| "grad_norm": 0.11419642716646194, | |
| "learning_rate": 5.418098510882016e-06, | |
| "loss": 0.199, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 25.010824742268042, | |
| "grad_norm": 4.453356742858887, | |
| "learning_rate": 5.406643757159221e-06, | |
| "loss": 0.0043, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 25.011855670103092, | |
| "grad_norm": 0.0850525051355362, | |
| "learning_rate": 5.395189003436427e-06, | |
| "loss": 0.0794, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 25.012886597938145, | |
| "grad_norm": 0.13754314184188843, | |
| "learning_rate": 5.383734249713631e-06, | |
| "loss": 0.2092, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 25.013917525773195, | |
| "grad_norm": 157.50311279296875, | |
| "learning_rate": 5.372279495990836e-06, | |
| "loss": 0.4186, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 25.01494845360825, | |
| "grad_norm": 0.975379228591919, | |
| "learning_rate": 5.360824742268042e-06, | |
| "loss": 0.1117, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 25.015979381443298, | |
| "grad_norm": 0.030048053711652756, | |
| "learning_rate": 5.349369988545247e-06, | |
| "loss": 0.1822, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 25.01701030927835, | |
| "grad_norm": 0.09299586713314056, | |
| "learning_rate": 5.337915234822451e-06, | |
| "loss": 0.2461, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 25.0180412371134, | |
| "grad_norm": 0.050019096583127975, | |
| "learning_rate": 5.326460481099657e-06, | |
| "loss": 0.024, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 25.019072164948454, | |
| "grad_norm": 152.73289489746094, | |
| "learning_rate": 5.315005727376862e-06, | |
| "loss": 0.2815, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 25.020103092783504, | |
| "grad_norm": 0.0027058201376348734, | |
| "learning_rate": 5.303550973654067e-06, | |
| "loss": 0.379, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 25.020103092783504, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 1.6197640895843506, | |
| "eval_runtime": 12.8821, | |
| "eval_samples_per_second": 3.493, | |
| "eval_steps_per_second": 0.932, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 26.001030927835053, | |
| "grad_norm": 0.036444660276174545, | |
| "learning_rate": 5.292096219931272e-06, | |
| "loss": 0.2972, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 26.002061855670103, | |
| "grad_norm": 8.088865280151367, | |
| "learning_rate": 5.280641466208477e-06, | |
| "loss": 0.0962, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 26.003092783505156, | |
| "grad_norm": 0.029521504417061806, | |
| "learning_rate": 5.269186712485682e-06, | |
| "loss": 0.0095, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 26.004123711340206, | |
| "grad_norm": 0.20875123143196106, | |
| "learning_rate": 5.257731958762888e-06, | |
| "loss": 0.4124, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 26.00515463917526, | |
| "grad_norm": 148.48397827148438, | |
| "learning_rate": 5.246277205040092e-06, | |
| "loss": 0.3919, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 26.00618556701031, | |
| "grad_norm": 0.03706370294094086, | |
| "learning_rate": 5.234822451317297e-06, | |
| "loss": 0.1514, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 26.007216494845363, | |
| "grad_norm": 0.0603543259203434, | |
| "learning_rate": 5.223367697594503e-06, | |
| "loss": 0.085, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 26.008247422680412, | |
| "grad_norm": 268.0271301269531, | |
| "learning_rate": 5.211912943871708e-06, | |
| "loss": 0.1849, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 26.009278350515466, | |
| "grad_norm": 0.05716723948717117, | |
| "learning_rate": 5.200458190148912e-06, | |
| "loss": 0.0813, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 26.010309278350515, | |
| "grad_norm": 36.733333587646484, | |
| "learning_rate": 5.189003436426118e-06, | |
| "loss": 0.1493, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 26.01134020618557, | |
| "grad_norm": 73.86190032958984, | |
| "learning_rate": 5.177548682703323e-06, | |
| "loss": 0.1373, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 26.01237113402062, | |
| "grad_norm": 0.04079816862940788, | |
| "learning_rate": 5.166093928980528e-06, | |
| "loss": 0.2272, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 26.013402061855672, | |
| "grad_norm": 0.07667230069637299, | |
| "learning_rate": 5.154639175257732e-06, | |
| "loss": 0.0554, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 26.01443298969072, | |
| "grad_norm": 0.8772750496864319, | |
| "learning_rate": 5.143184421534937e-06, | |
| "loss": 0.2251, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 26.015463917525775, | |
| "grad_norm": 0.012073386460542679, | |
| "learning_rate": 5.131729667812142e-06, | |
| "loss": 0.098, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 26.016494845360825, | |
| "grad_norm": 0.23029600083827972, | |
| "learning_rate": 5.120274914089347e-06, | |
| "loss": 0.1361, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 26.017525773195878, | |
| "grad_norm": 152.81813049316406, | |
| "learning_rate": 5.108820160366552e-06, | |
| "loss": 0.1956, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 26.018556701030928, | |
| "grad_norm": 0.01020987518131733, | |
| "learning_rate": 5.0973654066437575e-06, | |
| "loss": 0.3316, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 26.01958762886598, | |
| "grad_norm": 0.03010565973818302, | |
| "learning_rate": 5.085910652920962e-06, | |
| "loss": 0.0726, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 26.020103092783504, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 2.329340696334839, | |
| "eval_runtime": 12.9666, | |
| "eval_samples_per_second": 3.47, | |
| "eval_steps_per_second": 0.925, | |
| "step": 5265 | |
| }, | |
| { | |
| "epoch": 27.000515463917527, | |
| "grad_norm": 301.4482116699219, | |
| "learning_rate": 5.074455899198167e-06, | |
| "loss": 0.3906, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 27.001546391752576, | |
| "grad_norm": 0.012246874161064625, | |
| "learning_rate": 5.0630011454753726e-06, | |
| "loss": 0.3566, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 27.00257731958763, | |
| "grad_norm": 0.036673251539468765, | |
| "learning_rate": 5.051546391752578e-06, | |
| "loss": 0.0058, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 27.00360824742268, | |
| "grad_norm": 176.9713592529297, | |
| "learning_rate": 5.040091638029782e-06, | |
| "loss": 0.0528, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 27.004639175257733, | |
| "grad_norm": 110.79297637939453, | |
| "learning_rate": 5.028636884306988e-06, | |
| "loss": 0.1249, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 27.005670103092783, | |
| "grad_norm": 37.024574279785156, | |
| "learning_rate": 5.017182130584193e-06, | |
| "loss": 0.2464, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 27.006701030927836, | |
| "grad_norm": 0.11262047290802002, | |
| "learning_rate": 5.005727376861398e-06, | |
| "loss": 0.1313, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 27.007731958762886, | |
| "grad_norm": 0.6237982511520386, | |
| "learning_rate": 4.994272623138603e-06, | |
| "loss": 0.2988, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 27.00876288659794, | |
| "grad_norm": 0.011754998005926609, | |
| "learning_rate": 4.982817869415808e-06, | |
| "loss": 0.5804, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 27.00979381443299, | |
| "grad_norm": 0.027967087924480438, | |
| "learning_rate": 4.971363115693013e-06, | |
| "loss": 0.4306, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 27.010824742268042, | |
| "grad_norm": 0.041461341083049774, | |
| "learning_rate": 4.9599083619702185e-06, | |
| "loss": 0.0075, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 27.011855670103092, | |
| "grad_norm": 0.10922248661518097, | |
| "learning_rate": 4.948453608247423e-06, | |
| "loss": 0.5134, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 27.012886597938145, | |
| "grad_norm": 0.01793413981795311, | |
| "learning_rate": 4.936998854524628e-06, | |
| "loss": 0.4059, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 27.013917525773195, | |
| "grad_norm": 0.12545382976531982, | |
| "learning_rate": 4.9255441008018336e-06, | |
| "loss": 0.0593, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 27.01494845360825, | |
| "grad_norm": 0.23186025023460388, | |
| "learning_rate": 4.914089347079038e-06, | |
| "loss": 0.1651, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 27.015979381443298, | |
| "grad_norm": 24.21805191040039, | |
| "learning_rate": 4.902634593356243e-06, | |
| "loss": 0.5709, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 27.01701030927835, | |
| "grad_norm": 0.19743561744689941, | |
| "learning_rate": 4.891179839633448e-06, | |
| "loss": 0.2359, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 27.0180412371134, | |
| "grad_norm": 99.03250122070312, | |
| "learning_rate": 4.879725085910653e-06, | |
| "loss": 0.3897, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 27.019072164948454, | |
| "grad_norm": 0.35776469111442566, | |
| "learning_rate": 4.868270332187858e-06, | |
| "loss": 0.1425, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 27.020103092783504, | |
| "grad_norm": 0.001978857209905982, | |
| "learning_rate": 4.856815578465064e-06, | |
| "loss": 0.0099, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 27.020103092783504, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 1.765753984451294, | |
| "eval_runtime": 12.8308, | |
| "eval_samples_per_second": 3.507, | |
| "eval_steps_per_second": 0.935, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 28.001030927835053, | |
| "grad_norm": 0.10768745839595795, | |
| "learning_rate": 4.845360824742268e-06, | |
| "loss": 0.2311, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 28.002061855670103, | |
| "grad_norm": 57.12821960449219, | |
| "learning_rate": 4.833906071019473e-06, | |
| "loss": 0.1598, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 28.003092783505156, | |
| "grad_norm": 0.11094794422388077, | |
| "learning_rate": 4.822451317296679e-06, | |
| "loss": 0.2435, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 28.004123711340206, | |
| "grad_norm": 0.04346878081560135, | |
| "learning_rate": 4.810996563573884e-06, | |
| "loss": 0.4259, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 28.00515463917526, | |
| "grad_norm": 99.23379516601562, | |
| "learning_rate": 4.7995418098510884e-06, | |
| "loss": 0.4067, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 28.00618556701031, | |
| "grad_norm": 0.011886666528880596, | |
| "learning_rate": 4.788087056128294e-06, | |
| "loss": 0.3366, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 28.007216494845363, | |
| "grad_norm": 17.010530471801758, | |
| "learning_rate": 4.776632302405499e-06, | |
| "loss": 0.5229, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 28.008247422680412, | |
| "grad_norm": 84.538818359375, | |
| "learning_rate": 4.7651775486827035e-06, | |
| "loss": 0.2171, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 28.009278350515466, | |
| "grad_norm": 0.029075944796204567, | |
| "learning_rate": 4.753722794959909e-06, | |
| "loss": 0.2936, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 28.010309278350515, | |
| "grad_norm": 0.5513918399810791, | |
| "learning_rate": 4.742268041237113e-06, | |
| "loss": 0.1806, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 28.01134020618557, | |
| "grad_norm": 0.14319252967834473, | |
| "learning_rate": 4.7308132875143185e-06, | |
| "loss": 0.5482, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 28.01237113402062, | |
| "grad_norm": 0.33188721537590027, | |
| "learning_rate": 4.719358533791524e-06, | |
| "loss": 0.2692, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 28.013402061855672, | |
| "grad_norm": 0.03212364390492439, | |
| "learning_rate": 4.707903780068729e-06, | |
| "loss": 0.0034, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 28.01443298969072, | |
| "grad_norm": 0.03373891860246658, | |
| "learning_rate": 4.6964490263459336e-06, | |
| "loss": 0.0255, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 28.015463917525775, | |
| "grad_norm": 200.98361206054688, | |
| "learning_rate": 4.684994272623139e-06, | |
| "loss": 0.3354, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 28.016494845360825, | |
| "grad_norm": 41.92877960205078, | |
| "learning_rate": 4.673539518900344e-06, | |
| "loss": 0.0269, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 28.017525773195878, | |
| "grad_norm": 0.11387676745653152, | |
| "learning_rate": 4.6620847651775494e-06, | |
| "loss": 0.0015, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 28.018556701030928, | |
| "grad_norm": 0.05513579770922661, | |
| "learning_rate": 4.650630011454754e-06, | |
| "loss": 0.4027, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 28.01958762886598, | |
| "grad_norm": 0.003919053822755814, | |
| "learning_rate": 4.639175257731959e-06, | |
| "loss": 0.0346, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 28.020103092783504, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 1.5937350988388062, | |
| "eval_runtime": 11.439, | |
| "eval_samples_per_second": 3.934, | |
| "eval_steps_per_second": 1.049, | |
| "step": 5655 | |
| }, | |
| { | |
| "epoch": 29.000515463917527, | |
| "grad_norm": 0.547856867313385, | |
| "learning_rate": 4.6277205040091645e-06, | |
| "loss": 0.1152, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 29.001546391752576, | |
| "grad_norm": 0.010174279101192951, | |
| "learning_rate": 4.61626575028637e-06, | |
| "loss": 0.1109, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 29.00257731958763, | |
| "grad_norm": 0.08245531469583511, | |
| "learning_rate": 4.604810996563574e-06, | |
| "loss": 0.3028, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 29.00360824742268, | |
| "grad_norm": 0.2131088227033615, | |
| "learning_rate": 4.593356242840779e-06, | |
| "loss": 0.1538, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 29.004639175257733, | |
| "grad_norm": 128.2372589111328, | |
| "learning_rate": 4.581901489117984e-06, | |
| "loss": 0.4187, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 29.005670103092783, | |
| "grad_norm": 3.6024224758148193, | |
| "learning_rate": 4.570446735395189e-06, | |
| "loss": 0.1509, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 29.006701030927836, | |
| "grad_norm": 6.2508931159973145, | |
| "learning_rate": 4.5589919816723946e-06, | |
| "loss": 0.3448, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 29.007731958762886, | |
| "grad_norm": 25.978151321411133, | |
| "learning_rate": 4.547537227949599e-06, | |
| "loss": 0.378, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 29.00876288659794, | |
| "grad_norm": 97.8603286743164, | |
| "learning_rate": 4.536082474226804e-06, | |
| "loss": 0.3905, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 29.00979381443299, | |
| "grad_norm": 58.110443115234375, | |
| "learning_rate": 4.52462772050401e-06, | |
| "loss": 0.1606, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 29.010824742268042, | |
| "grad_norm": 0.03289901092648506, | |
| "learning_rate": 4.513172966781215e-06, | |
| "loss": 0.0116, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 29.011855670103092, | |
| "grad_norm": 0.4851415455341339, | |
| "learning_rate": 4.501718213058419e-06, | |
| "loss": 0.0014, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 29.012886597938145, | |
| "grad_norm": 0.6099609732627869, | |
| "learning_rate": 4.490263459335625e-06, | |
| "loss": 0.2822, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 29.013917525773195, | |
| "grad_norm": 0.035203512758016586, | |
| "learning_rate": 4.47880870561283e-06, | |
| "loss": 0.1903, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 29.01494845360825, | |
| "grad_norm": 0.075258269906044, | |
| "learning_rate": 4.467353951890035e-06, | |
| "loss": 0.2484, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 29.015979381443298, | |
| "grad_norm": 0.014334428124129772, | |
| "learning_rate": 4.45589919816724e-06, | |
| "loss": 0.2231, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 29.01701030927835, | |
| "grad_norm": 0.03960711136460304, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.203, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 29.0180412371134, | |
| "grad_norm": 0.082605741918087, | |
| "learning_rate": 4.4329896907216494e-06, | |
| "loss": 0.009, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 29.019072164948454, | |
| "grad_norm": 1.0183836221694946, | |
| "learning_rate": 4.421534936998855e-06, | |
| "loss": 0.3987, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 29.020103092783504, | |
| "grad_norm": 0.003887833096086979, | |
| "learning_rate": 4.41008018327606e-06, | |
| "loss": 0.0058, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 29.020103092783504, | |
| "eval_accuracy": 0.6444444444444445, | |
| "eval_loss": 2.351128578186035, | |
| "eval_runtime": 12.9569, | |
| "eval_samples_per_second": 3.473, | |
| "eval_steps_per_second": 0.926, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 30.001030927835053, | |
| "grad_norm": 5.273438453674316, | |
| "learning_rate": 4.3986254295532645e-06, | |
| "loss": 0.4555, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 30.002061855670103, | |
| "grad_norm": 0.054419275373220444, | |
| "learning_rate": 4.38717067583047e-06, | |
| "loss": 0.0564, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 30.003092783505156, | |
| "grad_norm": 23.50123405456543, | |
| "learning_rate": 4.375715922107675e-06, | |
| "loss": 0.0966, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 30.004123711340206, | |
| "grad_norm": 196.958251953125, | |
| "learning_rate": 4.36426116838488e-06, | |
| "loss": 0.133, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 30.00515463917526, | |
| "grad_norm": 52.535770416259766, | |
| "learning_rate": 4.352806414662085e-06, | |
| "loss": 0.2876, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 30.00618556701031, | |
| "grad_norm": 0.10040858387947083, | |
| "learning_rate": 4.34135166093929e-06, | |
| "loss": 0.2648, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 30.007216494845363, | |
| "grad_norm": 0.025616025552153587, | |
| "learning_rate": 4.329896907216495e-06, | |
| "loss": 0.3765, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 30.008247422680412, | |
| "grad_norm": 0.006266108714044094, | |
| "learning_rate": 4.318442153493701e-06, | |
| "loss": 0.1254, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 30.009278350515466, | |
| "grad_norm": 0.8512455224990845, | |
| "learning_rate": 4.306987399770905e-06, | |
| "loss": 0.0057, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 30.010309278350515, | |
| "grad_norm": 0.033568281680345535, | |
| "learning_rate": 4.2955326460481105e-06, | |
| "loss": 0.1417, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 30.01134020618557, | |
| "grad_norm": 30.20100212097168, | |
| "learning_rate": 4.284077892325315e-06, | |
| "loss": 0.0088, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 30.01237113402062, | |
| "grad_norm": 0.2173480987548828, | |
| "learning_rate": 4.27262313860252e-06, | |
| "loss": 0.1447, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 30.013402061855672, | |
| "grad_norm": 0.009785875678062439, | |
| "learning_rate": 4.2611683848797255e-06, | |
| "loss": 0.002, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 30.01443298969072, | |
| "grad_norm": 0.0599403902888298, | |
| "learning_rate": 4.24971363115693e-06, | |
| "loss": 0.0011, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 30.015463917525775, | |
| "grad_norm": 0.23365233838558197, | |
| "learning_rate": 4.238258877434135e-06, | |
| "loss": 0.2509, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 30.016494845360825, | |
| "grad_norm": 0.14375662803649902, | |
| "learning_rate": 4.2268041237113405e-06, | |
| "loss": 0.4676, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 30.017525773195878, | |
| "grad_norm": 0.03053143061697483, | |
| "learning_rate": 4.215349369988546e-06, | |
| "loss": 0.1415, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 30.018556701030928, | |
| "grad_norm": 0.08184775710105896, | |
| "learning_rate": 4.20389461626575e-06, | |
| "loss": 0.0848, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 30.01958762886598, | |
| "grad_norm": 0.6414709091186523, | |
| "learning_rate": 4.192439862542956e-06, | |
| "loss": 0.1163, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 30.020103092783504, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 1.7067806720733643, | |
| "eval_runtime": 13.9465, | |
| "eval_samples_per_second": 3.227, | |
| "eval_steps_per_second": 0.86, | |
| "step": 6045 | |
| }, | |
| { | |
| "epoch": 31.000515463917527, | |
| "grad_norm": 0.016645895317196846, | |
| "learning_rate": 4.180985108820161e-06, | |
| "loss": 0.4294, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 31.001546391752576, | |
| "grad_norm": 0.7525324821472168, | |
| "learning_rate": 4.169530355097366e-06, | |
| "loss": 0.4195, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 31.00257731958763, | |
| "grad_norm": 55.71934127807617, | |
| "learning_rate": 4.158075601374571e-06, | |
| "loss": 0.2149, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 31.00360824742268, | |
| "grad_norm": 0.006537964567542076, | |
| "learning_rate": 4.146620847651776e-06, | |
| "loss": 0.4038, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 31.004639175257733, | |
| "grad_norm": 2.2747881412506104, | |
| "learning_rate": 4.13516609392898e-06, | |
| "loss": 0.33, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 31.005670103092783, | |
| "grad_norm": 0.04077708348631859, | |
| "learning_rate": 4.123711340206186e-06, | |
| "loss": 0.0016, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 31.006701030927836, | |
| "grad_norm": 2.354952573776245, | |
| "learning_rate": 4.112256586483391e-06, | |
| "loss": 0.4812, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 31.007731958762886, | |
| "grad_norm": 137.84132385253906, | |
| "learning_rate": 4.100801832760596e-06, | |
| "loss": 0.2663, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 31.00876288659794, | |
| "grad_norm": 0.071143738925457, | |
| "learning_rate": 4.089347079037801e-06, | |
| "loss": 0.0608, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 31.00979381443299, | |
| "grad_norm": 0.9687093496322632, | |
| "learning_rate": 4.077892325315006e-06, | |
| "loss": 0.2852, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 31.010824742268042, | |
| "grad_norm": 0.061937566846609116, | |
| "learning_rate": 4.066437571592211e-06, | |
| "loss": 0.2822, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 31.011855670103092, | |
| "grad_norm": 51.97724914550781, | |
| "learning_rate": 4.054982817869416e-06, | |
| "loss": 0.4565, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 31.012886597938145, | |
| "grad_norm": 0.051208335906267166, | |
| "learning_rate": 4.043528064146621e-06, | |
| "loss": 0.0235, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 31.013917525773195, | |
| "grad_norm": 0.028698451817035675, | |
| "learning_rate": 4.032073310423826e-06, | |
| "loss": 0.0208, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 31.01494845360825, | |
| "grad_norm": 0.02642163261771202, | |
| "learning_rate": 4.020618556701032e-06, | |
| "loss": 0.0016, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 31.015979381443298, | |
| "grad_norm": 0.019867481663823128, | |
| "learning_rate": 4.009163802978236e-06, | |
| "loss": 0.3667, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 31.01701030927835, | |
| "grad_norm": 0.12207566946744919, | |
| "learning_rate": 3.997709049255441e-06, | |
| "loss": 0.0483, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 31.0180412371134, | |
| "grad_norm": 0.0024308476131409407, | |
| "learning_rate": 3.986254295532647e-06, | |
| "loss": 0.2048, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 31.019072164948454, | |
| "grad_norm": 0.3703158497810364, | |
| "learning_rate": 3.974799541809851e-06, | |
| "loss": 0.4534, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 31.020103092783504, | |
| "grad_norm": 0.07688494771718979, | |
| "learning_rate": 3.9633447880870564e-06, | |
| "loss": 0.0962, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 31.020103092783504, | |
| "eval_accuracy": 0.6888888888888889, | |
| "eval_loss": 1.8766827583312988, | |
| "eval_runtime": 12.7435, | |
| "eval_samples_per_second": 3.531, | |
| "eval_steps_per_second": 0.942, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 32.00103092783505, | |
| "grad_norm": 3.0947821140289307, | |
| "learning_rate": 3.951890034364262e-06, | |
| "loss": 0.003, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 32.00206185567011, | |
| "grad_norm": 0.021548155695199966, | |
| "learning_rate": 3.940435280641466e-06, | |
| "loss": 0.2464, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 32.00309278350515, | |
| "grad_norm": 0.010396094061434269, | |
| "learning_rate": 3.9289805269186715e-06, | |
| "loss": 0.1475, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 32.004123711340206, | |
| "grad_norm": 0.26135310530662537, | |
| "learning_rate": 3.917525773195877e-06, | |
| "loss": 0.0456, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 32.00515463917526, | |
| "grad_norm": 0.04791492596268654, | |
| "learning_rate": 3.906071019473082e-06, | |
| "loss": 0.0836, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 32.00618556701031, | |
| "grad_norm": 0.00865848921239376, | |
| "learning_rate": 3.8946162657502865e-06, | |
| "loss": 0.0589, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 32.00721649484536, | |
| "grad_norm": 0.04501219838857651, | |
| "learning_rate": 3.883161512027492e-06, | |
| "loss": 0.1143, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 32.00824742268041, | |
| "grad_norm": 0.049783822149038315, | |
| "learning_rate": 3.871706758304697e-06, | |
| "loss": 0.1852, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 32.009278350515466, | |
| "grad_norm": 0.031185345724225044, | |
| "learning_rate": 3.8602520045819016e-06, | |
| "loss": 0.3198, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 32.01030927835052, | |
| "grad_norm": 139.4656219482422, | |
| "learning_rate": 3.848797250859107e-06, | |
| "loss": 0.0602, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 32.011340206185565, | |
| "grad_norm": 155.99598693847656, | |
| "learning_rate": 3.837342497136312e-06, | |
| "loss": 0.133, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 32.01237113402062, | |
| "grad_norm": 0.06643953174352646, | |
| "learning_rate": 3.825887743413517e-06, | |
| "loss": 0.2172, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 32.01340206185567, | |
| "grad_norm": 149.49154663085938, | |
| "learning_rate": 3.814432989690722e-06, | |
| "loss": 0.2295, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 32.014432989690725, | |
| "grad_norm": 0.12383408844470978, | |
| "learning_rate": 3.8029782359679268e-06, | |
| "loss": 0.0736, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 32.01546391752577, | |
| "grad_norm": 46.82889938354492, | |
| "learning_rate": 3.791523482245132e-06, | |
| "loss": 0.2792, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 32.016494845360825, | |
| "grad_norm": 0.019881825894117355, | |
| "learning_rate": 3.780068728522337e-06, | |
| "loss": 0.1028, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 32.01752577319588, | |
| "grad_norm": 53.818206787109375, | |
| "learning_rate": 3.7686139747995422e-06, | |
| "loss": 0.5382, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 32.01855670103093, | |
| "grad_norm": 10.200157165527344, | |
| "learning_rate": 3.757159221076747e-06, | |
| "loss": 0.0028, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 32.01958762886598, | |
| "grad_norm": 0.9711626768112183, | |
| "learning_rate": 3.7457044673539524e-06, | |
| "loss": 0.2826, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 32.02010309278351, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 2.165703773498535, | |
| "eval_runtime": 14.5293, | |
| "eval_samples_per_second": 3.097, | |
| "eval_steps_per_second": 0.826, | |
| "step": 6435 | |
| }, | |
| { | |
| "epoch": 33.00051546391752, | |
| "grad_norm": 0.02483387477695942, | |
| "learning_rate": 3.7342497136311573e-06, | |
| "loss": 0.0014, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 33.001546391752576, | |
| "grad_norm": 65.55487060546875, | |
| "learning_rate": 3.7227949599083626e-06, | |
| "loss": 0.1579, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 33.00257731958763, | |
| "grad_norm": 0.09028110653162003, | |
| "learning_rate": 3.7113402061855674e-06, | |
| "loss": 0.2578, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 33.00360824742268, | |
| "grad_norm": 0.028494760394096375, | |
| "learning_rate": 3.6998854524627727e-06, | |
| "loss": 0.0175, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 33.00463917525773, | |
| "grad_norm": 0.015268037095665932, | |
| "learning_rate": 3.6884306987399776e-06, | |
| "loss": 0.1281, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 33.00567010309278, | |
| "grad_norm": 0.022972600534558296, | |
| "learning_rate": 3.6769759450171825e-06, | |
| "loss": 0.2622, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 33.006701030927836, | |
| "grad_norm": 0.022758277133107185, | |
| "learning_rate": 3.6655211912943874e-06, | |
| "loss": 0.0093, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 33.00773195876289, | |
| "grad_norm": 8.829773902893066, | |
| "learning_rate": 3.6540664375715922e-06, | |
| "loss": 0.597, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 33.008762886597935, | |
| "grad_norm": 0.006442221347242594, | |
| "learning_rate": 3.6426116838487975e-06, | |
| "loss": 0.1217, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 33.00979381443299, | |
| "grad_norm": 0.15540894865989685, | |
| "learning_rate": 3.6311569301260024e-06, | |
| "loss": 0.2546, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 33.01082474226804, | |
| "grad_norm": 1.448900580406189, | |
| "learning_rate": 3.6197021764032077e-06, | |
| "loss": 0.0054, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 33.011855670103095, | |
| "grad_norm": 0.1175546869635582, | |
| "learning_rate": 3.6082474226804126e-06, | |
| "loss": 0.0229, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 33.01288659793814, | |
| "grad_norm": 0.04827328771352768, | |
| "learning_rate": 3.596792668957618e-06, | |
| "loss": 0.2278, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 33.013917525773195, | |
| "grad_norm": 0.011951892636716366, | |
| "learning_rate": 3.5853379152348227e-06, | |
| "loss": 0.1539, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 33.01494845360825, | |
| "grad_norm": 0.05575776845216751, | |
| "learning_rate": 3.573883161512028e-06, | |
| "loss": 0.3902, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 33.0159793814433, | |
| "grad_norm": 43.80656051635742, | |
| "learning_rate": 3.562428407789233e-06, | |
| "loss": 0.011, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 33.01701030927835, | |
| "grad_norm": 1.3099844455718994, | |
| "learning_rate": 3.550973654066438e-06, | |
| "loss": 0.1933, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 33.0180412371134, | |
| "grad_norm": 0.03854267671704292, | |
| "learning_rate": 3.539518900343643e-06, | |
| "loss": 0.0416, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 33.019072164948454, | |
| "grad_norm": 0.01258633378893137, | |
| "learning_rate": 3.5280641466208484e-06, | |
| "loss": 0.2978, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 33.02010309278351, | |
| "grad_norm": 0.0009200758067891002, | |
| "learning_rate": 3.5166093928980532e-06, | |
| "loss": 0.1249, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 33.02010309278351, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 1.7385255098342896, | |
| "eval_runtime": 12.3632, | |
| "eval_samples_per_second": 3.64, | |
| "eval_steps_per_second": 0.971, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 34.00103092783505, | |
| "grad_norm": 0.1539161652326584, | |
| "learning_rate": 3.5051546391752577e-06, | |
| "loss": 0.1841, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 34.00206185567011, | |
| "grad_norm": 0.02649785578250885, | |
| "learning_rate": 3.493699885452463e-06, | |
| "loss": 0.2675, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 34.00309278350515, | |
| "grad_norm": 0.025071190670132637, | |
| "learning_rate": 3.482245131729668e-06, | |
| "loss": 0.0023, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 34.004123711340206, | |
| "grad_norm": 0.019204530864953995, | |
| "learning_rate": 3.470790378006873e-06, | |
| "loss": 0.4407, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 34.00515463917526, | |
| "grad_norm": 0.02421189844608307, | |
| "learning_rate": 3.459335624284078e-06, | |
| "loss": 0.0031, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 34.00618556701031, | |
| "grad_norm": 0.02116827294230461, | |
| "learning_rate": 3.4478808705612833e-06, | |
| "loss": 0.2766, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 34.00721649484536, | |
| "grad_norm": 0.0051218606531620026, | |
| "learning_rate": 3.436426116838488e-06, | |
| "loss": 0.6209, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 34.00824742268041, | |
| "grad_norm": 0.015079047530889511, | |
| "learning_rate": 3.4249713631156935e-06, | |
| "loss": 0.0015, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 34.009278350515466, | |
| "grad_norm": 1.001089096069336, | |
| "learning_rate": 3.4135166093928984e-06, | |
| "loss": 0.1645, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 34.01030927835052, | |
| "grad_norm": 0.0209248848259449, | |
| "learning_rate": 3.4020618556701037e-06, | |
| "loss": 0.0015, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 34.011340206185565, | |
| "grad_norm": 0.010426868684589863, | |
| "learning_rate": 3.3906071019473085e-06, | |
| "loss": 0.0104, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 34.01237113402062, | |
| "grad_norm": 0.06406555324792862, | |
| "learning_rate": 3.379152348224514e-06, | |
| "loss": 0.0415, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 34.01340206185567, | |
| "grad_norm": 222.54000854492188, | |
| "learning_rate": 3.3676975945017187e-06, | |
| "loss": 0.2297, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 34.014432989690725, | |
| "grad_norm": 0.2206907570362091, | |
| "learning_rate": 3.356242840778923e-06, | |
| "loss": 0.0033, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 34.01546391752577, | |
| "grad_norm": 0.0022467290982604027, | |
| "learning_rate": 3.3447880870561285e-06, | |
| "loss": 0.0908, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 34.016494845360825, | |
| "grad_norm": 281.9080505371094, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.2918, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 34.01752577319588, | |
| "grad_norm": 0.13616985082626343, | |
| "learning_rate": 3.3218785796105386e-06, | |
| "loss": 0.1153, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 34.01855670103093, | |
| "grad_norm": 1.4510610103607178, | |
| "learning_rate": 3.3104238258877435e-06, | |
| "loss": 0.0018, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 34.01958762886598, | |
| "grad_norm": 0.008665881119668484, | |
| "learning_rate": 3.298969072164949e-06, | |
| "loss": 0.2191, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 34.02010309278351, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 2.178891181945801, | |
| "eval_runtime": 12.7411, | |
| "eval_samples_per_second": 3.532, | |
| "eval_steps_per_second": 0.942, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 35.00051546391752, | |
| "grad_norm": 0.05293448269367218, | |
| "learning_rate": 3.2875143184421537e-06, | |
| "loss": 0.1271, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 35.001546391752576, | |
| "grad_norm": 0.0611007995903492, | |
| "learning_rate": 3.276059564719359e-06, | |
| "loss": 0.1393, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 35.00257731958763, | |
| "grad_norm": 0.024918990209698677, | |
| "learning_rate": 3.264604810996564e-06, | |
| "loss": 0.1459, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 35.00360824742268, | |
| "grad_norm": 0.28028416633605957, | |
| "learning_rate": 3.253150057273769e-06, | |
| "loss": 0.1449, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 35.00463917525773, | |
| "grad_norm": 0.0019702534191310406, | |
| "learning_rate": 3.241695303550974e-06, | |
| "loss": 0.0093, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 35.00567010309278, | |
| "grad_norm": 7.490848064422607, | |
| "learning_rate": 3.2302405498281793e-06, | |
| "loss": 0.122, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 35.006701030927836, | |
| "grad_norm": 0.013970672152936459, | |
| "learning_rate": 3.218785796105384e-06, | |
| "loss": 0.0232, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 35.00773195876289, | |
| "grad_norm": 0.031982772052288055, | |
| "learning_rate": 3.2073310423825895e-06, | |
| "loss": 0.086, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 35.008762886597935, | |
| "grad_norm": 0.010435467585921288, | |
| "learning_rate": 3.195876288659794e-06, | |
| "loss": 0.3166, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 35.00979381443299, | |
| "grad_norm": 86.65909576416016, | |
| "learning_rate": 3.1844215349369988e-06, | |
| "loss": 0.3312, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 35.01082474226804, | |
| "grad_norm": 122.04566955566406, | |
| "learning_rate": 3.172966781214204e-06, | |
| "loss": 0.2556, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 35.011855670103095, | |
| "grad_norm": 0.12888219952583313, | |
| "learning_rate": 3.161512027491409e-06, | |
| "loss": 0.3073, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 35.01288659793814, | |
| "grad_norm": 0.025842690840363503, | |
| "learning_rate": 3.1500572737686143e-06, | |
| "loss": 0.1798, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 35.013917525773195, | |
| "grad_norm": 0.011454744264483452, | |
| "learning_rate": 3.138602520045819e-06, | |
| "loss": 0.0072, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 35.01494845360825, | |
| "grad_norm": 167.35546875, | |
| "learning_rate": 3.1271477663230244e-06, | |
| "loss": 0.071, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 35.0159793814433, | |
| "grad_norm": 0.024680141359567642, | |
| "learning_rate": 3.1156930126002293e-06, | |
| "loss": 0.4104, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 35.01701030927835, | |
| "grad_norm": 0.008994595147669315, | |
| "learning_rate": 3.1042382588774346e-06, | |
| "loss": 0.1758, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 35.0180412371134, | |
| "grad_norm": 46.581077575683594, | |
| "learning_rate": 3.0927835051546395e-06, | |
| "loss": 0.1648, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 35.019072164948454, | |
| "grad_norm": 199.21475219726562, | |
| "learning_rate": 3.0813287514318448e-06, | |
| "loss": 0.264, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 35.02010309278351, | |
| "grad_norm": 0.001545836334116757, | |
| "learning_rate": 3.0698739977090496e-06, | |
| "loss": 0.0958, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 35.02010309278351, | |
| "eval_accuracy": 0.6444444444444445, | |
| "eval_loss": 2.472182035446167, | |
| "eval_runtime": 12.7933, | |
| "eval_samples_per_second": 3.517, | |
| "eval_steps_per_second": 0.938, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 36.00103092783505, | |
| "grad_norm": 0.045767877250909805, | |
| "learning_rate": 3.058419243986255e-06, | |
| "loss": 0.2266, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 36.00206185567011, | |
| "grad_norm": 0.014738555997610092, | |
| "learning_rate": 3.04696449026346e-06, | |
| "loss": 0.0054, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 36.00309278350515, | |
| "grad_norm": 0.00678382720798254, | |
| "learning_rate": 3.0355097365406643e-06, | |
| "loss": 0.3692, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 36.004123711340206, | |
| "grad_norm": 0.00942118652164936, | |
| "learning_rate": 3.0240549828178695e-06, | |
| "loss": 0.0014, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 36.00515463917526, | |
| "grad_norm": 0.024564573541283607, | |
| "learning_rate": 3.0126002290950744e-06, | |
| "loss": 0.1198, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 36.00618556701031, | |
| "grad_norm": 271.1005859375, | |
| "learning_rate": 3.0011454753722797e-06, | |
| "loss": 0.2814, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 36.00721649484536, | |
| "grad_norm": 0.043836403638124466, | |
| "learning_rate": 2.9896907216494846e-06, | |
| "loss": 0.0023, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 36.00824742268041, | |
| "grad_norm": 0.008552830666303635, | |
| "learning_rate": 2.97823596792669e-06, | |
| "loss": 0.0138, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 36.009278350515466, | |
| "grad_norm": 0.2735336124897003, | |
| "learning_rate": 2.9667812142038948e-06, | |
| "loss": 0.0027, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 36.01030927835052, | |
| "grad_norm": 0.0645265206694603, | |
| "learning_rate": 2.9553264604811e-06, | |
| "loss": 0.1183, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 36.011340206185565, | |
| "grad_norm": 0.035903312265872955, | |
| "learning_rate": 2.943871706758305e-06, | |
| "loss": 0.014, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 36.01237113402062, | |
| "grad_norm": 4.987162113189697, | |
| "learning_rate": 2.9324169530355102e-06, | |
| "loss": 0.1061, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 36.01340206185567, | |
| "grad_norm": 0.45441123843193054, | |
| "learning_rate": 2.920962199312715e-06, | |
| "loss": 0.0013, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 36.014432989690725, | |
| "grad_norm": 1.8816462755203247, | |
| "learning_rate": 2.9095074455899204e-06, | |
| "loss": 0.0026, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 36.01546391752577, | |
| "grad_norm": 0.005839156918227673, | |
| "learning_rate": 2.8980526918671253e-06, | |
| "loss": 0.0038, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 36.016494845360825, | |
| "grad_norm": 0.5239657163619995, | |
| "learning_rate": 2.8865979381443297e-06, | |
| "loss": 0.0005, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 36.01752577319588, | |
| "grad_norm": 60.700618743896484, | |
| "learning_rate": 2.875143184421535e-06, | |
| "loss": 0.1786, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 36.01855670103093, | |
| "grad_norm": 0.0037163153756409883, | |
| "learning_rate": 2.86368843069874e-06, | |
| "loss": 0.4697, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 36.01958762886598, | |
| "grad_norm": 0.007669605780392885, | |
| "learning_rate": 2.852233676975945e-06, | |
| "loss": 0.0006, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 36.02010309278351, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 1.917656421661377, | |
| "eval_runtime": 12.6565, | |
| "eval_samples_per_second": 3.555, | |
| "eval_steps_per_second": 0.948, | |
| "step": 7215 | |
| }, | |
| { | |
| "epoch": 37.00051546391752, | |
| "grad_norm": 0.019232606515288353, | |
| "learning_rate": 2.84077892325315e-06, | |
| "loss": 0.1498, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 37.001546391752576, | |
| "grad_norm": 0.006618044804781675, | |
| "learning_rate": 2.8293241695303553e-06, | |
| "loss": 0.0006, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 37.00257731958763, | |
| "grad_norm": 32.33464813232422, | |
| "learning_rate": 2.8178694158075602e-06, | |
| "loss": 0.1678, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 37.00360824742268, | |
| "grad_norm": 0.02818481996655464, | |
| "learning_rate": 2.8064146620847655e-06, | |
| "loss": 0.2838, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 37.00463917525773, | |
| "grad_norm": 136.29965209960938, | |
| "learning_rate": 2.7949599083619704e-06, | |
| "loss": 0.0919, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 37.00567010309278, | |
| "grad_norm": 0.004848357755690813, | |
| "learning_rate": 2.7835051546391757e-06, | |
| "loss": 0.1393, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 37.006701030927836, | |
| "grad_norm": 91.26074981689453, | |
| "learning_rate": 2.7720504009163806e-06, | |
| "loss": 0.293, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 37.00773195876289, | |
| "grad_norm": 0.008552628569304943, | |
| "learning_rate": 2.760595647193586e-06, | |
| "loss": 0.0011, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 37.008762886597935, | |
| "grad_norm": 0.011269760318100452, | |
| "learning_rate": 2.7491408934707907e-06, | |
| "loss": 0.0008, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 37.00979381443299, | |
| "grad_norm": 0.07644051313400269, | |
| "learning_rate": 2.737686139747996e-06, | |
| "loss": 0.2102, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 37.01082474226804, | |
| "grad_norm": 0.0022292486391961575, | |
| "learning_rate": 2.7262313860252005e-06, | |
| "loss": 0.1402, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 37.011855670103095, | |
| "grad_norm": 7.599360466003418, | |
| "learning_rate": 2.7147766323024053e-06, | |
| "loss": 0.0044, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 37.01288659793814, | |
| "grad_norm": 220.58099365234375, | |
| "learning_rate": 2.7033218785796106e-06, | |
| "loss": 0.4214, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 37.013917525773195, | |
| "grad_norm": 83.0787353515625, | |
| "learning_rate": 2.6918671248568155e-06, | |
| "loss": 0.0265, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 37.01494845360825, | |
| "grad_norm": 0.015676092356443405, | |
| "learning_rate": 2.680412371134021e-06, | |
| "loss": 0.0018, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 37.0159793814433, | |
| "grad_norm": 0.008032367564737797, | |
| "learning_rate": 2.6689576174112257e-06, | |
| "loss": 0.0151, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 37.01701030927835, | |
| "grad_norm": 0.36304762959480286, | |
| "learning_rate": 2.657502863688431e-06, | |
| "loss": 0.1789, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 37.0180412371134, | |
| "grad_norm": 0.010707261972129345, | |
| "learning_rate": 2.646048109965636e-06, | |
| "loss": 0.0672, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 37.019072164948454, | |
| "grad_norm": 110.2496566772461, | |
| "learning_rate": 2.634593356242841e-06, | |
| "loss": 0.664, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 37.02010309278351, | |
| "grad_norm": 0.13198626041412354, | |
| "learning_rate": 2.623138602520046e-06, | |
| "loss": 0.0036, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 37.02010309278351, | |
| "eval_accuracy": 0.6888888888888889, | |
| "eval_loss": 1.9590771198272705, | |
| "eval_runtime": 14.573, | |
| "eval_samples_per_second": 3.088, | |
| "eval_steps_per_second": 0.823, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 38.00103092783505, | |
| "grad_norm": 64.50994110107422, | |
| "learning_rate": 2.6116838487972513e-06, | |
| "loss": 0.3479, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 38.00206185567011, | |
| "grad_norm": 139.82247924804688, | |
| "learning_rate": 2.600229095074456e-06, | |
| "loss": 0.2922, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 38.00309278350515, | |
| "grad_norm": 0.011880365200340748, | |
| "learning_rate": 2.5887743413516615e-06, | |
| "loss": 0.0738, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 38.004123711340206, | |
| "grad_norm": 0.0019393068505451083, | |
| "learning_rate": 2.577319587628866e-06, | |
| "loss": 0.0053, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 38.00515463917526, | |
| "grad_norm": 0.007837596349418163, | |
| "learning_rate": 2.565864833906071e-06, | |
| "loss": 0.1809, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 38.00618556701031, | |
| "grad_norm": 0.051794491708278656, | |
| "learning_rate": 2.554410080183276e-06, | |
| "loss": 0.1372, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 38.00721649484536, | |
| "grad_norm": 0.11213437467813492, | |
| "learning_rate": 2.542955326460481e-06, | |
| "loss": 0.0021, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 38.00824742268041, | |
| "grad_norm": 0.008571717888116837, | |
| "learning_rate": 2.5315005727376863e-06, | |
| "loss": 0.2099, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 38.009278350515466, | |
| "grad_norm": 0.021522628143429756, | |
| "learning_rate": 2.520045819014891e-06, | |
| "loss": 0.214, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 38.01030927835052, | |
| "grad_norm": 0.24400968849658966, | |
| "learning_rate": 2.5085910652920964e-06, | |
| "loss": 0.0008, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 38.011340206185565, | |
| "grad_norm": 0.008205863647162914, | |
| "learning_rate": 2.4971363115693013e-06, | |
| "loss": 0.1451, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 38.01237113402062, | |
| "grad_norm": 0.0307331420481205, | |
| "learning_rate": 2.4856815578465066e-06, | |
| "loss": 0.1482, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 38.01340206185567, | |
| "grad_norm": 0.06798528879880905, | |
| "learning_rate": 2.4742268041237115e-06, | |
| "loss": 0.0751, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 38.014432989690725, | |
| "grad_norm": 0.0034336706157773733, | |
| "learning_rate": 2.4627720504009168e-06, | |
| "loss": 0.0006, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 38.01546391752577, | |
| "grad_norm": 0.14595551788806915, | |
| "learning_rate": 2.4513172966781217e-06, | |
| "loss": 0.0006, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 38.016494845360825, | |
| "grad_norm": 0.00959043949842453, | |
| "learning_rate": 2.4398625429553265e-06, | |
| "loss": 0.0011, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 38.01752577319588, | |
| "grad_norm": 77.62090301513672, | |
| "learning_rate": 2.428407789232532e-06, | |
| "loss": 0.5068, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 38.01855670103093, | |
| "grad_norm": 0.020476188510656357, | |
| "learning_rate": 2.4169530355097367e-06, | |
| "loss": 0.1332, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 38.01958762886598, | |
| "grad_norm": 0.06416749209165573, | |
| "learning_rate": 2.405498281786942e-06, | |
| "loss": 0.0009, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 38.02010309278351, | |
| "eval_accuracy": 0.6222222222222222, | |
| "eval_loss": 2.399278402328491, | |
| "eval_runtime": 12.9659, | |
| "eval_samples_per_second": 3.471, | |
| "eval_steps_per_second": 0.926, | |
| "step": 7605 | |
| }, | |
| { | |
| "epoch": 39.00051546391752, | |
| "grad_norm": 5.2809295654296875, | |
| "learning_rate": 2.394043528064147e-06, | |
| "loss": 0.0652, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 39.001546391752576, | |
| "grad_norm": 0.008623295463621616, | |
| "learning_rate": 2.3825887743413517e-06, | |
| "loss": 0.1178, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 39.00257731958763, | |
| "grad_norm": 0.03899654373526573, | |
| "learning_rate": 2.3711340206185566e-06, | |
| "loss": 0.0365, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 39.00360824742268, | |
| "grad_norm": 0.01230724435299635, | |
| "learning_rate": 2.359679266895762e-06, | |
| "loss": 0.1324, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 39.00463917525773, | |
| "grad_norm": 0.5138551592826843, | |
| "learning_rate": 2.3482245131729668e-06, | |
| "loss": 0.0991, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 39.00567010309278, | |
| "grad_norm": 0.004986596293747425, | |
| "learning_rate": 2.336769759450172e-06, | |
| "loss": 0.002, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 39.006701030927836, | |
| "grad_norm": 0.04998844861984253, | |
| "learning_rate": 2.325315005727377e-06, | |
| "loss": 0.0007, | |
| "step": 7670 | |
| }, | |
| { | |
| "epoch": 39.00773195876289, | |
| "grad_norm": 0.0053374143317341805, | |
| "learning_rate": 2.3138602520045822e-06, | |
| "loss": 0.0531, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 39.008762886597935, | |
| "grad_norm": 326.64031982421875, | |
| "learning_rate": 2.302405498281787e-06, | |
| "loss": 0.307, | |
| "step": 7690 | |
| }, | |
| { | |
| "epoch": 39.00979381443299, | |
| "grad_norm": 0.18461164832115173, | |
| "learning_rate": 2.290950744558992e-06, | |
| "loss": 0.0038, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 39.01082474226804, | |
| "grad_norm": 0.01692032255232334, | |
| "learning_rate": 2.2794959908361973e-06, | |
| "loss": 0.001, | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 39.011855670103095, | |
| "grad_norm": 0.012526489794254303, | |
| "learning_rate": 2.268041237113402e-06, | |
| "loss": 0.3061, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 39.01288659793814, | |
| "grad_norm": 0.009198295883834362, | |
| "learning_rate": 2.2565864833906075e-06, | |
| "loss": 0.1932, | |
| "step": 7730 | |
| }, | |
| { | |
| "epoch": 39.013917525773195, | |
| "grad_norm": 0.1381886750459671, | |
| "learning_rate": 2.2451317296678123e-06, | |
| "loss": 0.0878, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 39.01494845360825, | |
| "grad_norm": 0.03150659054517746, | |
| "learning_rate": 2.2336769759450176e-06, | |
| "loss": 0.1654, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 39.0159793814433, | |
| "grad_norm": 0.011908908374607563, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0004, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 39.01701030927835, | |
| "grad_norm": 16.51814842224121, | |
| "learning_rate": 2.2107674684994274e-06, | |
| "loss": 0.0024, | |
| "step": 7770 | |
| }, | |
| { | |
| "epoch": 39.0180412371134, | |
| "grad_norm": 0.05224655196070671, | |
| "learning_rate": 2.1993127147766322e-06, | |
| "loss": 0.0003, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 39.019072164948454, | |
| "grad_norm": 0.10637877881526947, | |
| "learning_rate": 2.1878579610538375e-06, | |
| "loss": 0.0116, | |
| "step": 7790 | |
| }, | |
| { | |
| "epoch": 39.02010309278351, | |
| "grad_norm": 0.08694580942392349, | |
| "learning_rate": 2.1764032073310424e-06, | |
| "loss": 0.0005, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 39.02010309278351, | |
| "eval_accuracy": 0.7777777777777778, | |
| "eval_loss": 1.7377581596374512, | |
| "eval_runtime": 12.9146, | |
| "eval_samples_per_second": 3.484, | |
| "eval_steps_per_second": 0.929, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 40.00103092783505, | |
| "grad_norm": 0.0032964874990284443, | |
| "learning_rate": 2.1649484536082477e-06, | |
| "loss": 0.057, | |
| "step": 7810 | |
| }, | |
| { | |
| "epoch": 40.00206185567011, | |
| "grad_norm": 0.01317548006772995, | |
| "learning_rate": 2.1534936998854526e-06, | |
| "loss": 0.0008, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 40.00309278350515, | |
| "grad_norm": 0.017642924562096596, | |
| "learning_rate": 2.1420389461626575e-06, | |
| "loss": 0.0005, | |
| "step": 7830 | |
| }, | |
| { | |
| "epoch": 40.004123711340206, | |
| "grad_norm": 0.012752017937600613, | |
| "learning_rate": 2.1305841924398628e-06, | |
| "loss": 0.0004, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 40.00515463917526, | |
| "grad_norm": 0.06961321830749512, | |
| "learning_rate": 2.1191294387170676e-06, | |
| "loss": 0.0003, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 40.00618556701031, | |
| "grad_norm": 0.09094371646642685, | |
| "learning_rate": 2.107674684994273e-06, | |
| "loss": 0.1611, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 40.00721649484536, | |
| "grad_norm": 0.008556496351957321, | |
| "learning_rate": 2.096219931271478e-06, | |
| "loss": 0.0007, | |
| "step": 7870 | |
| }, | |
| { | |
| "epoch": 40.00824742268041, | |
| "grad_norm": 309.84698486328125, | |
| "learning_rate": 2.084765177548683e-06, | |
| "loss": 0.2056, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 40.009278350515466, | |
| "grad_norm": 0.03041210025548935, | |
| "learning_rate": 2.073310423825888e-06, | |
| "loss": 0.0025, | |
| "step": 7890 | |
| }, | |
| { | |
| "epoch": 40.01030927835052, | |
| "grad_norm": 82.68245697021484, | |
| "learning_rate": 2.061855670103093e-06, | |
| "loss": 0.026, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 40.011340206185565, | |
| "grad_norm": 0.02870158664882183, | |
| "learning_rate": 2.050400916380298e-06, | |
| "loss": 0.0072, | |
| "step": 7910 | |
| }, | |
| { | |
| "epoch": 40.01237113402062, | |
| "grad_norm": 0.02945534512400627, | |
| "learning_rate": 2.038946162657503e-06, | |
| "loss": 0.1894, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 40.01340206185567, | |
| "grad_norm": 0.007657837588340044, | |
| "learning_rate": 2.027491408934708e-06, | |
| "loss": 0.0024, | |
| "step": 7930 | |
| }, | |
| { | |
| "epoch": 40.014432989690725, | |
| "grad_norm": 0.05909576267004013, | |
| "learning_rate": 2.016036655211913e-06, | |
| "loss": 0.1906, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 40.01546391752577, | |
| "grad_norm": 0.5790134072303772, | |
| "learning_rate": 2.004581901489118e-06, | |
| "loss": 0.0005, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 40.016494845360825, | |
| "grad_norm": 0.0051000709645450115, | |
| "learning_rate": 1.9931271477663233e-06, | |
| "loss": 0.1721, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 40.01752577319588, | |
| "grad_norm": 0.011354904621839523, | |
| "learning_rate": 1.9816723940435282e-06, | |
| "loss": 0.0046, | |
| "step": 7970 | |
| }, | |
| { | |
| "epoch": 40.01855670103093, | |
| "grad_norm": 327.5389099121094, | |
| "learning_rate": 1.970217640320733e-06, | |
| "loss": 0.4001, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 40.01958762886598, | |
| "grad_norm": 0.006849356461316347, | |
| "learning_rate": 1.9587628865979384e-06, | |
| "loss": 0.0014, | |
| "step": 7990 | |
| }, | |
| { | |
| "epoch": 40.02010309278351, | |
| "eval_accuracy": 0.6888888888888889, | |
| "eval_loss": 2.4453516006469727, | |
| "eval_runtime": 13.0464, | |
| "eval_samples_per_second": 3.449, | |
| "eval_steps_per_second": 0.92, | |
| "step": 7995 | |
| }, | |
| { | |
| "epoch": 41.00051546391752, | |
| "grad_norm": 0.013001556508243084, | |
| "learning_rate": 1.9473081328751433e-06, | |
| "loss": 0.2103, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 41.001546391752576, | |
| "grad_norm": 107.24781799316406, | |
| "learning_rate": 1.9358533791523486e-06, | |
| "loss": 0.3106, | |
| "step": 8010 | |
| }, | |
| { | |
| "epoch": 41.00257731958763, | |
| "grad_norm": 0.008363613858819008, | |
| "learning_rate": 1.9243986254295534e-06, | |
| "loss": 0.028, | |
| "step": 8020 | |
| }, | |
| { | |
| "epoch": 41.00360824742268, | |
| "grad_norm": 0.04029637202620506, | |
| "learning_rate": 1.9129438717067583e-06, | |
| "loss": 0.0264, | |
| "step": 8030 | |
| }, | |
| { | |
| "epoch": 41.00463917525773, | |
| "grad_norm": 0.06103358045220375, | |
| "learning_rate": 1.9014891179839634e-06, | |
| "loss": 0.0334, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 41.00567010309278, | |
| "grad_norm": 0.00733905378729105, | |
| "learning_rate": 1.8900343642611685e-06, | |
| "loss": 0.0014, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 41.006701030927836, | |
| "grad_norm": 338.4005432128906, | |
| "learning_rate": 1.8785796105383736e-06, | |
| "loss": 0.0489, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 41.00773195876289, | |
| "grad_norm": 0.006053756456822157, | |
| "learning_rate": 1.8671248568155786e-06, | |
| "loss": 0.1368, | |
| "step": 8070 | |
| }, | |
| { | |
| "epoch": 41.008762886597935, | |
| "grad_norm": 0.0014053594786673784, | |
| "learning_rate": 1.8556701030927837e-06, | |
| "loss": 0.1025, | |
| "step": 8080 | |
| }, | |
| { | |
| "epoch": 41.00979381443299, | |
| "grad_norm": 0.0035459971986711025, | |
| "learning_rate": 1.8442153493699888e-06, | |
| "loss": 0.0006, | |
| "step": 8090 | |
| }, | |
| { | |
| "epoch": 41.01082474226804, | |
| "grad_norm": 384.0328369140625, | |
| "learning_rate": 1.8327605956471937e-06, | |
| "loss": 0.3021, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 41.011855670103095, | |
| "grad_norm": 0.5014728903770447, | |
| "learning_rate": 1.8213058419243988e-06, | |
| "loss": 0.1137, | |
| "step": 8110 | |
| }, | |
| { | |
| "epoch": 41.01288659793814, | |
| "grad_norm": 0.0056852176785469055, | |
| "learning_rate": 1.8098510882016038e-06, | |
| "loss": 0.0004, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 41.013917525773195, | |
| "grad_norm": 0.013269063085317612, | |
| "learning_rate": 1.798396334478809e-06, | |
| "loss": 0.0002, | |
| "step": 8130 | |
| }, | |
| { | |
| "epoch": 41.01494845360825, | |
| "grad_norm": 56.177669525146484, | |
| "learning_rate": 1.786941580756014e-06, | |
| "loss": 0.1576, | |
| "step": 8140 | |
| }, | |
| { | |
| "epoch": 41.0159793814433, | |
| "grad_norm": 0.43153733015060425, | |
| "learning_rate": 1.775486827033219e-06, | |
| "loss": 0.1699, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 41.01701030927835, | |
| "grad_norm": 0.010211652144789696, | |
| "learning_rate": 1.7640320733104242e-06, | |
| "loss": 0.0006, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 41.0180412371134, | |
| "grad_norm": 0.004664810374379158, | |
| "learning_rate": 1.7525773195876288e-06, | |
| "loss": 0.087, | |
| "step": 8170 | |
| }, | |
| { | |
| "epoch": 41.019072164948454, | |
| "grad_norm": 124.0995864868164, | |
| "learning_rate": 1.741122565864834e-06, | |
| "loss": 0.1361, | |
| "step": 8180 | |
| }, | |
| { | |
| "epoch": 41.02010309278351, | |
| "grad_norm": 0.04815516620874405, | |
| "learning_rate": 1.729667812142039e-06, | |
| "loss": 0.1203, | |
| "step": 8190 | |
| }, | |
| { | |
| "epoch": 41.02010309278351, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 2.113752841949463, | |
| "eval_runtime": 12.9032, | |
| "eval_samples_per_second": 3.488, | |
| "eval_steps_per_second": 0.93, | |
| "step": 8190 | |
| }, | |
| { | |
| "epoch": 42.00103092783505, | |
| "grad_norm": 0.0071434988640248775, | |
| "learning_rate": 1.718213058419244e-06, | |
| "loss": 0.0003, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 42.00206185567011, | |
| "grad_norm": 0.020918577909469604, | |
| "learning_rate": 1.7067583046964492e-06, | |
| "loss": 0.0011, | |
| "step": 8210 | |
| }, | |
| { | |
| "epoch": 42.00309278350515, | |
| "grad_norm": 0.03154708817601204, | |
| "learning_rate": 1.6953035509736543e-06, | |
| "loss": 0.0815, | |
| "step": 8220 | |
| }, | |
| { | |
| "epoch": 42.004123711340206, | |
| "grad_norm": 0.013127969577908516, | |
| "learning_rate": 1.6838487972508594e-06, | |
| "loss": 0.0354, | |
| "step": 8230 | |
| }, | |
| { | |
| "epoch": 42.00515463917526, | |
| "grad_norm": 0.11450503766536713, | |
| "learning_rate": 1.6723940435280642e-06, | |
| "loss": 0.6204, | |
| "step": 8240 | |
| }, | |
| { | |
| "epoch": 42.00618556701031, | |
| "grad_norm": 0.007757487706840038, | |
| "learning_rate": 1.6609392898052693e-06, | |
| "loss": 0.1229, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 42.00721649484536, | |
| "grad_norm": 0.015443161129951477, | |
| "learning_rate": 1.6494845360824744e-06, | |
| "loss": 0.0002, | |
| "step": 8260 | |
| }, | |
| { | |
| "epoch": 42.00824742268041, | |
| "grad_norm": 0.002430438296869397, | |
| "learning_rate": 1.6380297823596795e-06, | |
| "loss": 0.0002, | |
| "step": 8270 | |
| }, | |
| { | |
| "epoch": 42.009278350515466, | |
| "grad_norm": 0.10335277765989304, | |
| "learning_rate": 1.6265750286368846e-06, | |
| "loss": 0.0185, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 42.01030927835052, | |
| "grad_norm": 0.002046901499852538, | |
| "learning_rate": 1.6151202749140896e-06, | |
| "loss": 0.0508, | |
| "step": 8290 | |
| }, | |
| { | |
| "epoch": 42.011340206185565, | |
| "grad_norm": 0.005286916624754667, | |
| "learning_rate": 1.6036655211912947e-06, | |
| "loss": 0.0702, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 42.01237113402062, | |
| "grad_norm": 0.007195398211479187, | |
| "learning_rate": 1.5922107674684994e-06, | |
| "loss": 0.3414, | |
| "step": 8310 | |
| }, | |
| { | |
| "epoch": 42.01340206185567, | |
| "grad_norm": 0.002511281054466963, | |
| "learning_rate": 1.5807560137457045e-06, | |
| "loss": 0.2097, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 42.014432989690725, | |
| "grad_norm": 201.3845977783203, | |
| "learning_rate": 1.5693012600229096e-06, | |
| "loss": 0.2369, | |
| "step": 8330 | |
| }, | |
| { | |
| "epoch": 42.01546391752577, | |
| "grad_norm": 0.04441210627555847, | |
| "learning_rate": 1.5578465063001146e-06, | |
| "loss": 0.0554, | |
| "step": 8340 | |
| }, | |
| { | |
| "epoch": 42.016494845360825, | |
| "grad_norm": 12.391925811767578, | |
| "learning_rate": 1.5463917525773197e-06, | |
| "loss": 0.0019, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 42.01752577319588, | |
| "grad_norm": 0.1355491429567337, | |
| "learning_rate": 1.5349369988545248e-06, | |
| "loss": 0.0007, | |
| "step": 8360 | |
| }, | |
| { | |
| "epoch": 42.01855670103093, | |
| "grad_norm": 0.0035951670724898577, | |
| "learning_rate": 1.52348224513173e-06, | |
| "loss": 0.1423, | |
| "step": 8370 | |
| }, | |
| { | |
| "epoch": 42.01958762886598, | |
| "grad_norm": 0.007604878395795822, | |
| "learning_rate": 1.5120274914089348e-06, | |
| "loss": 0.0138, | |
| "step": 8380 | |
| }, | |
| { | |
| "epoch": 42.02010309278351, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 2.1768667697906494, | |
| "eval_runtime": 12.9363, | |
| "eval_samples_per_second": 3.479, | |
| "eval_steps_per_second": 0.928, | |
| "step": 8385 | |
| }, | |
| { | |
| "epoch": 43.00051546391752, | |
| "grad_norm": 0.026959659531712532, | |
| "learning_rate": 1.5005727376861399e-06, | |
| "loss": 0.0004, | |
| "step": 8390 | |
| }, | |
| { | |
| "epoch": 43.001546391752576, | |
| "grad_norm": 0.09655001759529114, | |
| "learning_rate": 1.489117983963345e-06, | |
| "loss": 0.2277, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 43.00257731958763, | |
| "grad_norm": 1.289469838142395, | |
| "learning_rate": 1.47766323024055e-06, | |
| "loss": 0.1235, | |
| "step": 8410 | |
| }, | |
| { | |
| "epoch": 43.00360824742268, | |
| "grad_norm": 0.0012255634646862745, | |
| "learning_rate": 1.4662084765177551e-06, | |
| "loss": 0.0079, | |
| "step": 8420 | |
| }, | |
| { | |
| "epoch": 43.00463917525773, | |
| "grad_norm": 0.0007930409628897905, | |
| "learning_rate": 1.4547537227949602e-06, | |
| "loss": 0.0002, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 43.00567010309278, | |
| "grad_norm": 0.024416925385594368, | |
| "learning_rate": 1.4432989690721649e-06, | |
| "loss": 0.3317, | |
| "step": 8440 | |
| }, | |
| { | |
| "epoch": 43.006701030927836, | |
| "grad_norm": 0.0007845965446904302, | |
| "learning_rate": 1.43184421534937e-06, | |
| "loss": 0.0002, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 43.00773195876289, | |
| "grad_norm": 0.001673469552770257, | |
| "learning_rate": 1.420389461626575e-06, | |
| "loss": 0.0028, | |
| "step": 8460 | |
| }, | |
| { | |
| "epoch": 43.008762886597935, | |
| "grad_norm": 0.00685878423973918, | |
| "learning_rate": 1.4089347079037801e-06, | |
| "loss": 0.0008, | |
| "step": 8470 | |
| }, | |
| { | |
| "epoch": 43.00979381443299, | |
| "grad_norm": 0.0035130823962390423, | |
| "learning_rate": 1.3974799541809852e-06, | |
| "loss": 0.0007, | |
| "step": 8480 | |
| }, | |
| { | |
| "epoch": 43.01082474226804, | |
| "grad_norm": 198.6305694580078, | |
| "learning_rate": 1.3860252004581903e-06, | |
| "loss": 0.1222, | |
| "step": 8490 | |
| }, | |
| { | |
| "epoch": 43.011855670103095, | |
| "grad_norm": 0.0031229022424668074, | |
| "learning_rate": 1.3745704467353954e-06, | |
| "loss": 0.0002, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 43.01288659793814, | |
| "grad_norm": 0.004219813738018274, | |
| "learning_rate": 1.3631156930126002e-06, | |
| "loss": 0.0001, | |
| "step": 8510 | |
| }, | |
| { | |
| "epoch": 43.013917525773195, | |
| "grad_norm": 0.0015190464910119772, | |
| "learning_rate": 1.3516609392898053e-06, | |
| "loss": 0.014, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 43.01494845360825, | |
| "grad_norm": 0.003646484576165676, | |
| "learning_rate": 1.3402061855670104e-06, | |
| "loss": 0.1198, | |
| "step": 8530 | |
| }, | |
| { | |
| "epoch": 43.0159793814433, | |
| "grad_norm": 0.0039468565955758095, | |
| "learning_rate": 1.3287514318442155e-06, | |
| "loss": 0.0002, | |
| "step": 8540 | |
| }, | |
| { | |
| "epoch": 43.01701030927835, | |
| "grad_norm": 0.0018218251643702388, | |
| "learning_rate": 1.3172966781214206e-06, | |
| "loss": 0.0013, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 43.0180412371134, | |
| "grad_norm": 104.12612915039062, | |
| "learning_rate": 1.3058419243986257e-06, | |
| "loss": 0.2044, | |
| "step": 8560 | |
| }, | |
| { | |
| "epoch": 43.019072164948454, | |
| "grad_norm": 0.0026801032945513725, | |
| "learning_rate": 1.2943871706758307e-06, | |
| "loss": 0.0003, | |
| "step": 8570 | |
| }, | |
| { | |
| "epoch": 43.02010309278351, | |
| "grad_norm": 0.0005498857935890555, | |
| "learning_rate": 1.2829324169530354e-06, | |
| "loss": 0.3569, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 43.02010309278351, | |
| "eval_accuracy": 0.6222222222222222, | |
| "eval_loss": 2.6946287155151367, | |
| "eval_runtime": 13.5653, | |
| "eval_samples_per_second": 3.317, | |
| "eval_steps_per_second": 0.885, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 44.00103092783505, | |
| "grad_norm": 0.0034866807982325554, | |
| "learning_rate": 1.2714776632302405e-06, | |
| "loss": 0.0771, | |
| "step": 8590 | |
| }, | |
| { | |
| "epoch": 44.00206185567011, | |
| "grad_norm": 0.0050008767284452915, | |
| "learning_rate": 1.2600229095074456e-06, | |
| "loss": 0.1379, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 44.00309278350515, | |
| "grad_norm": 0.003171220887452364, | |
| "learning_rate": 1.2485681557846507e-06, | |
| "loss": 0.0002, | |
| "step": 8610 | |
| }, | |
| { | |
| "epoch": 44.004123711340206, | |
| "grad_norm": 0.00555449491366744, | |
| "learning_rate": 1.2371134020618557e-06, | |
| "loss": 0.2303, | |
| "step": 8620 | |
| }, | |
| { | |
| "epoch": 44.00515463917526, | |
| "grad_norm": 0.00758061558008194, | |
| "learning_rate": 1.2256586483390608e-06, | |
| "loss": 0.0019, | |
| "step": 8630 | |
| }, | |
| { | |
| "epoch": 44.00618556701031, | |
| "grad_norm": 35.78764724731445, | |
| "learning_rate": 1.214203894616266e-06, | |
| "loss": 0.2258, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 44.00721649484536, | |
| "grad_norm": 0.002628314308822155, | |
| "learning_rate": 1.202749140893471e-06, | |
| "loss": 0.1071, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 44.00824742268041, | |
| "grad_norm": 0.0065179308876395226, | |
| "learning_rate": 1.1912943871706759e-06, | |
| "loss": 0.0002, | |
| "step": 8660 | |
| }, | |
| { | |
| "epoch": 44.009278350515466, | |
| "grad_norm": 0.00975864939391613, | |
| "learning_rate": 1.179839633447881e-06, | |
| "loss": 0.0002, | |
| "step": 8670 | |
| }, | |
| { | |
| "epoch": 44.01030927835052, | |
| "grad_norm": 0.008498983457684517, | |
| "learning_rate": 1.168384879725086e-06, | |
| "loss": 0.0009, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 44.011340206185565, | |
| "grad_norm": 0.0022127425763756037, | |
| "learning_rate": 1.1569301260022911e-06, | |
| "loss": 0.0182, | |
| "step": 8690 | |
| }, | |
| { | |
| "epoch": 44.01237113402062, | |
| "grad_norm": 0.015520663000643253, | |
| "learning_rate": 1.145475372279496e-06, | |
| "loss": 0.1718, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 44.01340206185567, | |
| "grad_norm": 0.00714871846139431, | |
| "learning_rate": 1.134020618556701e-06, | |
| "loss": 0.0001, | |
| "step": 8710 | |
| }, | |
| { | |
| "epoch": 44.014432989690725, | |
| "grad_norm": 98.3997573852539, | |
| "learning_rate": 1.1225658648339062e-06, | |
| "loss": 0.3241, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 44.01546391752577, | |
| "grad_norm": 0.004079570062458515, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0002, | |
| "step": 8730 | |
| }, | |
| { | |
| "epoch": 44.016494845360825, | |
| "grad_norm": 0.002131837885826826, | |
| "learning_rate": 1.0996563573883161e-06, | |
| "loss": 0.0002, | |
| "step": 8740 | |
| }, | |
| { | |
| "epoch": 44.01752577319588, | |
| "grad_norm": 0.027215346693992615, | |
| "learning_rate": 1.0882016036655212e-06, | |
| "loss": 0.388, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 44.01855670103093, | |
| "grad_norm": 0.025429489091038704, | |
| "learning_rate": 1.0767468499427263e-06, | |
| "loss": 0.1207, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 44.01958762886598, | |
| "grad_norm": 0.5874564051628113, | |
| "learning_rate": 1.0652920962199314e-06, | |
| "loss": 0.0002, | |
| "step": 8770 | |
| }, | |
| { | |
| "epoch": 44.02010309278351, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 2.156606435775757, | |
| "eval_runtime": 13.2361, | |
| "eval_samples_per_second": 3.4, | |
| "eval_steps_per_second": 0.907, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 45.00051546391752, | |
| "grad_norm": 0.01159296091645956, | |
| "learning_rate": 1.0538373424971365e-06, | |
| "loss": 0.0002, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 45.001546391752576, | |
| "grad_norm": 0.002699700416997075, | |
| "learning_rate": 1.0423825887743415e-06, | |
| "loss": 0.1022, | |
| "step": 8790 | |
| }, | |
| { | |
| "epoch": 45.00257731958763, | |
| "grad_norm": 0.002690671943128109, | |
| "learning_rate": 1.0309278350515464e-06, | |
| "loss": 0.0003, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 45.00360824742268, | |
| "grad_norm": 0.011441020295023918, | |
| "learning_rate": 1.0194730813287515e-06, | |
| "loss": 0.0027, | |
| "step": 8810 | |
| }, | |
| { | |
| "epoch": 45.00463917525773, | |
| "grad_norm": 0.3645267188549042, | |
| "learning_rate": 1.0080183276059566e-06, | |
| "loss": 0.0002, | |
| "step": 8820 | |
| }, | |
| { | |
| "epoch": 45.00567010309278, | |
| "grad_norm": 0.014001118950545788, | |
| "learning_rate": 9.965635738831617e-07, | |
| "loss": 0.0002, | |
| "step": 8830 | |
| }, | |
| { | |
| "epoch": 45.006701030927836, | |
| "grad_norm": 0.0017673630500212312, | |
| "learning_rate": 9.851088201603665e-07, | |
| "loss": 0.0006, | |
| "step": 8840 | |
| }, | |
| { | |
| "epoch": 45.00773195876289, | |
| "grad_norm": 44.564510345458984, | |
| "learning_rate": 9.736540664375716e-07, | |
| "loss": 0.4445, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 45.008762886597935, | |
| "grad_norm": 0.09444551169872284, | |
| "learning_rate": 9.621993127147767e-07, | |
| "loss": 0.0002, | |
| "step": 8860 | |
| }, | |
| { | |
| "epoch": 45.00979381443299, | |
| "grad_norm": 0.001571907545439899, | |
| "learning_rate": 9.507445589919817e-07, | |
| "loss": 0.2272, | |
| "step": 8870 | |
| }, | |
| { | |
| "epoch": 45.01082474226804, | |
| "grad_norm": 117.25080108642578, | |
| "learning_rate": 9.392898052691868e-07, | |
| "loss": 0.0253, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 45.011855670103095, | |
| "grad_norm": 0.02719367854297161, | |
| "learning_rate": 9.278350515463919e-07, | |
| "loss": 0.2072, | |
| "step": 8890 | |
| }, | |
| { | |
| "epoch": 45.01288659793814, | |
| "grad_norm": 0.011369436047971249, | |
| "learning_rate": 9.163802978235968e-07, | |
| "loss": 0.0002, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 45.013917525773195, | |
| "grad_norm": 0.009269513189792633, | |
| "learning_rate": 9.049255441008019e-07, | |
| "loss": 0.0184, | |
| "step": 8910 | |
| }, | |
| { | |
| "epoch": 45.01494845360825, | |
| "grad_norm": 0.003293866291642189, | |
| "learning_rate": 8.93470790378007e-07, | |
| "loss": 0.0084, | |
| "step": 8920 | |
| }, | |
| { | |
| "epoch": 45.0159793814433, | |
| "grad_norm": 0.012929639779031277, | |
| "learning_rate": 8.820160366552121e-07, | |
| "loss": 0.0005, | |
| "step": 8930 | |
| }, | |
| { | |
| "epoch": 45.01701030927835, | |
| "grad_norm": 0.012119736522436142, | |
| "learning_rate": 8.70561282932417e-07, | |
| "loss": 0.2072, | |
| "step": 8940 | |
| }, | |
| { | |
| "epoch": 45.0180412371134, | |
| "grad_norm": 0.1019548624753952, | |
| "learning_rate": 8.59106529209622e-07, | |
| "loss": 0.0002, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 45.019072164948454, | |
| "grad_norm": 0.005744592752307653, | |
| "learning_rate": 8.476517754868271e-07, | |
| "loss": 0.1167, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 45.02010309278351, | |
| "grad_norm": 0.08731329441070557, | |
| "learning_rate": 8.361970217640321e-07, | |
| "loss": 0.0924, | |
| "step": 8970 | |
| }, | |
| { | |
| "epoch": 45.02010309278351, | |
| "eval_accuracy": 0.6666666666666666, | |
| "eval_loss": 2.4635937213897705, | |
| "eval_runtime": 12.8852, | |
| "eval_samples_per_second": 3.492, | |
| "eval_steps_per_second": 0.931, | |
| "step": 8970 | |
| }, | |
| { | |
| "epoch": 46.00103092783505, | |
| "grad_norm": 0.006049647461622953, | |
| "learning_rate": 8.247422680412372e-07, | |
| "loss": 0.3291, | |
| "step": 8980 | |
| }, | |
| { | |
| "epoch": 46.00206185567011, | |
| "grad_norm": 206.76785278320312, | |
| "learning_rate": 8.132875143184423e-07, | |
| "loss": 0.1158, | |
| "step": 8990 | |
| }, | |
| { | |
| "epoch": 46.00309278350515, | |
| "grad_norm": 0.002335500903427601, | |
| "learning_rate": 8.018327605956474e-07, | |
| "loss": 0.0002, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 46.004123711340206, | |
| "grad_norm": 0.02907603606581688, | |
| "learning_rate": 7.903780068728522e-07, | |
| "loss": 0.0002, | |
| "step": 9010 | |
| }, | |
| { | |
| "epoch": 46.00515463917526, | |
| "grad_norm": 0.004002581350505352, | |
| "learning_rate": 7.789232531500573e-07, | |
| "loss": 0.3073, | |
| "step": 9020 | |
| }, | |
| { | |
| "epoch": 46.00618556701031, | |
| "grad_norm": 0.005749888252466917, | |
| "learning_rate": 7.674684994272624e-07, | |
| "loss": 0.3535, | |
| "step": 9030 | |
| }, | |
| { | |
| "epoch": 46.00721649484536, | |
| "grad_norm": 0.008823657408356667, | |
| "learning_rate": 7.560137457044674e-07, | |
| "loss": 0.0001, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 46.00824742268041, | |
| "grad_norm": 0.005698314867913723, | |
| "learning_rate": 7.445589919816725e-07, | |
| "loss": 0.0002, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 46.009278350515466, | |
| "grad_norm": 161.3025665283203, | |
| "learning_rate": 7.331042382588776e-07, | |
| "loss": 0.1556, | |
| "step": 9060 | |
| }, | |
| { | |
| "epoch": 46.01030927835052, | |
| "grad_norm": 0.010940884239971638, | |
| "learning_rate": 7.216494845360824e-07, | |
| "loss": 0.0938, | |
| "step": 9070 | |
| }, | |
| { | |
| "epoch": 46.011340206185565, | |
| "grad_norm": 0.004464665427803993, | |
| "learning_rate": 7.101947308132875e-07, | |
| "loss": 0.0002, | |
| "step": 9080 | |
| }, | |
| { | |
| "epoch": 46.01237113402062, | |
| "grad_norm": 0.0026364498771727085, | |
| "learning_rate": 6.987399770904926e-07, | |
| "loss": 0.0002, | |
| "step": 9090 | |
| }, | |
| { | |
| "epoch": 46.01340206185567, | |
| "grad_norm": 0.009777194820344448, | |
| "learning_rate": 6.872852233676977e-07, | |
| "loss": 0.0002, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 46.014432989690725, | |
| "grad_norm": 0.008579901419579983, | |
| "learning_rate": 6.758304696449027e-07, | |
| "loss": 0.0001, | |
| "step": 9110 | |
| }, | |
| { | |
| "epoch": 46.01546391752577, | |
| "grad_norm": 0.6798976063728333, | |
| "learning_rate": 6.643757159221077e-07, | |
| "loss": 0.015, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 46.016494845360825, | |
| "grad_norm": 0.032176949083805084, | |
| "learning_rate": 6.529209621993128e-07, | |
| "loss": 0.0002, | |
| "step": 9130 | |
| }, | |
| { | |
| "epoch": 46.01752577319588, | |
| "grad_norm": 0.010497340932488441, | |
| "learning_rate": 6.414662084765177e-07, | |
| "loss": 0.0002, | |
| "step": 9140 | |
| }, | |
| { | |
| "epoch": 46.01855670103093, | |
| "grad_norm": 0.022227320820093155, | |
| "learning_rate": 6.300114547537228e-07, | |
| "loss": 0.0006, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 46.01958762886598, | |
| "grad_norm": 0.02829979918897152, | |
| "learning_rate": 6.185567010309279e-07, | |
| "loss": 0.0004, | |
| "step": 9160 | |
| }, | |
| { | |
| "epoch": 46.02010309278351, | |
| "eval_accuracy": 0.7333333333333333, | |
| "eval_loss": 2.1075925827026367, | |
| "eval_runtime": 12.8105, | |
| "eval_samples_per_second": 3.513, | |
| "eval_steps_per_second": 0.937, | |
| "step": 9165 | |
| }, | |
| { | |
| "epoch": 47.00051546391752, | |
| "grad_norm": 0.004454383160918951, | |
| "learning_rate": 6.07101947308133e-07, | |
| "loss": 0.0002, | |
| "step": 9170 | |
| }, | |
| { | |
| "epoch": 47.001546391752576, | |
| "grad_norm": 0.007718723267316818, | |
| "learning_rate": 5.956471935853379e-07, | |
| "loss": 0.0004, | |
| "step": 9180 | |
| }, | |
| { | |
| "epoch": 47.00257731958763, | |
| "grad_norm": 0.0052038333378732204, | |
| "learning_rate": 5.84192439862543e-07, | |
| "loss": 0.1885, | |
| "step": 9190 | |
| }, | |
| { | |
| "epoch": 47.00360824742268, | |
| "grad_norm": 150.0712127685547, | |
| "learning_rate": 5.72737686139748e-07, | |
| "loss": 0.0726, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 47.00463917525773, | |
| "grad_norm": 0.002846953459084034, | |
| "learning_rate": 5.612829324169531e-07, | |
| "loss": 0.0002, | |
| "step": 9210 | |
| }, | |
| { | |
| "epoch": 47.00567010309278, | |
| "grad_norm": 0.0032731653191149235, | |
| "learning_rate": 5.498281786941581e-07, | |
| "loss": 0.0014, | |
| "step": 9220 | |
| }, | |
| { | |
| "epoch": 47.006701030927836, | |
| "grad_norm": 0.04309961199760437, | |
| "learning_rate": 5.383734249713631e-07, | |
| "loss": 0.0002, | |
| "step": 9230 | |
| }, | |
| { | |
| "epoch": 47.00773195876289, | |
| "grad_norm": 0.01399976946413517, | |
| "learning_rate": 5.269186712485682e-07, | |
| "loss": 0.0002, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 47.008762886597935, | |
| "grad_norm": 0.007444604765623808, | |
| "learning_rate": 5.154639175257732e-07, | |
| "loss": 0.0002, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 47.00979381443299, | |
| "grad_norm": 0.0034925006330013275, | |
| "learning_rate": 5.040091638029783e-07, | |
| "loss": 0.04, | |
| "step": 9260 | |
| }, | |
| { | |
| "epoch": 47.01082474226804, | |
| "grad_norm": 0.0037354633677750826, | |
| "learning_rate": 4.925544100801833e-07, | |
| "loss": 0.0002, | |
| "step": 9270 | |
| }, | |
| { | |
| "epoch": 47.011855670103095, | |
| "grad_norm": 0.0180855430662632, | |
| "learning_rate": 4.810996563573884e-07, | |
| "loss": 0.0004, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 47.01288659793814, | |
| "grad_norm": 0.007617347873747349, | |
| "learning_rate": 4.696449026345934e-07, | |
| "loss": 0.183, | |
| "step": 9290 | |
| }, | |
| { | |
| "epoch": 47.013917525773195, | |
| "grad_norm": 0.011123509146273136, | |
| "learning_rate": 4.581901489117984e-07, | |
| "loss": 0.0002, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 47.01494845360825, | |
| "grad_norm": 0.019915536046028137, | |
| "learning_rate": 4.467353951890035e-07, | |
| "loss": 0.0002, | |
| "step": 9310 | |
| }, | |
| { | |
| "epoch": 47.0159793814433, | |
| "grad_norm": 0.003200843231752515, | |
| "learning_rate": 4.352806414662085e-07, | |
| "loss": 0.0004, | |
| "step": 9320 | |
| }, | |
| { | |
| "epoch": 47.01701030927835, | |
| "grad_norm": 0.018746715039014816, | |
| "learning_rate": 4.2382588774341357e-07, | |
| "loss": 0.1612, | |
| "step": 9330 | |
| }, | |
| { | |
| "epoch": 47.0180412371134, | |
| "grad_norm": 0.004534402396529913, | |
| "learning_rate": 4.123711340206186e-07, | |
| "loss": 0.1429, | |
| "step": 9340 | |
| }, | |
| { | |
| "epoch": 47.019072164948454, | |
| "grad_norm": 0.014469140209257603, | |
| "learning_rate": 4.009163802978237e-07, | |
| "loss": 0.0001, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 47.02010309278351, | |
| "grad_norm": 0.0009225418325513601, | |
| "learning_rate": 3.8946162657502866e-07, | |
| "loss": 0.0645, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 47.02010309278351, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 2.1502678394317627, | |
| "eval_runtime": 12.792, | |
| "eval_samples_per_second": 3.518, | |
| "eval_steps_per_second": 0.938, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 48.00103092783505, | |
| "grad_norm": 0.008684216067194939, | |
| "learning_rate": 3.780068728522337e-07, | |
| "loss": 0.0002, | |
| "step": 9370 | |
| }, | |
| { | |
| "epoch": 48.00206185567011, | |
| "grad_norm": 0.002957036718726158, | |
| "learning_rate": 3.665521191294388e-07, | |
| "loss": 0.0059, | |
| "step": 9380 | |
| }, | |
| { | |
| "epoch": 48.00309278350515, | |
| "grad_norm": 0.0021803213749080896, | |
| "learning_rate": 3.5509736540664376e-07, | |
| "loss": 0.0002, | |
| "step": 9390 | |
| }, | |
| { | |
| "epoch": 48.004123711340206, | |
| "grad_norm": 0.00486848596483469, | |
| "learning_rate": 3.4364261168384884e-07, | |
| "loss": 0.0001, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 48.00515463917526, | |
| "grad_norm": 0.0031498922035098076, | |
| "learning_rate": 3.3218785796105387e-07, | |
| "loss": 0.1926, | |
| "step": 9410 | |
| }, | |
| { | |
| "epoch": 48.00618556701031, | |
| "grad_norm": 0.004292685072869062, | |
| "learning_rate": 3.2073310423825885e-07, | |
| "loss": 0.0026, | |
| "step": 9420 | |
| }, | |
| { | |
| "epoch": 48.00721649484536, | |
| "grad_norm": 0.003267282620072365, | |
| "learning_rate": 3.0927835051546394e-07, | |
| "loss": 0.0006, | |
| "step": 9430 | |
| }, | |
| { | |
| "epoch": 48.00824742268041, | |
| "grad_norm": 0.022121211513876915, | |
| "learning_rate": 2.9782359679266897e-07, | |
| "loss": 0.0001, | |
| "step": 9440 | |
| }, | |
| { | |
| "epoch": 48.009278350515466, | |
| "grad_norm": 375.67889404296875, | |
| "learning_rate": 2.86368843069874e-07, | |
| "loss": 0.0801, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 48.01030927835052, | |
| "grad_norm": 23.7979736328125, | |
| "learning_rate": 2.7491408934707903e-07, | |
| "loss": 0.0009, | |
| "step": 9460 | |
| }, | |
| { | |
| "epoch": 48.011340206185565, | |
| "grad_norm": 0.0010315030813217163, | |
| "learning_rate": 2.634593356242841e-07, | |
| "loss": 0.0671, | |
| "step": 9470 | |
| }, | |
| { | |
| "epoch": 48.01237113402062, | |
| "grad_norm": 0.01132703386247158, | |
| "learning_rate": 2.5200458190148915e-07, | |
| "loss": 0.0056, | |
| "step": 9480 | |
| }, | |
| { | |
| "epoch": 48.01340206185567, | |
| "grad_norm": 0.0034062955528497696, | |
| "learning_rate": 2.405498281786942e-07, | |
| "loss": 0.0002, | |
| "step": 9490 | |
| }, | |
| { | |
| "epoch": 48.014432989690725, | |
| "grad_norm": 0.23974573612213135, | |
| "learning_rate": 2.290950744558992e-07, | |
| "loss": 0.0002, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 48.01546391752577, | |
| "grad_norm": 0.006117957644164562, | |
| "learning_rate": 2.1764032073310424e-07, | |
| "loss": 0.3933, | |
| "step": 9510 | |
| }, | |
| { | |
| "epoch": 48.016494845360825, | |
| "grad_norm": 0.23375919461250305, | |
| "learning_rate": 2.061855670103093e-07, | |
| "loss": 0.0328, | |
| "step": 9520 | |
| }, | |
| { | |
| "epoch": 48.01752577319588, | |
| "grad_norm": 0.003966511692851782, | |
| "learning_rate": 1.9473081328751433e-07, | |
| "loss": 0.0002, | |
| "step": 9530 | |
| }, | |
| { | |
| "epoch": 48.01855670103093, | |
| "grad_norm": 0.015851015225052834, | |
| "learning_rate": 1.832760595647194e-07, | |
| "loss": 0.0002, | |
| "step": 9540 | |
| }, | |
| { | |
| "epoch": 48.01958762886598, | |
| "grad_norm": 0.07893943041563034, | |
| "learning_rate": 1.7182130584192442e-07, | |
| "loss": 0.1121, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 48.02010309278351, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 2.2610628604888916, | |
| "eval_runtime": 12.1107, | |
| "eval_samples_per_second": 3.716, | |
| "eval_steps_per_second": 0.991, | |
| "step": 9555 | |
| }, | |
| { | |
| "epoch": 49.00051546391752, | |
| "grad_norm": 85.26998901367188, | |
| "learning_rate": 1.6036655211912943e-07, | |
| "loss": 0.1761, | |
| "step": 9560 | |
| }, | |
| { | |
| "epoch": 49.001546391752576, | |
| "grad_norm": 0.005270041059702635, | |
| "learning_rate": 1.4891179839633448e-07, | |
| "loss": 0.1176, | |
| "step": 9570 | |
| }, | |
| { | |
| "epoch": 49.00257731958763, | |
| "grad_norm": 0.004510932136327028, | |
| "learning_rate": 1.3745704467353952e-07, | |
| "loss": 0.0033, | |
| "step": 9580 | |
| }, | |
| { | |
| "epoch": 49.00360824742268, | |
| "grad_norm": 0.005600025877356529, | |
| "learning_rate": 1.2600229095074457e-07, | |
| "loss": 0.2291, | |
| "step": 9590 | |
| }, | |
| { | |
| "epoch": 49.00463917525773, | |
| "grad_norm": 0.03189009428024292, | |
| "learning_rate": 1.145475372279496e-07, | |
| "loss": 0.0002, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 49.00567010309278, | |
| "grad_norm": 0.06988941878080368, | |
| "learning_rate": 1.0309278350515465e-07, | |
| "loss": 0.0002, | |
| "step": 9610 | |
| }, | |
| { | |
| "epoch": 49.006701030927836, | |
| "grad_norm": 0.0760478675365448, | |
| "learning_rate": 9.16380297823597e-08, | |
| "loss": 0.0005, | |
| "step": 9620 | |
| }, | |
| { | |
| "epoch": 49.00773195876289, | |
| "grad_norm": 113.97074890136719, | |
| "learning_rate": 8.018327605956471e-08, | |
| "loss": 0.1573, | |
| "step": 9630 | |
| }, | |
| { | |
| "epoch": 49.008762886597935, | |
| "grad_norm": 0.008412045426666737, | |
| "learning_rate": 6.872852233676976e-08, | |
| "loss": 0.2028, | |
| "step": 9640 | |
| }, | |
| { | |
| "epoch": 49.00979381443299, | |
| "grad_norm": 0.014376115053892136, | |
| "learning_rate": 5.72737686139748e-08, | |
| "loss": 0.0003, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 49.01082474226804, | |
| "grad_norm": 0.02352799102663994, | |
| "learning_rate": 4.581901489117985e-08, | |
| "loss": 0.1283, | |
| "step": 9660 | |
| }, | |
| { | |
| "epoch": 49.011855670103095, | |
| "grad_norm": 0.0021438777912408113, | |
| "learning_rate": 3.436426116838488e-08, | |
| "loss": 0.1895, | |
| "step": 9670 | |
| }, | |
| { | |
| "epoch": 49.01288659793814, | |
| "grad_norm": 44.23455810546875, | |
| "learning_rate": 2.2909507445589924e-08, | |
| "loss": 0.1931, | |
| "step": 9680 | |
| }, | |
| { | |
| "epoch": 49.013917525773195, | |
| "grad_norm": 0.00817310530692339, | |
| "learning_rate": 1.1454753722794962e-08, | |
| "loss": 0.0002, | |
| "step": 9690 | |
| }, | |
| { | |
| "epoch": 49.01494845360825, | |
| "grad_norm": 0.005044811405241489, | |
| "learning_rate": 0.0, | |
| "loss": 0.1268, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 49.01494845360825, | |
| "eval_accuracy": 0.7111111111111111, | |
| "eval_loss": 2.172813653945923, | |
| "eval_runtime": 13.8199, | |
| "eval_samples_per_second": 3.256, | |
| "eval_steps_per_second": 0.868, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 49.01494845360825, | |
| "step": 9700, | |
| "total_flos": 1.6972634016516597e+20, | |
| "train_loss": 0.2964441666583838, | |
| "train_runtime": 22323.5867, | |
| "train_samples_per_second": 1.738, | |
| "train_steps_per_second": 0.435 | |
| }, | |
| { | |
| "epoch": 49.01494845360825, | |
| "eval_accuracy": 0.8666666666666667, | |
| "eval_loss": 0.36465370655059814, | |
| "eval_runtime": 12.1222, | |
| "eval_samples_per_second": 3.712, | |
| "eval_steps_per_second": 0.99, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 49.01494845360825, | |
| "eval_accuracy": 0.8666666666666667, | |
| "eval_loss": 0.36465373635292053, | |
| "eval_runtime": 12.3017, | |
| "eval_samples_per_second": 3.658, | |
| "eval_steps_per_second": 0.975, | |
| "step": 9700 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 9700, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.6972634016516597e+20, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |