| { | |
| "best_metric": 0.8051581382751465, | |
| "best_model_checkpoint": "wav2vec2-vocal-disable-classification/checkpoint-513", | |
| "epoch": 9.0, | |
| "eval_steps": 1, | |
| "global_step": 513, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.017543859649122806, | |
| "grad_norm": 1.56613290309906, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 1.6056, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03508771929824561, | |
| "grad_norm": 1.8094905614852905, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 1.6101, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.05263157894736842, | |
| "grad_norm": 1.138840913772583, | |
| "learning_rate": 1.8e-06, | |
| "loss": 1.609, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.07017543859649122, | |
| "grad_norm": 3.695634603500366, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.6097, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.08771929824561403, | |
| "grad_norm": 1.9893696308135986, | |
| "learning_rate": 3e-06, | |
| "loss": 1.6028, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.10526315789473684, | |
| "grad_norm": 2.095263957977295, | |
| "learning_rate": 3.6e-06, | |
| "loss": 1.6016, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.12280701754385964, | |
| "grad_norm": 1.7659621238708496, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 1.5994, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.14035087719298245, | |
| "grad_norm": 2.041477680206299, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 1.5951, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.15789473684210525, | |
| "grad_norm": 1.6106765270233154, | |
| "learning_rate": 5.4e-06, | |
| "loss": 1.5957, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.17543859649122806, | |
| "grad_norm": 1.5744407176971436, | |
| "learning_rate": 6e-06, | |
| "loss": 1.5951, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.19298245614035087, | |
| "grad_norm": 1.0527958869934082, | |
| "learning_rate": 6.6e-06, | |
| "loss": 1.5994, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.21052631578947367, | |
| "grad_norm": 1.5314522981643677, | |
| "learning_rate": 7.2e-06, | |
| "loss": 1.5766, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.22807017543859648, | |
| "grad_norm": 1.0843933820724487, | |
| "learning_rate": 7.8e-06, | |
| "loss": 1.5976, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.24561403508771928, | |
| "grad_norm": 1.8420531749725342, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.5803, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.2631578947368421, | |
| "grad_norm": 2.5668458938598633, | |
| "learning_rate": 9e-06, | |
| "loss": 1.5839, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.2807017543859649, | |
| "grad_norm": 1.5706558227539062, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 1.5522, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.2982456140350877, | |
| "grad_norm": 1.722334384918213, | |
| "learning_rate": 1.02e-05, | |
| "loss": 1.5391, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.3157894736842105, | |
| "grad_norm": 1.317153811454773, | |
| "learning_rate": 1.08e-05, | |
| "loss": 1.6246, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 1.3377695083618164, | |
| "learning_rate": 1.1400000000000001e-05, | |
| "loss": 1.6166, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.3508771929824561, | |
| "grad_norm": 3.216266632080078, | |
| "learning_rate": 1.2e-05, | |
| "loss": 1.5151, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3684210526315789, | |
| "grad_norm": 1.3238260746002197, | |
| "learning_rate": 1.26e-05, | |
| "loss": 1.5195, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.38596491228070173, | |
| "grad_norm": 2.946779251098633, | |
| "learning_rate": 1.32e-05, | |
| "loss": 1.503, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.40350877192982454, | |
| "grad_norm": 1.427848219871521, | |
| "learning_rate": 1.3800000000000002e-05, | |
| "loss": 1.5538, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.42105263157894735, | |
| "grad_norm": 3.5313634872436523, | |
| "learning_rate": 1.44e-05, | |
| "loss": 1.545, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.43859649122807015, | |
| "grad_norm": 1.3384391069412231, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.5406, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.45614035087719296, | |
| "grad_norm": 1.635528802871704, | |
| "learning_rate": 1.56e-05, | |
| "loss": 1.4864, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.47368421052631576, | |
| "grad_norm": 1.3669475317001343, | |
| "learning_rate": 1.62e-05, | |
| "loss": 1.5609, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.49122807017543857, | |
| "grad_norm": 1.532422661781311, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 1.4829, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.5087719298245614, | |
| "grad_norm": 14.834467887878418, | |
| "learning_rate": 1.74e-05, | |
| "loss": 1.4049, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.5263157894736842, | |
| "grad_norm": 1.5839215517044067, | |
| "learning_rate": 1.8e-05, | |
| "loss": 1.4429, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.543859649122807, | |
| "grad_norm": 8.556641578674316, | |
| "learning_rate": 1.86e-05, | |
| "loss": 1.5076, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.5614035087719298, | |
| "grad_norm": 2.6420958042144775, | |
| "learning_rate": 1.9200000000000003e-05, | |
| "loss": 1.6325, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.5789473684210527, | |
| "grad_norm": 1.5453814268112183, | |
| "learning_rate": 1.98e-05, | |
| "loss": 1.5331, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.5964912280701754, | |
| "grad_norm": 2.1342880725860596, | |
| "learning_rate": 2.04e-05, | |
| "loss": 1.5563, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.6140350877192983, | |
| "grad_norm": 7.298281669616699, | |
| "learning_rate": 2.1e-05, | |
| "loss": 1.3024, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.631578947368421, | |
| "grad_norm": 2.423535108566284, | |
| "learning_rate": 2.16e-05, | |
| "loss": 1.5851, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.6491228070175439, | |
| "grad_norm": 2.8099899291992188, | |
| "learning_rate": 2.22e-05, | |
| "loss": 1.664, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 3.804816961288452, | |
| "learning_rate": 2.2800000000000002e-05, | |
| "loss": 1.4642, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.6842105263157895, | |
| "grad_norm": 5.346778392791748, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 1.5016, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.7017543859649122, | |
| "grad_norm": 5.4332733154296875, | |
| "learning_rate": 2.4e-05, | |
| "loss": 1.4937, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.7192982456140351, | |
| "grad_norm": 3.2267627716064453, | |
| "learning_rate": 2.4599999999999998e-05, | |
| "loss": 1.4857, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.7368421052631579, | |
| "grad_norm": 9.248668670654297, | |
| "learning_rate": 2.52e-05, | |
| "loss": 1.4574, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.7543859649122807, | |
| "grad_norm": 2.983009099960327, | |
| "learning_rate": 2.58e-05, | |
| "loss": 1.4802, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.7719298245614035, | |
| "grad_norm": 9.322035789489746, | |
| "learning_rate": 2.64e-05, | |
| "loss": 1.3646, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.7894736842105263, | |
| "grad_norm": 3.1577627658843994, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 1.3842, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.8070175438596491, | |
| "grad_norm": 3.0866541862487793, | |
| "learning_rate": 2.7600000000000003e-05, | |
| "loss": 1.478, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.8245614035087719, | |
| "grad_norm": 2.434828281402588, | |
| "learning_rate": 2.8199999999999998e-05, | |
| "loss": 1.2678, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.8421052631578947, | |
| "grad_norm": 19.617595672607422, | |
| "learning_rate": 2.88e-05, | |
| "loss": 1.5149, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.8596491228070176, | |
| "grad_norm": 3.927741527557373, | |
| "learning_rate": 2.94e-05, | |
| "loss": 1.3689, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.8771929824561403, | |
| "grad_norm": 4.532490253448486, | |
| "learning_rate": 3e-05, | |
| "loss": 1.5975, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.8947368421052632, | |
| "grad_norm": 4.9621500968933105, | |
| "learning_rate": 2.9972477064220187e-05, | |
| "loss": 1.4361, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.9122807017543859, | |
| "grad_norm": 6.061293125152588, | |
| "learning_rate": 2.994495412844037e-05, | |
| "loss": 1.4083, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.9298245614035088, | |
| "grad_norm": 4.741385459899902, | |
| "learning_rate": 2.991743119266055e-05, | |
| "loss": 1.4497, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.9473684210526315, | |
| "grad_norm": 3.2169480323791504, | |
| "learning_rate": 2.9889908256880734e-05, | |
| "loss": 1.4465, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.9649122807017544, | |
| "grad_norm": 7.039127349853516, | |
| "learning_rate": 2.986238532110092e-05, | |
| "loss": 1.2431, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.9824561403508771, | |
| "grad_norm": 6.163113594055176, | |
| "learning_rate": 2.9834862385321102e-05, | |
| "loss": 1.2167, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 3.4295806884765625, | |
| "learning_rate": 2.9807339449541285e-05, | |
| "loss": 1.3198, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4482758620689655, | |
| "eval_loss": 1.3703213930130005, | |
| "eval_roc_auc": 0.7240004822764974, | |
| "eval_runtime": 9.1707, | |
| "eval_samples_per_second": 28.46, | |
| "eval_steps_per_second": 1.854, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 1.0175438596491229, | |
| "grad_norm": 2.1605288982391357, | |
| "learning_rate": 2.9779816513761467e-05, | |
| "loss": 1.3161, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 1.0350877192982457, | |
| "grad_norm": 2.469620943069458, | |
| "learning_rate": 2.9752293577981653e-05, | |
| "loss": 1.4468, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 1.0526315789473684, | |
| "grad_norm": 5.3216376304626465, | |
| "learning_rate": 2.9724770642201836e-05, | |
| "loss": 1.3517, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.0701754385964912, | |
| "grad_norm": 5.191178798675537, | |
| "learning_rate": 2.9697247706422018e-05, | |
| "loss": 1.3841, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 1.087719298245614, | |
| "grad_norm": 2.3336193561553955, | |
| "learning_rate": 2.9669724770642204e-05, | |
| "loss": 1.3057, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 1.1052631578947367, | |
| "grad_norm": 42.809783935546875, | |
| "learning_rate": 2.9642201834862386e-05, | |
| "loss": 1.2346, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 1.1228070175438596, | |
| "grad_norm": 6.859128952026367, | |
| "learning_rate": 2.9614678899082572e-05, | |
| "loss": 1.4394, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 1.1403508771929824, | |
| "grad_norm": 9.006386756896973, | |
| "learning_rate": 2.958715596330275e-05, | |
| "loss": 1.2954, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.1578947368421053, | |
| "grad_norm": 2.543628454208374, | |
| "learning_rate": 2.9559633027522937e-05, | |
| "loss": 1.0557, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 1.1754385964912282, | |
| "grad_norm": 2.736093044281006, | |
| "learning_rate": 2.953211009174312e-05, | |
| "loss": 1.2225, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 1.1929824561403508, | |
| "grad_norm": 4.434782028198242, | |
| "learning_rate": 2.9504587155963306e-05, | |
| "loss": 1.1875, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 1.2105263157894737, | |
| "grad_norm": 8.687559127807617, | |
| "learning_rate": 2.9477064220183485e-05, | |
| "loss": 0.937, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.2280701754385965, | |
| "grad_norm": 3.4958977699279785, | |
| "learning_rate": 2.944954128440367e-05, | |
| "loss": 1.2851, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.2456140350877192, | |
| "grad_norm": 12.490559577941895, | |
| "learning_rate": 2.9422018348623853e-05, | |
| "loss": 1.5578, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.263157894736842, | |
| "grad_norm": 3.466176748275757, | |
| "learning_rate": 2.939449541284404e-05, | |
| "loss": 0.9698, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.280701754385965, | |
| "grad_norm": 9.730321884155273, | |
| "learning_rate": 2.936697247706422e-05, | |
| "loss": 1.2044, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.2982456140350878, | |
| "grad_norm": 12.908202171325684, | |
| "learning_rate": 2.9339449541284404e-05, | |
| "loss": 1.3778, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.3157894736842106, | |
| "grad_norm": 7.409069538116455, | |
| "learning_rate": 2.931192660550459e-05, | |
| "loss": 1.6065, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 12.185145378112793, | |
| "learning_rate": 2.9284403669724772e-05, | |
| "loss": 1.2463, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.3508771929824561, | |
| "grad_norm": 16.505674362182617, | |
| "learning_rate": 2.9256880733944955e-05, | |
| "loss": 1.0027, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.368421052631579, | |
| "grad_norm": 9.078436851501465, | |
| "learning_rate": 2.9229357798165137e-05, | |
| "loss": 1.2037, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.3859649122807016, | |
| "grad_norm": 23.13191032409668, | |
| "learning_rate": 2.9201834862385323e-05, | |
| "loss": 1.0528, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.4035087719298245, | |
| "grad_norm": 7.651644229888916, | |
| "learning_rate": 2.9174311926605506e-05, | |
| "loss": 1.0373, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.4210526315789473, | |
| "grad_norm": 15.649460792541504, | |
| "learning_rate": 2.9146788990825688e-05, | |
| "loss": 1.3094, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.4385964912280702, | |
| "grad_norm": 14.335933685302734, | |
| "learning_rate": 2.9119266055045874e-05, | |
| "loss": 1.229, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.456140350877193, | |
| "grad_norm": 13.704193115234375, | |
| "learning_rate": 2.9091743119266056e-05, | |
| "loss": 1.2517, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.4736842105263157, | |
| "grad_norm": 6.872323036193848, | |
| "learning_rate": 2.9064220183486242e-05, | |
| "loss": 1.0614, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.4912280701754386, | |
| "grad_norm": 58.63323211669922, | |
| "learning_rate": 2.903669724770642e-05, | |
| "loss": 1.2651, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.5087719298245614, | |
| "grad_norm": 16.493391036987305, | |
| "learning_rate": 2.9009174311926607e-05, | |
| "loss": 1.0977, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.526315789473684, | |
| "grad_norm": 76.30412292480469, | |
| "learning_rate": 2.898165137614679e-05, | |
| "loss": 1.3534, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.543859649122807, | |
| "grad_norm": 35.07970428466797, | |
| "learning_rate": 2.8954128440366975e-05, | |
| "loss": 1.1015, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.5614035087719298, | |
| "grad_norm": 41.934513092041016, | |
| "learning_rate": 2.8926605504587155e-05, | |
| "loss": 1.252, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.5789473684210527, | |
| "grad_norm": 6.429330825805664, | |
| "learning_rate": 2.889908256880734e-05, | |
| "loss": 0.9942, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.5964912280701755, | |
| "grad_norm": 4.915737152099609, | |
| "learning_rate": 2.8871559633027523e-05, | |
| "loss": 1.0583, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.6140350877192984, | |
| "grad_norm": 5.171088695526123, | |
| "learning_rate": 2.884403669724771e-05, | |
| "loss": 0.9167, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.631578947368421, | |
| "grad_norm": 4.840869426727295, | |
| "learning_rate": 2.881651376146789e-05, | |
| "loss": 1.3165, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.6491228070175439, | |
| "grad_norm": 8.278257369995117, | |
| "learning_rate": 2.8788990825688074e-05, | |
| "loss": 1.0294, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 7.380605697631836, | |
| "learning_rate": 2.876146788990826e-05, | |
| "loss": 1.16, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.6842105263157894, | |
| "grad_norm": 27.351318359375, | |
| "learning_rate": 2.8733944954128442e-05, | |
| "loss": 1.6427, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.7017543859649122, | |
| "grad_norm": 5.766673564910889, | |
| "learning_rate": 2.8706422018348625e-05, | |
| "loss": 1.1461, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.719298245614035, | |
| "grad_norm": 6.586304664611816, | |
| "learning_rate": 2.8678899082568807e-05, | |
| "loss": 1.1368, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.736842105263158, | |
| "grad_norm": 13.888910293579102, | |
| "learning_rate": 2.8651376146788993e-05, | |
| "loss": 0.8426, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.7543859649122808, | |
| "grad_norm": 14.992203712463379, | |
| "learning_rate": 2.8623853211009175e-05, | |
| "loss": 1.1494, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.7719298245614035, | |
| "grad_norm": 8.557571411132812, | |
| "learning_rate": 2.8596330275229358e-05, | |
| "loss": 1.1022, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.7894736842105263, | |
| "grad_norm": 19.743364334106445, | |
| "learning_rate": 2.856880733944954e-05, | |
| "loss": 1.0737, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.807017543859649, | |
| "grad_norm": 13.77326488494873, | |
| "learning_rate": 2.8541284403669726e-05, | |
| "loss": 0.8547, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.8245614035087718, | |
| "grad_norm": 4.906980514526367, | |
| "learning_rate": 2.8513761467889912e-05, | |
| "loss": 0.8529, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.8421052631578947, | |
| "grad_norm": 2.8242599964141846, | |
| "learning_rate": 2.848623853211009e-05, | |
| "loss": 1.2058, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.8596491228070176, | |
| "grad_norm": 3.802095890045166, | |
| "learning_rate": 2.8458715596330277e-05, | |
| "loss": 0.9236, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.8771929824561404, | |
| "grad_norm": 4.197514057159424, | |
| "learning_rate": 2.843119266055046e-05, | |
| "loss": 0.8856, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.8947368421052633, | |
| "grad_norm": 13.207415580749512, | |
| "learning_rate": 2.8403669724770645e-05, | |
| "loss": 0.769, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.912280701754386, | |
| "grad_norm": 4.843390941619873, | |
| "learning_rate": 2.8376146788990824e-05, | |
| "loss": 1.1706, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.9298245614035088, | |
| "grad_norm": 13.065381050109863, | |
| "learning_rate": 2.834862385321101e-05, | |
| "loss": 1.0655, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.9473684210526314, | |
| "grad_norm": 15.791166305541992, | |
| "learning_rate": 2.8321100917431193e-05, | |
| "loss": 1.2378, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.9649122807017543, | |
| "grad_norm": 11.551010131835938, | |
| "learning_rate": 2.829357798165138e-05, | |
| "loss": 0.9768, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.9824561403508771, | |
| "grad_norm": 14.838546752929688, | |
| "learning_rate": 2.826605504587156e-05, | |
| "loss": 1.0073, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 2.385023355484009, | |
| "learning_rate": 2.8238532110091744e-05, | |
| "loss": 0.9486, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6091954022988506, | |
| "eval_loss": 1.0561790466308594, | |
| "eval_roc_auc": 0.8297140297597192, | |
| "eval_runtime": 10.2393, | |
| "eval_samples_per_second": 25.49, | |
| "eval_steps_per_second": 1.66, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 2.017543859649123, | |
| "grad_norm": 13.100618362426758, | |
| "learning_rate": 2.821100917431193e-05, | |
| "loss": 0.9394, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 2.0350877192982457, | |
| "grad_norm": 22.331451416015625, | |
| "learning_rate": 2.8183486238532112e-05, | |
| "loss": 1.0603, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 2.0526315789473686, | |
| "grad_norm": 4.956656455993652, | |
| "learning_rate": 2.8155963302752294e-05, | |
| "loss": 0.9566, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 2.0701754385964914, | |
| "grad_norm": 19.40747833251953, | |
| "learning_rate": 2.8128440366972477e-05, | |
| "loss": 1.1504, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 2.087719298245614, | |
| "grad_norm": 13.512128829956055, | |
| "learning_rate": 2.8100917431192663e-05, | |
| "loss": 1.194, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 2.1052631578947367, | |
| "grad_norm": 6.616446495056152, | |
| "learning_rate": 2.8073394495412845e-05, | |
| "loss": 1.0321, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.1228070175438596, | |
| "grad_norm": 2.3998453617095947, | |
| "learning_rate": 2.8045871559633028e-05, | |
| "loss": 1.0636, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 2.1403508771929824, | |
| "grad_norm": 3.214811086654663, | |
| "learning_rate": 2.801834862385321e-05, | |
| "loss": 0.7859, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 2.1578947368421053, | |
| "grad_norm": 2.0244710445404053, | |
| "learning_rate": 2.7990825688073396e-05, | |
| "loss": 0.8801, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 2.175438596491228, | |
| "grad_norm": 3.1682400703430176, | |
| "learning_rate": 2.7963302752293582e-05, | |
| "loss": 0.9413, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 2.192982456140351, | |
| "grad_norm": 3.509514570236206, | |
| "learning_rate": 2.793577981651376e-05, | |
| "loss": 0.8083, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 2.2105263157894735, | |
| "grad_norm": 6.175449848175049, | |
| "learning_rate": 2.7908256880733947e-05, | |
| "loss": 1.0884, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 2.2280701754385963, | |
| "grad_norm": 4.469012260437012, | |
| "learning_rate": 2.788073394495413e-05, | |
| "loss": 0.8281, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 2.245614035087719, | |
| "grad_norm": 18.924325942993164, | |
| "learning_rate": 2.7853211009174315e-05, | |
| "loss": 0.9676, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 2.263157894736842, | |
| "grad_norm": 24.372833251953125, | |
| "learning_rate": 2.7825688073394494e-05, | |
| "loss": 1.1955, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 2.280701754385965, | |
| "grad_norm": 1.3634321689605713, | |
| "learning_rate": 2.779816513761468e-05, | |
| "loss": 0.9889, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 2.2982456140350878, | |
| "grad_norm": 17.57159996032715, | |
| "learning_rate": 2.7770642201834863e-05, | |
| "loss": 1.1324, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 2.3157894736842106, | |
| "grad_norm": 16.63871192932129, | |
| "learning_rate": 2.774311926605505e-05, | |
| "loss": 1.1951, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 2.3333333333333335, | |
| "grad_norm": 16.466339111328125, | |
| "learning_rate": 2.7715596330275228e-05, | |
| "loss": 0.9375, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 2.3508771929824563, | |
| "grad_norm": 13.808109283447266, | |
| "learning_rate": 2.7688073394495413e-05, | |
| "loss": 1.0418, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 2.3684210526315788, | |
| "grad_norm": 5.8455729484558105, | |
| "learning_rate": 2.76605504587156e-05, | |
| "loss": 1.0906, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 2.3859649122807016, | |
| "grad_norm": 18.494245529174805, | |
| "learning_rate": 2.7633027522935782e-05, | |
| "loss": 0.9786, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 2.4035087719298245, | |
| "grad_norm": 21.47800636291504, | |
| "learning_rate": 2.7605504587155964e-05, | |
| "loss": 0.8138, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 2.4210526315789473, | |
| "grad_norm": 8.400023460388184, | |
| "learning_rate": 2.7577981651376147e-05, | |
| "loss": 1.027, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 2.43859649122807, | |
| "grad_norm": 7.198583602905273, | |
| "learning_rate": 2.7550458715596333e-05, | |
| "loss": 0.7104, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.456140350877193, | |
| "grad_norm": 23.315410614013672, | |
| "learning_rate": 2.7522935779816515e-05, | |
| "loss": 1.2115, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.473684210526316, | |
| "grad_norm": 18.734041213989258, | |
| "learning_rate": 2.7495412844036697e-05, | |
| "loss": 0.9739, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.4912280701754383, | |
| "grad_norm": 10.015871047973633, | |
| "learning_rate": 2.746788990825688e-05, | |
| "loss": 0.9796, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.5087719298245617, | |
| "grad_norm": 11.979524612426758, | |
| "learning_rate": 2.7440366972477066e-05, | |
| "loss": 0.8025, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.526315789473684, | |
| "grad_norm": 9.438429832458496, | |
| "learning_rate": 2.741284403669725e-05, | |
| "loss": 0.8838, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.543859649122807, | |
| "grad_norm": 43.801063537597656, | |
| "learning_rate": 2.738532110091743e-05, | |
| "loss": 0.9144, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.56140350877193, | |
| "grad_norm": 19.495887756347656, | |
| "learning_rate": 2.7357798165137617e-05, | |
| "loss": 0.8573, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.5789473684210527, | |
| "grad_norm": 21.922138214111328, | |
| "learning_rate": 2.73302752293578e-05, | |
| "loss": 0.9611, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.5964912280701755, | |
| "grad_norm": 16.30131721496582, | |
| "learning_rate": 2.7302752293577985e-05, | |
| "loss": 1.3519, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.6140350877192984, | |
| "grad_norm": 43.10526657104492, | |
| "learning_rate": 2.7275229357798164e-05, | |
| "loss": 1.0194, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.6315789473684212, | |
| "grad_norm": 18.148771286010742, | |
| "learning_rate": 2.724770642201835e-05, | |
| "loss": 1.0759, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.6491228070175437, | |
| "grad_norm": 28.365219116210938, | |
| "learning_rate": 2.7220183486238532e-05, | |
| "loss": 0.9998, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 10.812620162963867, | |
| "learning_rate": 2.7192660550458718e-05, | |
| "loss": 0.9368, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.6842105263157894, | |
| "grad_norm": 32.16633605957031, | |
| "learning_rate": 2.7165137614678897e-05, | |
| "loss": 1.0304, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.7017543859649122, | |
| "grad_norm": 63.48967742919922, | |
| "learning_rate": 2.7137614678899083e-05, | |
| "loss": 1.0795, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.719298245614035, | |
| "grad_norm": 4.874389171600342, | |
| "learning_rate": 2.711009174311927e-05, | |
| "loss": 1.062, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.736842105263158, | |
| "grad_norm": 18.235118865966797, | |
| "learning_rate": 2.708256880733945e-05, | |
| "loss": 1.3184, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.754385964912281, | |
| "grad_norm": 22.853351593017578, | |
| "learning_rate": 2.7055045871559634e-05, | |
| "loss": 1.0108, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.7719298245614032, | |
| "grad_norm": 50.805885314941406, | |
| "learning_rate": 2.7027522935779817e-05, | |
| "loss": 0.914, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.7894736842105265, | |
| "grad_norm": 5.718901634216309, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 0.7322, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.807017543859649, | |
| "grad_norm": 60.18952178955078, | |
| "learning_rate": 2.6972477064220185e-05, | |
| "loss": 1.088, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.824561403508772, | |
| "grad_norm": 5.959671974182129, | |
| "learning_rate": 2.6944954128440367e-05, | |
| "loss": 0.5275, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.8421052631578947, | |
| "grad_norm": 2.963286876678467, | |
| "learning_rate": 2.691743119266055e-05, | |
| "loss": 0.8271, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.8596491228070176, | |
| "grad_norm": 18.080677032470703, | |
| "learning_rate": 2.6889908256880736e-05, | |
| "loss": 0.7593, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.8771929824561404, | |
| "grad_norm": 3.0679006576538086, | |
| "learning_rate": 2.6862385321100918e-05, | |
| "loss": 0.582, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.8947368421052633, | |
| "grad_norm": 27.97750473022461, | |
| "learning_rate": 2.68348623853211e-05, | |
| "loss": 1.0477, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.912280701754386, | |
| "grad_norm": 9.784193992614746, | |
| "learning_rate": 2.6807339449541286e-05, | |
| "loss": 0.7083, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.9298245614035086, | |
| "grad_norm": 7.2379045486450195, | |
| "learning_rate": 2.677981651376147e-05, | |
| "loss": 0.6767, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.9473684210526314, | |
| "grad_norm": 17.974803924560547, | |
| "learning_rate": 2.6752293577981655e-05, | |
| "loss": 1.117, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.9649122807017543, | |
| "grad_norm": 8.451967239379883, | |
| "learning_rate": 2.6724770642201834e-05, | |
| "loss": 1.0163, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.982456140350877, | |
| "grad_norm": 16.834774017333984, | |
| "learning_rate": 2.669724770642202e-05, | |
| "loss": 0.642, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 10.743306159973145, | |
| "learning_rate": 2.6669724770642202e-05, | |
| "loss": 0.8485, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.5938697318007663, | |
| "eval_loss": 1.042966604232788, | |
| "eval_roc_auc": 0.8370073235789949, | |
| "eval_runtime": 9.8627, | |
| "eval_samples_per_second": 26.463, | |
| "eval_steps_per_second": 1.724, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 3.017543859649123, | |
| "grad_norm": 7.450799942016602, | |
| "learning_rate": 2.6642201834862388e-05, | |
| "loss": 0.7936, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 3.0350877192982457, | |
| "grad_norm": 13.938980102539062, | |
| "learning_rate": 2.6614678899082567e-05, | |
| "loss": 0.9981, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 3.0526315789473686, | |
| "grad_norm": 10.046002388000488, | |
| "learning_rate": 2.6587155963302753e-05, | |
| "loss": 0.9664, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 3.0701754385964914, | |
| "grad_norm": 2.47877836227417, | |
| "learning_rate": 2.6559633027522936e-05, | |
| "loss": 0.4564, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 3.087719298245614, | |
| "grad_norm": 4.682964324951172, | |
| "learning_rate": 2.653211009174312e-05, | |
| "loss": 0.7715, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 3.1052631578947367, | |
| "grad_norm": 31.68276023864746, | |
| "learning_rate": 2.6504587155963304e-05, | |
| "loss": 1.0283, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 3.1228070175438596, | |
| "grad_norm": 6.329732418060303, | |
| "learning_rate": 2.6477064220183486e-05, | |
| "loss": 0.6996, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 3.1403508771929824, | |
| "grad_norm": 14.793739318847656, | |
| "learning_rate": 2.6449541284403672e-05, | |
| "loss": 0.5858, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 3.1578947368421053, | |
| "grad_norm": 123.91106414794922, | |
| "learning_rate": 2.6422018348623855e-05, | |
| "loss": 0.8345, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 3.175438596491228, | |
| "grad_norm": 10.042440414428711, | |
| "learning_rate": 2.6394495412844037e-05, | |
| "loss": 0.8356, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 3.192982456140351, | |
| "grad_norm": 15.220785140991211, | |
| "learning_rate": 2.636697247706422e-05, | |
| "loss": 0.5512, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 3.2105263157894735, | |
| "grad_norm": 9.333229064941406, | |
| "learning_rate": 2.6339449541284406e-05, | |
| "loss": 0.6247, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 3.2280701754385963, | |
| "grad_norm": 13.827474594116211, | |
| "learning_rate": 2.6311926605504588e-05, | |
| "loss": 1.0353, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 3.245614035087719, | |
| "grad_norm": 4.803915500640869, | |
| "learning_rate": 2.628440366972477e-05, | |
| "loss": 0.8724, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 3.263157894736842, | |
| "grad_norm": 9.242655754089355, | |
| "learning_rate": 2.6256880733944953e-05, | |
| "loss": 0.7395, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 3.280701754385965, | |
| "grad_norm": 82.34434509277344, | |
| "learning_rate": 2.622935779816514e-05, | |
| "loss": 0.7797, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 3.2982456140350878, | |
| "grad_norm": 29.72227668762207, | |
| "learning_rate": 2.6201834862385325e-05, | |
| "loss": 0.9626, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 3.3157894736842106, | |
| "grad_norm": 30.35354232788086, | |
| "learning_rate": 2.6174311926605504e-05, | |
| "loss": 1.0387, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 9.3469820022583, | |
| "learning_rate": 2.614678899082569e-05, | |
| "loss": 0.9555, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 3.3508771929824563, | |
| "grad_norm": 10.0740327835083, | |
| "learning_rate": 2.6119266055045872e-05, | |
| "loss": 0.7608, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 3.3684210526315788, | |
| "grad_norm": 16.590267181396484, | |
| "learning_rate": 2.6091743119266058e-05, | |
| "loss": 0.8894, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 3.3859649122807016, | |
| "grad_norm": 6.453067779541016, | |
| "learning_rate": 2.6064220183486237e-05, | |
| "loss": 0.6409, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 3.4035087719298245, | |
| "grad_norm": 18.96304702758789, | |
| "learning_rate": 2.6036697247706423e-05, | |
| "loss": 1.1674, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 3.4210526315789473, | |
| "grad_norm": 4.519999027252197, | |
| "learning_rate": 2.6009174311926605e-05, | |
| "loss": 0.9088, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 3.43859649122807, | |
| "grad_norm": 14.560205459594727, | |
| "learning_rate": 2.598165137614679e-05, | |
| "loss": 1.0394, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 3.456140350877193, | |
| "grad_norm": 9.643004417419434, | |
| "learning_rate": 2.5954128440366974e-05, | |
| "loss": 0.7166, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 3.473684210526316, | |
| "grad_norm": 6.686179161071777, | |
| "learning_rate": 2.5926605504587156e-05, | |
| "loss": 0.9138, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 3.4912280701754383, | |
| "grad_norm": 6.065926551818848, | |
| "learning_rate": 2.5899082568807342e-05, | |
| "loss": 0.5985, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 3.5087719298245617, | |
| "grad_norm": 8.297146797180176, | |
| "learning_rate": 2.5871559633027525e-05, | |
| "loss": 0.8459, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 3.526315789473684, | |
| "grad_norm": 28.6617431640625, | |
| "learning_rate": 2.5844036697247707e-05, | |
| "loss": 0.4553, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 3.543859649122807, | |
| "grad_norm": 8.133190155029297, | |
| "learning_rate": 2.581651376146789e-05, | |
| "loss": 0.8616, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 3.56140350877193, | |
| "grad_norm": 7.848640441894531, | |
| "learning_rate": 2.5788990825688075e-05, | |
| "loss": 1.0041, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 3.5789473684210527, | |
| "grad_norm": 12.316092491149902, | |
| "learning_rate": 2.5761467889908258e-05, | |
| "loss": 0.8715, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 3.5964912280701755, | |
| "grad_norm": 7.93389892578125, | |
| "learning_rate": 2.573394495412844e-05, | |
| "loss": 0.7744, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 3.6140350877192984, | |
| "grad_norm": 9.929866790771484, | |
| "learning_rate": 2.5706422018348623e-05, | |
| "loss": 0.6355, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 3.6315789473684212, | |
| "grad_norm": 11.27926254272461, | |
| "learning_rate": 2.567889908256881e-05, | |
| "loss": 1.0616, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 3.6491228070175437, | |
| "grad_norm": 12.538276672363281, | |
| "learning_rate": 2.5651376146788994e-05, | |
| "loss": 0.8681, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 3.6666666666666665, | |
| "grad_norm": 14.14212703704834, | |
| "learning_rate": 2.5623853211009174e-05, | |
| "loss": 0.8615, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 3.6842105263157894, | |
| "grad_norm": 12.915414810180664, | |
| "learning_rate": 2.559633027522936e-05, | |
| "loss": 0.905, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 3.7017543859649122, | |
| "grad_norm": 6.382973670959473, | |
| "learning_rate": 2.5568807339449542e-05, | |
| "loss": 0.8229, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 3.719298245614035, | |
| "grad_norm": 8.958497047424316, | |
| "learning_rate": 2.5541284403669728e-05, | |
| "loss": 0.8367, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 3.736842105263158, | |
| "grad_norm": 13.200749397277832, | |
| "learning_rate": 2.5513761467889907e-05, | |
| "loss": 0.7738, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 3.754385964912281, | |
| "grad_norm": 25.975767135620117, | |
| "learning_rate": 2.5486238532110093e-05, | |
| "loss": 1.1707, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 3.7719298245614032, | |
| "grad_norm": 46.98295211791992, | |
| "learning_rate": 2.5458715596330275e-05, | |
| "loss": 0.815, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 3.7894736842105265, | |
| "grad_norm": 6.342257499694824, | |
| "learning_rate": 2.543119266055046e-05, | |
| "loss": 0.9045, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 3.807017543859649, | |
| "grad_norm": 11.122331619262695, | |
| "learning_rate": 2.540366972477064e-05, | |
| "loss": 0.4915, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 3.824561403508772, | |
| "grad_norm": 11.39393424987793, | |
| "learning_rate": 2.5376146788990826e-05, | |
| "loss": 0.8584, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 3.8421052631578947, | |
| "grad_norm": 16.47450828552246, | |
| "learning_rate": 2.5348623853211012e-05, | |
| "loss": 0.9062, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 3.8596491228070176, | |
| "grad_norm": 8.9651517868042, | |
| "learning_rate": 2.5321100917431194e-05, | |
| "loss": 0.6332, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 3.8771929824561404, | |
| "grad_norm": 14.864767074584961, | |
| "learning_rate": 2.5293577981651377e-05, | |
| "loss": 0.7327, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 3.8947368421052633, | |
| "grad_norm": 13.695717811584473, | |
| "learning_rate": 2.526605504587156e-05, | |
| "loss": 0.8439, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 3.912280701754386, | |
| "grad_norm": 10.765061378479004, | |
| "learning_rate": 2.5238532110091745e-05, | |
| "loss": 1.0394, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 3.9298245614035086, | |
| "grad_norm": 15.805130958557129, | |
| "learning_rate": 2.5211009174311928e-05, | |
| "loss": 1.3535, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 3.9473684210526314, | |
| "grad_norm": 19.387144088745117, | |
| "learning_rate": 2.518348623853211e-05, | |
| "loss": 0.7274, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 3.9649122807017543, | |
| "grad_norm": 37.748497009277344, | |
| "learning_rate": 2.5155963302752293e-05, | |
| "loss": 0.7626, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 3.982456140350877, | |
| "grad_norm": 16.551725387573242, | |
| "learning_rate": 2.512844036697248e-05, | |
| "loss": 0.9978, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 25.66643524169922, | |
| "learning_rate": 2.510091743119266e-05, | |
| "loss": 0.7405, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.6245210727969349, | |
| "eval_loss": 0.9546777606010437, | |
| "eval_roc_auc": 0.8586184139970431, | |
| "eval_runtime": 9.814, | |
| "eval_samples_per_second": 26.595, | |
| "eval_steps_per_second": 1.732, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 4.017543859649122, | |
| "grad_norm": 4.81406307220459, | |
| "learning_rate": 2.5073394495412843e-05, | |
| "loss": 0.5165, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 4.035087719298246, | |
| "grad_norm": 9.769265174865723, | |
| "learning_rate": 2.504587155963303e-05, | |
| "loss": 0.6222, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 4.052631578947368, | |
| "grad_norm": 8.312880516052246, | |
| "learning_rate": 2.5018348623853212e-05, | |
| "loss": 0.6313, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 4.0701754385964914, | |
| "grad_norm": 5.13502311706543, | |
| "learning_rate": 2.4990825688073398e-05, | |
| "loss": 0.7425, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 4.087719298245614, | |
| "grad_norm": 19.665512084960938, | |
| "learning_rate": 2.4963302752293577e-05, | |
| "loss": 1.0384, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 4.105263157894737, | |
| "grad_norm": 29.155986785888672, | |
| "learning_rate": 2.4935779816513763e-05, | |
| "loss": 1.0976, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 4.12280701754386, | |
| "grad_norm": 11.831360816955566, | |
| "learning_rate": 2.4908256880733945e-05, | |
| "loss": 0.7877, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 4.140350877192983, | |
| "grad_norm": 19.49347686767578, | |
| "learning_rate": 2.488073394495413e-05, | |
| "loss": 0.6279, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 4.157894736842105, | |
| "grad_norm": 6.922645092010498, | |
| "learning_rate": 2.485321100917431e-05, | |
| "loss": 0.7153, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 4.175438596491228, | |
| "grad_norm": 3.6549301147460938, | |
| "learning_rate": 2.4825688073394496e-05, | |
| "loss": 0.873, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 4.192982456140351, | |
| "grad_norm": 10.724910736083984, | |
| "learning_rate": 2.4798165137614682e-05, | |
| "loss": 0.7728, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 4.2105263157894735, | |
| "grad_norm": 3.7120726108551025, | |
| "learning_rate": 2.4770642201834864e-05, | |
| "loss": 0.7132, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 4.228070175438597, | |
| "grad_norm": 29.73471450805664, | |
| "learning_rate": 2.4743119266055047e-05, | |
| "loss": 0.5634, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 4.245614035087719, | |
| "grad_norm": 5.868150234222412, | |
| "learning_rate": 2.471559633027523e-05, | |
| "loss": 0.6967, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 4.2631578947368425, | |
| "grad_norm": 9.828577995300293, | |
| "learning_rate": 2.4688073394495415e-05, | |
| "loss": 0.8814, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 4.280701754385965, | |
| "grad_norm": 19.289398193359375, | |
| "learning_rate": 2.4660550458715597e-05, | |
| "loss": 0.6907, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 4.298245614035087, | |
| "grad_norm": 18.95272445678711, | |
| "learning_rate": 2.463302752293578e-05, | |
| "loss": 0.7413, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 4.315789473684211, | |
| "grad_norm": 9.175381660461426, | |
| "learning_rate": 2.4605504587155962e-05, | |
| "loss": 0.5549, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 4.333333333333333, | |
| "grad_norm": 11.017579078674316, | |
| "learning_rate": 2.4577981651376148e-05, | |
| "loss": 0.5834, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 4.350877192982456, | |
| "grad_norm": 4.641144752502441, | |
| "learning_rate": 2.455045871559633e-05, | |
| "loss": 0.7171, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 4.368421052631579, | |
| "grad_norm": 30.48017120361328, | |
| "learning_rate": 2.4522935779816513e-05, | |
| "loss": 0.5314, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 4.385964912280702, | |
| "grad_norm": 5.151516914367676, | |
| "learning_rate": 2.44954128440367e-05, | |
| "loss": 0.822, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 4.4035087719298245, | |
| "grad_norm": 4.152652263641357, | |
| "learning_rate": 2.446788990825688e-05, | |
| "loss": 0.8574, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 4.421052631578947, | |
| "grad_norm": 26.325111389160156, | |
| "learning_rate": 2.4440366972477067e-05, | |
| "loss": 0.5536, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 4.43859649122807, | |
| "grad_norm": 3.4374260902404785, | |
| "learning_rate": 2.4412844036697247e-05, | |
| "loss": 0.7328, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 4.456140350877193, | |
| "grad_norm": 7.868347644805908, | |
| "learning_rate": 2.4385321100917432e-05, | |
| "loss": 0.8698, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 4.473684210526316, | |
| "grad_norm": 14.990694046020508, | |
| "learning_rate": 2.4357798165137615e-05, | |
| "loss": 0.9928, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 4.491228070175438, | |
| "grad_norm": 55.8467903137207, | |
| "learning_rate": 2.43302752293578e-05, | |
| "loss": 0.8542, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 4.508771929824562, | |
| "grad_norm": 15.516489028930664, | |
| "learning_rate": 2.430275229357798e-05, | |
| "loss": 0.935, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 4.526315789473684, | |
| "grad_norm": 31.769515991210938, | |
| "learning_rate": 2.4275229357798166e-05, | |
| "loss": 0.5695, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 4.543859649122807, | |
| "grad_norm": 6.501729488372803, | |
| "learning_rate": 2.4247706422018348e-05, | |
| "loss": 0.7384, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 4.56140350877193, | |
| "grad_norm": 13.194774627685547, | |
| "learning_rate": 2.4220183486238534e-05, | |
| "loss": 0.7311, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 4.578947368421053, | |
| "grad_norm": 9.523067474365234, | |
| "learning_rate": 2.4192660550458717e-05, | |
| "loss": 0.8254, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 4.5964912280701755, | |
| "grad_norm": 40.60873794555664, | |
| "learning_rate": 2.41651376146789e-05, | |
| "loss": 0.7756, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 4.614035087719298, | |
| "grad_norm": 20.81313133239746, | |
| "learning_rate": 2.4137614678899085e-05, | |
| "loss": 0.7474, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 4.631578947368421, | |
| "grad_norm": 30.596458435058594, | |
| "learning_rate": 2.4110091743119267e-05, | |
| "loss": 0.8261, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 4.649122807017544, | |
| "grad_norm": 36.82753372192383, | |
| "learning_rate": 2.408256880733945e-05, | |
| "loss": 0.7422, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 4.666666666666667, | |
| "grad_norm": 7.798766136169434, | |
| "learning_rate": 2.4055045871559632e-05, | |
| "loss": 0.7433, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 4.684210526315789, | |
| "grad_norm": 10.442023277282715, | |
| "learning_rate": 2.4027522935779818e-05, | |
| "loss": 0.9382, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 4.701754385964913, | |
| "grad_norm": 7.129739761352539, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.4544, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 4.719298245614035, | |
| "grad_norm": 6.29140567779541, | |
| "learning_rate": 2.3972477064220183e-05, | |
| "loss": 0.5828, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 4.7368421052631575, | |
| "grad_norm": 14.20146369934082, | |
| "learning_rate": 2.3944954128440366e-05, | |
| "loss": 0.7544, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 4.754385964912281, | |
| "grad_norm": 25.83724594116211, | |
| "learning_rate": 2.391743119266055e-05, | |
| "loss": 0.8725, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 4.771929824561403, | |
| "grad_norm": 19.661861419677734, | |
| "learning_rate": 2.3889908256880737e-05, | |
| "loss": 0.9125, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 4.7894736842105265, | |
| "grad_norm": 7.706132888793945, | |
| "learning_rate": 2.3862385321100916e-05, | |
| "loss": 0.4359, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 4.807017543859649, | |
| "grad_norm": 13.709455490112305, | |
| "learning_rate": 2.3834862385321102e-05, | |
| "loss": 0.6747, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 4.824561403508772, | |
| "grad_norm": 27.501550674438477, | |
| "learning_rate": 2.3807339449541285e-05, | |
| "loss": 0.5231, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 4.842105263157895, | |
| "grad_norm": 3.0259737968444824, | |
| "learning_rate": 2.377981651376147e-05, | |
| "loss": 0.5482, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 4.859649122807017, | |
| "grad_norm": 22.053667068481445, | |
| "learning_rate": 2.375229357798165e-05, | |
| "loss": 0.6844, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 4.87719298245614, | |
| "grad_norm": 21.116025924682617, | |
| "learning_rate": 2.3724770642201836e-05, | |
| "loss": 0.5924, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 4.894736842105263, | |
| "grad_norm": 45.17483901977539, | |
| "learning_rate": 2.3697247706422018e-05, | |
| "loss": 0.9959, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 4.912280701754386, | |
| "grad_norm": 2.0751166343688965, | |
| "learning_rate": 2.3669724770642204e-05, | |
| "loss": 0.4215, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 4.9298245614035086, | |
| "grad_norm": 17.698915481567383, | |
| "learning_rate": 2.3642201834862386e-05, | |
| "loss": 0.8212, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 4.947368421052632, | |
| "grad_norm": 11.810864448547363, | |
| "learning_rate": 2.361467889908257e-05, | |
| "loss": 0.7421, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 4.964912280701754, | |
| "grad_norm": 25.751745223999023, | |
| "learning_rate": 2.3587155963302755e-05, | |
| "loss": 0.8161, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 4.982456140350877, | |
| "grad_norm": 9.799457550048828, | |
| "learning_rate": 2.3559633027522937e-05, | |
| "loss": 0.6857, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 10.592196464538574, | |
| "learning_rate": 2.353211009174312e-05, | |
| "loss": 0.8668, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.7011494252873564, | |
| "eval_loss": 0.8508906364440918, | |
| "eval_roc_auc": 0.8698378210083979, | |
| "eval_runtime": 9.9199, | |
| "eval_samples_per_second": 26.311, | |
| "eval_steps_per_second": 1.714, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 5.017543859649122, | |
| "grad_norm": 29.6278133392334, | |
| "learning_rate": 2.3504587155963302e-05, | |
| "loss": 0.8195, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 5.035087719298246, | |
| "grad_norm": 8.088125228881836, | |
| "learning_rate": 2.3477064220183488e-05, | |
| "loss": 0.4163, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 5.052631578947368, | |
| "grad_norm": 2.1910946369171143, | |
| "learning_rate": 2.344954128440367e-05, | |
| "loss": 0.4273, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 5.0701754385964914, | |
| "grad_norm": 27.934329986572266, | |
| "learning_rate": 2.3422018348623853e-05, | |
| "loss": 0.7804, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 5.087719298245614, | |
| "grad_norm": 28.2197265625, | |
| "learning_rate": 2.3394495412844035e-05, | |
| "loss": 0.5328, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 5.105263157894737, | |
| "grad_norm": 20.080032348632812, | |
| "learning_rate": 2.336697247706422e-05, | |
| "loss": 0.9274, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 5.12280701754386, | |
| "grad_norm": 8.307439804077148, | |
| "learning_rate": 2.3339449541284407e-05, | |
| "loss": 0.7575, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 5.140350877192983, | |
| "grad_norm": 15.012224197387695, | |
| "learning_rate": 2.3311926605504586e-05, | |
| "loss": 0.6881, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 5.157894736842105, | |
| "grad_norm": 19.6754093170166, | |
| "learning_rate": 2.3284403669724772e-05, | |
| "loss": 0.5243, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 5.175438596491228, | |
| "grad_norm": 2.4565722942352295, | |
| "learning_rate": 2.3256880733944955e-05, | |
| "loss": 0.5858, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 5.192982456140351, | |
| "grad_norm": 12.843107223510742, | |
| "learning_rate": 2.322935779816514e-05, | |
| "loss": 0.7095, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 5.2105263157894735, | |
| "grad_norm": 3.9343743324279785, | |
| "learning_rate": 2.320183486238532e-05, | |
| "loss": 0.6592, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 5.228070175438597, | |
| "grad_norm": 21.421518325805664, | |
| "learning_rate": 2.3174311926605505e-05, | |
| "loss": 0.6792, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 5.245614035087719, | |
| "grad_norm": 29.099624633789062, | |
| "learning_rate": 2.3146788990825688e-05, | |
| "loss": 0.8901, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 5.2631578947368425, | |
| "grad_norm": 28.176443099975586, | |
| "learning_rate": 2.3119266055045874e-05, | |
| "loss": 0.677, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 5.280701754385965, | |
| "grad_norm": 49.84172439575195, | |
| "learning_rate": 2.3091743119266053e-05, | |
| "loss": 0.6627, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 5.298245614035087, | |
| "grad_norm": 11.457085609436035, | |
| "learning_rate": 2.306422018348624e-05, | |
| "loss": 0.3993, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 5.315789473684211, | |
| "grad_norm": 187.01239013671875, | |
| "learning_rate": 2.3036697247706425e-05, | |
| "loss": 0.6346, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 5.333333333333333, | |
| "grad_norm": 3.9368183612823486, | |
| "learning_rate": 2.3009174311926607e-05, | |
| "loss": 0.8296, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 5.350877192982456, | |
| "grad_norm": 100.2872085571289, | |
| "learning_rate": 2.298165137614679e-05, | |
| "loss": 0.8781, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 5.368421052631579, | |
| "grad_norm": 24.209882736206055, | |
| "learning_rate": 2.2954128440366972e-05, | |
| "loss": 0.5137, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 5.385964912280702, | |
| "grad_norm": 3.6952366828918457, | |
| "learning_rate": 2.2926605504587158e-05, | |
| "loss": 0.6469, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 5.4035087719298245, | |
| "grad_norm": 29.038558959960938, | |
| "learning_rate": 2.289908256880734e-05, | |
| "loss": 0.8628, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 5.421052631578947, | |
| "grad_norm": 40.4047737121582, | |
| "learning_rate": 2.2871559633027523e-05, | |
| "loss": 1.065, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 5.43859649122807, | |
| "grad_norm": 43.313846588134766, | |
| "learning_rate": 2.2844036697247705e-05, | |
| "loss": 0.5773, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 5.456140350877193, | |
| "grad_norm": 7.541952133178711, | |
| "learning_rate": 2.281651376146789e-05, | |
| "loss": 0.7124, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 5.473684210526316, | |
| "grad_norm": 9.793838500976562, | |
| "learning_rate": 2.2788990825688074e-05, | |
| "loss": 0.5441, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 5.491228070175438, | |
| "grad_norm": 17.21137809753418, | |
| "learning_rate": 2.2761467889908256e-05, | |
| "loss": 0.5927, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 5.508771929824562, | |
| "grad_norm": 23.36025047302246, | |
| "learning_rate": 2.2733944954128442e-05, | |
| "loss": 0.8349, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 5.526315789473684, | |
| "grad_norm": 10.170068740844727, | |
| "learning_rate": 2.2706422018348624e-05, | |
| "loss": 0.7946, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 5.543859649122807, | |
| "grad_norm": 7.996306419372559, | |
| "learning_rate": 2.267889908256881e-05, | |
| "loss": 0.7086, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 5.56140350877193, | |
| "grad_norm": 4.8672027587890625, | |
| "learning_rate": 2.265137614678899e-05, | |
| "loss": 0.4384, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 5.578947368421053, | |
| "grad_norm": 14.543272972106934, | |
| "learning_rate": 2.2623853211009175e-05, | |
| "loss": 0.9153, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 5.5964912280701755, | |
| "grad_norm": 28.248517990112305, | |
| "learning_rate": 2.2596330275229358e-05, | |
| "loss": 0.4361, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 5.614035087719298, | |
| "grad_norm": 10.02695083618164, | |
| "learning_rate": 2.2568807339449544e-05, | |
| "loss": 0.5584, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 5.631578947368421, | |
| "grad_norm": 255.10897827148438, | |
| "learning_rate": 2.2541284403669723e-05, | |
| "loss": 0.812, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 5.649122807017544, | |
| "grad_norm": 29.070514678955078, | |
| "learning_rate": 2.251376146788991e-05, | |
| "loss": 0.7201, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 5.666666666666667, | |
| "grad_norm": 70.94801330566406, | |
| "learning_rate": 2.2486238532110094e-05, | |
| "loss": 0.7983, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 5.684210526315789, | |
| "grad_norm": 9.982966423034668, | |
| "learning_rate": 2.2458715596330277e-05, | |
| "loss": 0.3869, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 5.701754385964913, | |
| "grad_norm": 44.79756546020508, | |
| "learning_rate": 2.243119266055046e-05, | |
| "loss": 0.5092, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 5.719298245614035, | |
| "grad_norm": 14.900089263916016, | |
| "learning_rate": 2.2403669724770642e-05, | |
| "loss": 0.7798, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 5.7368421052631575, | |
| "grad_norm": 5.473045825958252, | |
| "learning_rate": 2.2376146788990828e-05, | |
| "loss": 0.527, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 5.754385964912281, | |
| "grad_norm": 5.292675495147705, | |
| "learning_rate": 2.234862385321101e-05, | |
| "loss": 0.599, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 5.771929824561403, | |
| "grad_norm": 7.409575462341309, | |
| "learning_rate": 2.2321100917431193e-05, | |
| "loss": 0.6566, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 5.7894736842105265, | |
| "grad_norm": 16.337003707885742, | |
| "learning_rate": 2.2293577981651375e-05, | |
| "loss": 0.421, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 5.807017543859649, | |
| "grad_norm": 10.473221778869629, | |
| "learning_rate": 2.226605504587156e-05, | |
| "loss": 0.6791, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 5.824561403508772, | |
| "grad_norm": 24.943050384521484, | |
| "learning_rate": 2.2238532110091743e-05, | |
| "loss": 1.0399, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 5.842105263157895, | |
| "grad_norm": 3.1360652446746826, | |
| "learning_rate": 2.2211009174311926e-05, | |
| "loss": 0.4585, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 5.859649122807017, | |
| "grad_norm": 5.742803573608398, | |
| "learning_rate": 2.2183486238532112e-05, | |
| "loss": 0.7424, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 5.87719298245614, | |
| "grad_norm": 10.02749252319336, | |
| "learning_rate": 2.2155963302752294e-05, | |
| "loss": 0.5505, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 5.894736842105263, | |
| "grad_norm": 11.549918174743652, | |
| "learning_rate": 2.212844036697248e-05, | |
| "loss": 0.7165, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 5.912280701754386, | |
| "grad_norm": 11.704695701599121, | |
| "learning_rate": 2.210091743119266e-05, | |
| "loss": 0.5897, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 5.9298245614035086, | |
| "grad_norm": 8.045721054077148, | |
| "learning_rate": 2.2073394495412845e-05, | |
| "loss": 0.5606, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 5.947368421052632, | |
| "grad_norm": 6.550736904144287, | |
| "learning_rate": 2.2045871559633028e-05, | |
| "loss": 0.4498, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 5.964912280701754, | |
| "grad_norm": 7.8976335525512695, | |
| "learning_rate": 2.2018348623853213e-05, | |
| "loss": 0.6578, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 5.982456140350877, | |
| "grad_norm": 19.857566833496094, | |
| "learning_rate": 2.1990825688073392e-05, | |
| "loss": 0.6241, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 389.2672119140625, | |
| "learning_rate": 2.196330275229358e-05, | |
| "loss": 0.5834, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.6436781609195402, | |
| "eval_loss": 0.9111022353172302, | |
| "eval_roc_auc": 0.8661937516704385, | |
| "eval_runtime": 8.4523, | |
| "eval_samples_per_second": 30.879, | |
| "eval_steps_per_second": 2.011, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 6.017543859649122, | |
| "grad_norm": 6.004997253417969, | |
| "learning_rate": 2.193577981651376e-05, | |
| "loss": 0.4383, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 6.035087719298246, | |
| "grad_norm": 32.85868453979492, | |
| "learning_rate": 2.1908256880733947e-05, | |
| "loss": 0.8127, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 6.052631578947368, | |
| "grad_norm": 22.06620216369629, | |
| "learning_rate": 2.188073394495413e-05, | |
| "loss": 0.5588, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 6.0701754385964914, | |
| "grad_norm": 12.496269226074219, | |
| "learning_rate": 2.185321100917431e-05, | |
| "loss": 0.7202, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 6.087719298245614, | |
| "grad_norm": 39.94806671142578, | |
| "learning_rate": 2.1825688073394497e-05, | |
| "loss": 0.4362, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 6.105263157894737, | |
| "grad_norm": 5.034066677093506, | |
| "learning_rate": 2.179816513761468e-05, | |
| "loss": 0.6235, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 6.12280701754386, | |
| "grad_norm": 98.4699935913086, | |
| "learning_rate": 2.1770642201834862e-05, | |
| "loss": 0.8505, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 6.140350877192983, | |
| "grad_norm": 22.5294132232666, | |
| "learning_rate": 2.1743119266055045e-05, | |
| "loss": 0.6741, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 6.157894736842105, | |
| "grad_norm": 33.31329345703125, | |
| "learning_rate": 2.171559633027523e-05, | |
| "loss": 0.6783, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 6.175438596491228, | |
| "grad_norm": 2.9709410667419434, | |
| "learning_rate": 2.1688073394495413e-05, | |
| "loss": 0.3269, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 6.192982456140351, | |
| "grad_norm": 14.349902153015137, | |
| "learning_rate": 2.1660550458715596e-05, | |
| "loss": 0.5928, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 6.2105263157894735, | |
| "grad_norm": 6.400623321533203, | |
| "learning_rate": 2.163302752293578e-05, | |
| "loss": 0.5708, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 6.228070175438597, | |
| "grad_norm": 40.32659912109375, | |
| "learning_rate": 2.1605504587155964e-05, | |
| "loss": 0.4459, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 6.245614035087719, | |
| "grad_norm": 16.54581069946289, | |
| "learning_rate": 2.157798165137615e-05, | |
| "loss": 1.0475, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 6.2631578947368425, | |
| "grad_norm": 7.048498630523682, | |
| "learning_rate": 2.155045871559633e-05, | |
| "loss": 0.6128, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 6.280701754385965, | |
| "grad_norm": 18.782760620117188, | |
| "learning_rate": 2.1522935779816515e-05, | |
| "loss": 0.6483, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 6.298245614035087, | |
| "grad_norm": 13.411852836608887, | |
| "learning_rate": 2.1495412844036697e-05, | |
| "loss": 0.6957, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 6.315789473684211, | |
| "grad_norm": 3.7218034267425537, | |
| "learning_rate": 2.1467889908256883e-05, | |
| "loss": 0.268, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 6.333333333333333, | |
| "grad_norm": 5.883011817932129, | |
| "learning_rate": 2.1440366972477062e-05, | |
| "loss": 0.5415, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 6.350877192982456, | |
| "grad_norm": 6.700387477874756, | |
| "learning_rate": 2.1412844036697248e-05, | |
| "loss": 0.4707, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 6.368421052631579, | |
| "grad_norm": 12.57919692993164, | |
| "learning_rate": 2.138532110091743e-05, | |
| "loss": 0.5245, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 6.385964912280702, | |
| "grad_norm": 23.14158058166504, | |
| "learning_rate": 2.1357798165137616e-05, | |
| "loss": 0.4275, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 6.4035087719298245, | |
| "grad_norm": 20.404285430908203, | |
| "learning_rate": 2.13302752293578e-05, | |
| "loss": 0.7325, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 6.421052631578947, | |
| "grad_norm": 7.9764580726623535, | |
| "learning_rate": 2.130275229357798e-05, | |
| "loss": 0.469, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 6.43859649122807, | |
| "grad_norm": 11.24958324432373, | |
| "learning_rate": 2.1275229357798167e-05, | |
| "loss": 0.7704, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 6.456140350877193, | |
| "grad_norm": 3.0611751079559326, | |
| "learning_rate": 2.124770642201835e-05, | |
| "loss": 0.3323, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 6.473684210526316, | |
| "grad_norm": 40.933162689208984, | |
| "learning_rate": 2.1220183486238532e-05, | |
| "loss": 0.5066, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 6.491228070175438, | |
| "grad_norm": 9.126625061035156, | |
| "learning_rate": 2.1192660550458715e-05, | |
| "loss": 0.4907, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 6.508771929824562, | |
| "grad_norm": 9.465023040771484, | |
| "learning_rate": 2.11651376146789e-05, | |
| "loss": 0.5455, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 6.526315789473684, | |
| "grad_norm": 24.881996154785156, | |
| "learning_rate": 2.1137614678899083e-05, | |
| "loss": 0.6284, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 6.543859649122807, | |
| "grad_norm": 32.151466369628906, | |
| "learning_rate": 2.1110091743119266e-05, | |
| "loss": 0.4083, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 6.56140350877193, | |
| "grad_norm": 12.968864440917969, | |
| "learning_rate": 2.1082568807339448e-05, | |
| "loss": 0.5154, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 6.578947368421053, | |
| "grad_norm": 64.264892578125, | |
| "learning_rate": 2.1055045871559634e-05, | |
| "loss": 0.2809, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 6.5964912280701755, | |
| "grad_norm": 19.623029708862305, | |
| "learning_rate": 2.102752293577982e-05, | |
| "loss": 0.7864, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 6.614035087719298, | |
| "grad_norm": 7.40114688873291, | |
| "learning_rate": 2.1e-05, | |
| "loss": 0.5327, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 6.631578947368421, | |
| "grad_norm": 15.018710136413574, | |
| "learning_rate": 2.0972477064220185e-05, | |
| "loss": 0.5948, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 6.649122807017544, | |
| "grad_norm": 6.5761847496032715, | |
| "learning_rate": 2.0944954128440367e-05, | |
| "loss": 0.5606, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 6.666666666666667, | |
| "grad_norm": 21.176044464111328, | |
| "learning_rate": 2.0917431192660553e-05, | |
| "loss": 0.8495, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 6.684210526315789, | |
| "grad_norm": 15.794927597045898, | |
| "learning_rate": 2.0889908256880732e-05, | |
| "loss": 0.2922, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 6.701754385964913, | |
| "grad_norm": 21.976320266723633, | |
| "learning_rate": 2.0862385321100918e-05, | |
| "loss": 0.4193, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 6.719298245614035, | |
| "grad_norm": 127.4170150756836, | |
| "learning_rate": 2.08348623853211e-05, | |
| "loss": 0.7539, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 6.7368421052631575, | |
| "grad_norm": 23.519792556762695, | |
| "learning_rate": 2.0807339449541286e-05, | |
| "loss": 0.8324, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 6.754385964912281, | |
| "grad_norm": 34.533729553222656, | |
| "learning_rate": 2.0779816513761465e-05, | |
| "loss": 0.635, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 6.771929824561403, | |
| "grad_norm": 48.43922805786133, | |
| "learning_rate": 2.075229357798165e-05, | |
| "loss": 0.7622, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 6.7894736842105265, | |
| "grad_norm": 9.6075439453125, | |
| "learning_rate": 2.0724770642201837e-05, | |
| "loss": 0.4885, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 6.807017543859649, | |
| "grad_norm": 25.955398559570312, | |
| "learning_rate": 2.069724770642202e-05, | |
| "loss": 0.2685, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 6.824561403508772, | |
| "grad_norm": 17.847700119018555, | |
| "learning_rate": 2.0669724770642202e-05, | |
| "loss": 0.3958, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 6.842105263157895, | |
| "grad_norm": 13.096657752990723, | |
| "learning_rate": 2.0642201834862385e-05, | |
| "loss": 0.9942, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 6.859649122807017, | |
| "grad_norm": 11.541616439819336, | |
| "learning_rate": 2.061467889908257e-05, | |
| "loss": 0.5704, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 6.87719298245614, | |
| "grad_norm": 3.565450668334961, | |
| "learning_rate": 2.0587155963302753e-05, | |
| "loss": 0.6809, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 6.894736842105263, | |
| "grad_norm": 27.333837509155273, | |
| "learning_rate": 2.0559633027522935e-05, | |
| "loss": 0.6584, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 6.912280701754386, | |
| "grad_norm": 15.244169235229492, | |
| "learning_rate": 2.0532110091743118e-05, | |
| "loss": 0.6753, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 6.9298245614035086, | |
| "grad_norm": 18.803529739379883, | |
| "learning_rate": 2.0504587155963304e-05, | |
| "loss": 0.5975, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 6.947368421052632, | |
| "grad_norm": 7.317320346832275, | |
| "learning_rate": 2.047706422018349e-05, | |
| "loss": 0.9889, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 6.964912280701754, | |
| "grad_norm": 19.616382598876953, | |
| "learning_rate": 2.044954128440367e-05, | |
| "loss": 0.6709, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 6.982456140350877, | |
| "grad_norm": 46.652469635009766, | |
| "learning_rate": 2.0422018348623855e-05, | |
| "loss": 0.5268, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "grad_norm": 11.762057304382324, | |
| "learning_rate": 2.0394495412844037e-05, | |
| "loss": 0.9838, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.6819923371647509, | |
| "eval_loss": 0.8196430802345276, | |
| "eval_roc_auc": 0.8813437942252584, | |
| "eval_runtime": 9.6385, | |
| "eval_samples_per_second": 27.079, | |
| "eval_steps_per_second": 1.764, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 7.017543859649122, | |
| "grad_norm": 14.474690437316895, | |
| "learning_rate": 2.0366972477064223e-05, | |
| "loss": 0.3408, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 7.035087719298246, | |
| "grad_norm": 19.78782081604004, | |
| "learning_rate": 2.0339449541284402e-05, | |
| "loss": 0.4244, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 7.052631578947368, | |
| "grad_norm": 10.25345230102539, | |
| "learning_rate": 2.0311926605504588e-05, | |
| "loss": 0.6707, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 7.0701754385964914, | |
| "grad_norm": 2.85701060295105, | |
| "learning_rate": 2.028440366972477e-05, | |
| "loss": 0.2397, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 7.087719298245614, | |
| "grad_norm": 6.676547527313232, | |
| "learning_rate": 2.0256880733944956e-05, | |
| "loss": 0.3686, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 7.105263157894737, | |
| "grad_norm": 165.5542755126953, | |
| "learning_rate": 2.0229357798165135e-05, | |
| "loss": 0.5689, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 7.12280701754386, | |
| "grad_norm": 14.226078033447266, | |
| "learning_rate": 2.020183486238532e-05, | |
| "loss": 0.3423, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 7.140350877192983, | |
| "grad_norm": 5.127176761627197, | |
| "learning_rate": 2.0174311926605507e-05, | |
| "loss": 0.6358, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 7.157894736842105, | |
| "grad_norm": 21.368282318115234, | |
| "learning_rate": 2.014678899082569e-05, | |
| "loss": 0.5103, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 7.175438596491228, | |
| "grad_norm": 3.635981321334839, | |
| "learning_rate": 2.0119266055045872e-05, | |
| "loss": 0.3426, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 7.192982456140351, | |
| "grad_norm": 8.354199409484863, | |
| "learning_rate": 2.0091743119266054e-05, | |
| "loss": 0.4433, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 7.2105263157894735, | |
| "grad_norm": 125.07445526123047, | |
| "learning_rate": 2.006422018348624e-05, | |
| "loss": 0.5754, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 7.228070175438597, | |
| "grad_norm": 18.86847496032715, | |
| "learning_rate": 2.0036697247706423e-05, | |
| "loss": 0.6004, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 7.245614035087719, | |
| "grad_norm": 13.290031433105469, | |
| "learning_rate": 2.0009174311926605e-05, | |
| "loss": 0.6924, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 7.2631578947368425, | |
| "grad_norm": 7.521588325500488, | |
| "learning_rate": 1.9981651376146788e-05, | |
| "loss": 0.4925, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 7.280701754385965, | |
| "grad_norm": 14.16098690032959, | |
| "learning_rate": 1.9954128440366974e-05, | |
| "loss": 0.6182, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 7.298245614035087, | |
| "grad_norm": 12.972576141357422, | |
| "learning_rate": 1.9926605504587156e-05, | |
| "loss": 0.4994, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 7.315789473684211, | |
| "grad_norm": 7.475390911102295, | |
| "learning_rate": 1.989908256880734e-05, | |
| "loss": 0.5975, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 7.333333333333333, | |
| "grad_norm": 19.82378387451172, | |
| "learning_rate": 1.9871559633027524e-05, | |
| "loss": 0.4584, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 7.350877192982456, | |
| "grad_norm": 7.333564281463623, | |
| "learning_rate": 1.9844036697247707e-05, | |
| "loss": 0.5651, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 7.368421052631579, | |
| "grad_norm": 12.829773902893066, | |
| "learning_rate": 1.9816513761467893e-05, | |
| "loss": 0.4198, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 7.385964912280702, | |
| "grad_norm": 3.876220464706421, | |
| "learning_rate": 1.9788990825688072e-05, | |
| "loss": 0.2843, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 7.4035087719298245, | |
| "grad_norm": 10.361215591430664, | |
| "learning_rate": 1.9761467889908258e-05, | |
| "loss": 0.4935, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 7.421052631578947, | |
| "grad_norm": 59.4860954284668, | |
| "learning_rate": 1.973394495412844e-05, | |
| "loss": 0.5763, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 7.43859649122807, | |
| "grad_norm": 10.748738288879395, | |
| "learning_rate": 1.9706422018348626e-05, | |
| "loss": 0.7249, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 7.456140350877193, | |
| "grad_norm": 19.900938034057617, | |
| "learning_rate": 1.9678899082568805e-05, | |
| "loss": 0.3448, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 7.473684210526316, | |
| "grad_norm": 9.298110961914062, | |
| "learning_rate": 1.965137614678899e-05, | |
| "loss": 0.424, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 7.491228070175438, | |
| "grad_norm": 6.225700855255127, | |
| "learning_rate": 1.9623853211009173e-05, | |
| "loss": 0.5622, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 7.508771929824562, | |
| "grad_norm": 7.592733860015869, | |
| "learning_rate": 1.959633027522936e-05, | |
| "loss": 0.327, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 7.526315789473684, | |
| "grad_norm": 15.926929473876953, | |
| "learning_rate": 1.9568807339449542e-05, | |
| "loss": 0.5397, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 7.543859649122807, | |
| "grad_norm": 5.76393461227417, | |
| "learning_rate": 1.9541284403669724e-05, | |
| "loss": 0.6615, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 7.56140350877193, | |
| "grad_norm": 9.405485153198242, | |
| "learning_rate": 1.951376146788991e-05, | |
| "loss": 0.5491, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 7.578947368421053, | |
| "grad_norm": 6.3222479820251465, | |
| "learning_rate": 1.9486238532110093e-05, | |
| "loss": 0.5539, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 7.5964912280701755, | |
| "grad_norm": 15.603589057922363, | |
| "learning_rate": 1.9458715596330275e-05, | |
| "loss": 0.352, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 7.614035087719298, | |
| "grad_norm": 54.22720718383789, | |
| "learning_rate": 1.9431192660550458e-05, | |
| "loss": 0.5936, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 7.631578947368421, | |
| "grad_norm": 7.901265621185303, | |
| "learning_rate": 1.9403669724770643e-05, | |
| "loss": 0.6319, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 7.649122807017544, | |
| "grad_norm": 16.199270248413086, | |
| "learning_rate": 1.9376146788990826e-05, | |
| "loss": 0.5468, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 7.666666666666667, | |
| "grad_norm": 84.76512908935547, | |
| "learning_rate": 1.934862385321101e-05, | |
| "loss": 0.5745, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 7.684210526315789, | |
| "grad_norm": 8.624265670776367, | |
| "learning_rate": 1.9321100917431194e-05, | |
| "loss": 0.5141, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 7.701754385964913, | |
| "grad_norm": 4.701018333435059, | |
| "learning_rate": 1.9293577981651377e-05, | |
| "loss": 0.2706, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 7.719298245614035, | |
| "grad_norm": 3.979341745376587, | |
| "learning_rate": 1.9266055045871563e-05, | |
| "loss": 0.4677, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 7.7368421052631575, | |
| "grad_norm": 10.237481117248535, | |
| "learning_rate": 1.923853211009174e-05, | |
| "loss": 0.6246, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 7.754385964912281, | |
| "grad_norm": 16.742145538330078, | |
| "learning_rate": 1.9211009174311927e-05, | |
| "loss": 0.5573, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 7.771929824561403, | |
| "grad_norm": 7.850118637084961, | |
| "learning_rate": 1.918348623853211e-05, | |
| "loss": 0.3705, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 7.7894736842105265, | |
| "grad_norm": 7.423269271850586, | |
| "learning_rate": 1.9155963302752296e-05, | |
| "loss": 0.5847, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 7.807017543859649, | |
| "grad_norm": 3.7411139011383057, | |
| "learning_rate": 1.9128440366972475e-05, | |
| "loss": 0.3861, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 7.824561403508772, | |
| "grad_norm": 131.05946350097656, | |
| "learning_rate": 1.910091743119266e-05, | |
| "loss": 0.5242, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 7.842105263157895, | |
| "grad_norm": 12.686647415161133, | |
| "learning_rate": 1.9073394495412843e-05, | |
| "loss": 0.2422, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 7.859649122807017, | |
| "grad_norm": 7.073160171508789, | |
| "learning_rate": 1.904587155963303e-05, | |
| "loss": 0.2323, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 7.87719298245614, | |
| "grad_norm": 4.771604061126709, | |
| "learning_rate": 1.901834862385321e-05, | |
| "loss": 0.2788, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 7.894736842105263, | |
| "grad_norm": 11.123143196105957, | |
| "learning_rate": 1.8990825688073394e-05, | |
| "loss": 0.3865, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 7.912280701754386, | |
| "grad_norm": 19.361055374145508, | |
| "learning_rate": 1.896330275229358e-05, | |
| "loss": 0.4511, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 7.9298245614035086, | |
| "grad_norm": 8.789932250976562, | |
| "learning_rate": 1.8935779816513762e-05, | |
| "loss": 0.3365, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 7.947368421052632, | |
| "grad_norm": 12.437773704528809, | |
| "learning_rate": 1.8908256880733945e-05, | |
| "loss": 0.3662, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 7.964912280701754, | |
| "grad_norm": 17.096485137939453, | |
| "learning_rate": 1.8880733944954127e-05, | |
| "loss": 0.5647, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 7.982456140350877, | |
| "grad_norm": 20.25951385498047, | |
| "learning_rate": 1.8853211009174313e-05, | |
| "loss": 0.4497, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 14.128129005432129, | |
| "learning_rate": 1.8825688073394496e-05, | |
| "loss": 0.3699, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.6743295019157088, | |
| "eval_loss": 0.9114387035369873, | |
| "eval_roc_auc": 0.8783069858814981, | |
| "eval_runtime": 9.7288, | |
| "eval_samples_per_second": 26.827, | |
| "eval_steps_per_second": 1.747, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 8.017543859649123, | |
| "grad_norm": 14.421273231506348, | |
| "learning_rate": 1.8798165137614678e-05, | |
| "loss": 0.4197, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 8.035087719298245, | |
| "grad_norm": 7.326334476470947, | |
| "learning_rate": 1.877064220183486e-05, | |
| "loss": 0.3367, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 8.052631578947368, | |
| "grad_norm": 9.997421264648438, | |
| "learning_rate": 1.8743119266055047e-05, | |
| "loss": 0.4693, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 8.070175438596491, | |
| "grad_norm": 16.052030563354492, | |
| "learning_rate": 1.8715596330275232e-05, | |
| "loss": 0.3885, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 8.087719298245615, | |
| "grad_norm": 5.081548690795898, | |
| "learning_rate": 1.868807339449541e-05, | |
| "loss": 0.3361, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 8.105263157894736, | |
| "grad_norm": 12.288408279418945, | |
| "learning_rate": 1.8660550458715597e-05, | |
| "loss": 0.4407, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 8.12280701754386, | |
| "grad_norm": 20.765512466430664, | |
| "learning_rate": 1.863302752293578e-05, | |
| "loss": 0.3498, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 8.140350877192983, | |
| "grad_norm": 16.51128578186035, | |
| "learning_rate": 1.8605504587155966e-05, | |
| "loss": 0.5519, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 8.157894736842104, | |
| "grad_norm": 9.807088851928711, | |
| "learning_rate": 1.8577981651376145e-05, | |
| "loss": 0.4753, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 8.175438596491228, | |
| "grad_norm": 27.620849609375, | |
| "learning_rate": 1.855045871559633e-05, | |
| "loss": 0.4292, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 8.192982456140351, | |
| "grad_norm": 3.228712320327759, | |
| "learning_rate": 1.8522935779816513e-05, | |
| "loss": 0.2354, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 8.210526315789474, | |
| "grad_norm": 19.046974182128906, | |
| "learning_rate": 1.84954128440367e-05, | |
| "loss": 0.7982, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 8.228070175438596, | |
| "grad_norm": 35.34560775756836, | |
| "learning_rate": 1.8467889908256878e-05, | |
| "loss": 0.6045, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 8.24561403508772, | |
| "grad_norm": 19.82710838317871, | |
| "learning_rate": 1.8440366972477064e-05, | |
| "loss": 0.2748, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 8.263157894736842, | |
| "grad_norm": 107.26939392089844, | |
| "learning_rate": 1.841284403669725e-05, | |
| "loss": 0.5046, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 8.280701754385966, | |
| "grad_norm": 9.757340431213379, | |
| "learning_rate": 1.8385321100917432e-05, | |
| "loss": 0.2989, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 8.298245614035087, | |
| "grad_norm": 42.89109802246094, | |
| "learning_rate": 1.8357798165137615e-05, | |
| "loss": 0.4882, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 8.31578947368421, | |
| "grad_norm": 27.338111877441406, | |
| "learning_rate": 1.8330275229357797e-05, | |
| "loss": 0.4523, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "grad_norm": 3.047417163848877, | |
| "learning_rate": 1.8302752293577983e-05, | |
| "loss": 0.2369, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 8.350877192982455, | |
| "grad_norm": 27.468341827392578, | |
| "learning_rate": 1.8275229357798166e-05, | |
| "loss": 0.4376, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 8.368421052631579, | |
| "grad_norm": 4.080908298492432, | |
| "learning_rate": 1.8247706422018348e-05, | |
| "loss": 0.3097, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 8.385964912280702, | |
| "grad_norm": 15.838717460632324, | |
| "learning_rate": 1.822018348623853e-05, | |
| "loss": 0.424, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 8.403508771929825, | |
| "grad_norm": 9.841987609863281, | |
| "learning_rate": 1.8192660550458716e-05, | |
| "loss": 0.5209, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 8.421052631578947, | |
| "grad_norm": 11.688602447509766, | |
| "learning_rate": 1.8165137614678902e-05, | |
| "loss": 0.2871, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 8.43859649122807, | |
| "grad_norm": 42.60127639770508, | |
| "learning_rate": 1.813761467889908e-05, | |
| "loss": 0.7017, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 8.456140350877194, | |
| "grad_norm": 21.921663284301758, | |
| "learning_rate": 1.8110091743119267e-05, | |
| "loss": 0.6333, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 8.473684210526315, | |
| "grad_norm": 14.230140686035156, | |
| "learning_rate": 1.808256880733945e-05, | |
| "loss": 0.4494, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 8.491228070175438, | |
| "grad_norm": 4.411128997802734, | |
| "learning_rate": 1.8055045871559636e-05, | |
| "loss": 0.2, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 8.508771929824562, | |
| "grad_norm": 4.632619857788086, | |
| "learning_rate": 1.8027522935779815e-05, | |
| "loss": 0.3362, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 8.526315789473685, | |
| "grad_norm": 11.009140014648438, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.2433, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 8.543859649122806, | |
| "grad_norm": 35.23042297363281, | |
| "learning_rate": 1.7972477064220183e-05, | |
| "loss": 0.4174, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 8.56140350877193, | |
| "grad_norm": 6.151736259460449, | |
| "learning_rate": 1.794495412844037e-05, | |
| "loss": 0.3024, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 8.578947368421053, | |
| "grad_norm": 25.69590950012207, | |
| "learning_rate": 1.7917431192660548e-05, | |
| "loss": 0.7038, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 8.596491228070175, | |
| "grad_norm": 20.76173973083496, | |
| "learning_rate": 1.7889908256880734e-05, | |
| "loss": 0.6203, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 8.614035087719298, | |
| "grad_norm": 17.6284122467041, | |
| "learning_rate": 1.786238532110092e-05, | |
| "loss": 0.3738, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 8.631578947368421, | |
| "grad_norm": 6.477032661437988, | |
| "learning_rate": 1.7834862385321102e-05, | |
| "loss": 0.4933, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 8.649122807017545, | |
| "grad_norm": 11.045804023742676, | |
| "learning_rate": 1.7807339449541285e-05, | |
| "loss": 0.4582, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 8.666666666666666, | |
| "grad_norm": 7.45139741897583, | |
| "learning_rate": 1.7779816513761467e-05, | |
| "loss": 0.2789, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 8.68421052631579, | |
| "grad_norm": 6.092661380767822, | |
| "learning_rate": 1.7752293577981653e-05, | |
| "loss": 0.5297, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 8.701754385964913, | |
| "grad_norm": 9.822498321533203, | |
| "learning_rate": 1.7724770642201835e-05, | |
| "loss": 0.4994, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 8.719298245614034, | |
| "grad_norm": 25.944725036621094, | |
| "learning_rate": 1.7697247706422018e-05, | |
| "loss": 0.5637, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 8.736842105263158, | |
| "grad_norm": 7.313636302947998, | |
| "learning_rate": 1.76697247706422e-05, | |
| "loss": 0.3809, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 8.75438596491228, | |
| "grad_norm": 28.630786895751953, | |
| "learning_rate": 1.7642201834862386e-05, | |
| "loss": 0.2553, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 8.771929824561404, | |
| "grad_norm": 13.779499053955078, | |
| "learning_rate": 1.761467889908257e-05, | |
| "loss": 0.6088, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 8.789473684210526, | |
| "grad_norm": 13.866922378540039, | |
| "learning_rate": 1.758715596330275e-05, | |
| "loss": 0.3418, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 8.807017543859649, | |
| "grad_norm": 9.200101852416992, | |
| "learning_rate": 1.7559633027522937e-05, | |
| "loss": 0.5261, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 8.824561403508772, | |
| "grad_norm": 4.6678853034973145, | |
| "learning_rate": 1.753211009174312e-05, | |
| "loss": 0.2657, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 8.842105263157894, | |
| "grad_norm": 13.394104957580566, | |
| "learning_rate": 1.7504587155963305e-05, | |
| "loss": 0.4874, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 8.859649122807017, | |
| "grad_norm": 8.163538932800293, | |
| "learning_rate": 1.7477064220183484e-05, | |
| "loss": 0.4425, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 8.87719298245614, | |
| "grad_norm": 11.24685001373291, | |
| "learning_rate": 1.744954128440367e-05, | |
| "loss": 0.4143, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 8.894736842105264, | |
| "grad_norm": 15.51749324798584, | |
| "learning_rate": 1.7422018348623853e-05, | |
| "loss": 0.2075, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 8.912280701754385, | |
| "grad_norm": 18.03982162475586, | |
| "learning_rate": 1.739449541284404e-05, | |
| "loss": 0.3895, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 8.929824561403509, | |
| "grad_norm": 30.238689422607422, | |
| "learning_rate": 1.7366972477064218e-05, | |
| "loss": 0.502, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 8.947368421052632, | |
| "grad_norm": 17.355592727661133, | |
| "learning_rate": 1.7339449541284404e-05, | |
| "loss": 0.3032, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 8.964912280701755, | |
| "grad_norm": 8.36916732788086, | |
| "learning_rate": 1.7311926605504586e-05, | |
| "loss": 0.4566, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 8.982456140350877, | |
| "grad_norm": 4.937262535095215, | |
| "learning_rate": 1.7284403669724772e-05, | |
| "loss": 0.2615, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "grad_norm": 61.639869689941406, | |
| "learning_rate": 1.7256880733944954e-05, | |
| "loss": 0.6482, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.7126436781609196, | |
| "eval_loss": 0.8051581382751465, | |
| "eval_roc_auc": 0.8928142130921077, | |
| "eval_runtime": 9.5241, | |
| "eval_samples_per_second": 27.404, | |
| "eval_steps_per_second": 1.785, | |
| "step": 513 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1140, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.23308947617632e+17, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |