| { | |
| "best_metric": 0.011418873444199562, | |
| "best_model_checkpoint": "beit-banana-diseases/checkpoint-264", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 264, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 21.393463134765625, | |
| "learning_rate": 1.8518518518518519e-06, | |
| "loss": 1.9713, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 22.38655662536621, | |
| "learning_rate": 3.7037037037037037e-06, | |
| "loss": 1.9772, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 21.854021072387695, | |
| "learning_rate": 5.555555555555556e-06, | |
| "loss": 1.7078, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 24.193124771118164, | |
| "learning_rate": 7.4074074074074075e-06, | |
| "loss": 2.1819, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 20.156389236450195, | |
| "learning_rate": 9.259259259259259e-06, | |
| "loss": 1.7843, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 21.425861358642578, | |
| "learning_rate": 1.1111111111111112e-05, | |
| "loss": 1.5904, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 23.77149200439453, | |
| "learning_rate": 1.2962962962962962e-05, | |
| "loss": 1.8766, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 21.238325119018555, | |
| "learning_rate": 1.4814814814814815e-05, | |
| "loss": 1.6849, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 21.050870895385742, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 1.9811, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 21.381715774536133, | |
| "learning_rate": 1.8518518518518518e-05, | |
| "loss": 1.7029, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 23.12903594970703, | |
| "learning_rate": 2.037037037037037e-05, | |
| "loss": 1.5096, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 21.264657974243164, | |
| "learning_rate": 2.2222222222222223e-05, | |
| "loss": 1.748, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 22.664464950561523, | |
| "learning_rate": 2.4074074074074074e-05, | |
| "loss": 1.5399, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 23.9726619720459, | |
| "learning_rate": 2.5925925925925925e-05, | |
| "loss": 1.6543, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 25.409692764282227, | |
| "learning_rate": 2.777777777777778e-05, | |
| "loss": 1.4427, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 19.801799774169922, | |
| "learning_rate": 2.962962962962963e-05, | |
| "loss": 1.415, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 18.942399978637695, | |
| "learning_rate": 3.148148148148148e-05, | |
| "loss": 1.3031, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 17.01711654663086, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 1.382, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 25.648284912109375, | |
| "learning_rate": 3.518518518518519e-05, | |
| "loss": 1.5428, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 17.678979873657227, | |
| "learning_rate": 3.7037037037037037e-05, | |
| "loss": 1.1447, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 20.149110794067383, | |
| "learning_rate": 3.888888888888889e-05, | |
| "loss": 1.0376, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 22.123849868774414, | |
| "learning_rate": 4.074074074074074e-05, | |
| "loss": 1.037, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 16.500776290893555, | |
| "learning_rate": 4.259259259259259e-05, | |
| "loss": 0.7835, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 24.026493072509766, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "loss": 1.1297, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 15.726734161376953, | |
| "learning_rate": 4.62962962962963e-05, | |
| "loss": 0.872, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 27.0452823638916, | |
| "learning_rate": 4.814814814814815e-05, | |
| "loss": 1.4377, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 19.147762298583984, | |
| "learning_rate": 5e-05, | |
| "loss": 0.8577, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 18.310501098632812, | |
| "learning_rate": 4.9789029535864986e-05, | |
| "loss": 0.7565, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 39.3529052734375, | |
| "learning_rate": 4.957805907172996e-05, | |
| "loss": 1.1609, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 18.195653915405273, | |
| "learning_rate": 4.936708860759494e-05, | |
| "loss": 0.7501, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 18.973045349121094, | |
| "learning_rate": 4.9156118143459915e-05, | |
| "loss": 0.5452, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 34.03499221801758, | |
| "learning_rate": 4.89451476793249e-05, | |
| "loss": 0.7635, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 20.804288864135742, | |
| "learning_rate": 4.8734177215189874e-05, | |
| "loss": 0.5612, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 17.79737663269043, | |
| "learning_rate": 4.852320675105486e-05, | |
| "loss": 0.6733, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 16.90382194519043, | |
| "learning_rate": 4.8312236286919834e-05, | |
| "loss": 0.6375, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 10.320564270019531, | |
| "learning_rate": 4.810126582278481e-05, | |
| "loss": 0.2782, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 23.366121292114258, | |
| "learning_rate": 4.789029535864979e-05, | |
| "loss": 0.5793, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 10.37669563293457, | |
| "learning_rate": 4.767932489451477e-05, | |
| "loss": 0.2511, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 12.599074363708496, | |
| "learning_rate": 4.7468354430379746e-05, | |
| "loss": 0.3908, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 12.652225494384766, | |
| "learning_rate": 4.725738396624473e-05, | |
| "loss": 0.382, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 12.52493953704834, | |
| "learning_rate": 4.704641350210971e-05, | |
| "loss": 0.5196, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 24.851703643798828, | |
| "learning_rate": 4.683544303797468e-05, | |
| "loss": 0.759, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 20.065210342407227, | |
| "learning_rate": 4.6624472573839666e-05, | |
| "loss": 0.4902, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 6.980092525482178, | |
| "learning_rate": 4.641350210970464e-05, | |
| "loss": 0.1772, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 10.875155448913574, | |
| "learning_rate": 4.6202531645569625e-05, | |
| "loss": 0.2554, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 8.317543029785156, | |
| "learning_rate": 4.59915611814346e-05, | |
| "loss": 0.1368, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 15.295212745666504, | |
| "learning_rate": 4.5780590717299585e-05, | |
| "loss": 0.539, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 9.764492988586426, | |
| "learning_rate": 4.556962025316456e-05, | |
| "loss": 0.3554, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 18.125457763671875, | |
| "learning_rate": 4.535864978902954e-05, | |
| "loss": 0.5302, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 7.70007848739624, | |
| "learning_rate": 4.5147679324894514e-05, | |
| "loss": 0.2058, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 15.114970207214355, | |
| "learning_rate": 4.49367088607595e-05, | |
| "loss": 0.4506, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 3.854482889175415, | |
| "learning_rate": 4.4725738396624474e-05, | |
| "loss": 0.0844, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 15.301838874816895, | |
| "learning_rate": 4.451476793248946e-05, | |
| "loss": 0.443, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 12.818358421325684, | |
| "learning_rate": 4.430379746835443e-05, | |
| "loss": 0.4188, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 20.202260971069336, | |
| "learning_rate": 4.409282700421941e-05, | |
| "loss": 0.518, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 10.593685150146484, | |
| "learning_rate": 4.388185654008439e-05, | |
| "loss": 0.2015, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 6.216127872467041, | |
| "learning_rate": 4.367088607594937e-05, | |
| "loss": 0.1223, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 8.717305183410645, | |
| "learning_rate": 4.345991561181435e-05, | |
| "loss": 0.1161, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 5.678544998168945, | |
| "learning_rate": 4.324894514767933e-05, | |
| "loss": 0.1481, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 12.164253234863281, | |
| "learning_rate": 4.3037974683544305e-05, | |
| "loss": 0.3593, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 9.91303825378418, | |
| "learning_rate": 4.282700421940928e-05, | |
| "loss": 0.2332, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 11.6513090133667, | |
| "learning_rate": 4.2616033755274265e-05, | |
| "loss": 0.2646, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 10.128318786621094, | |
| "learning_rate": 4.240506329113924e-05, | |
| "loss": 0.166, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 7.344198226928711, | |
| "learning_rate": 4.2194092827004224e-05, | |
| "loss": 0.1824, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 14.342442512512207, | |
| "learning_rate": 4.19831223628692e-05, | |
| "loss": 0.2132, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 9.733878135681152, | |
| "learning_rate": 4.177215189873418e-05, | |
| "loss": 0.1968, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 6.932378768920898, | |
| "learning_rate": 4.1561181434599153e-05, | |
| "loss": 0.1487, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 1.614574909210205, | |
| "learning_rate": 4.135021097046414e-05, | |
| "loss": 0.0385, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 12.623132705688477, | |
| "learning_rate": 4.113924050632912e-05, | |
| "loss": 0.1805, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 8.182653427124023, | |
| "learning_rate": 4.0928270042194096e-05, | |
| "loss": 0.1422, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 2.2609386444091797, | |
| "learning_rate": 4.071729957805907e-05, | |
| "loss": 0.0578, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 1.1661690473556519, | |
| "learning_rate": 4.050632911392405e-05, | |
| "loss": 0.0243, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 18.96497917175293, | |
| "learning_rate": 4.029535864978903e-05, | |
| "loss": 0.3047, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 16.800785064697266, | |
| "learning_rate": 4.008438818565401e-05, | |
| "loss": 0.4561, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 25.772275924682617, | |
| "learning_rate": 3.987341772151899e-05, | |
| "loss": 0.5251, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 10.185676574707031, | |
| "learning_rate": 3.966244725738397e-05, | |
| "loss": 0.2006, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 6.298079013824463, | |
| "learning_rate": 3.945147679324895e-05, | |
| "loss": 0.0788, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 13.123534202575684, | |
| "learning_rate": 3.924050632911392e-05, | |
| "loss": 0.2158, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 4.510250568389893, | |
| "learning_rate": 3.9029535864978904e-05, | |
| "loss": 0.0525, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 8.665478706359863, | |
| "learning_rate": 3.881856540084388e-05, | |
| "loss": 0.1001, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 5.559294700622559, | |
| "learning_rate": 3.8607594936708864e-05, | |
| "loss": 0.0602, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 5.936877250671387, | |
| "learning_rate": 3.839662447257384e-05, | |
| "loss": 0.0586, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 4.139342784881592, | |
| "learning_rate": 3.8185654008438823e-05, | |
| "loss": 0.0709, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 1.722638487815857, | |
| "learning_rate": 3.79746835443038e-05, | |
| "loss": 0.0264, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 7.886575222015381, | |
| "learning_rate": 3.7763713080168776e-05, | |
| "loss": 0.1474, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 3.067516803741455, | |
| "learning_rate": 3.755274261603376e-05, | |
| "loss": 0.0431, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 7.004971027374268, | |
| "learning_rate": 3.7341772151898736e-05, | |
| "loss": 0.0518, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 42.861572265625, | |
| "learning_rate": 3.713080168776372e-05, | |
| "loss": 0.686, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.935064935064935, | |
| "eval_f1_macro": 0.9315281668222845, | |
| "eval_f1_micro": 0.935064935064935, | |
| "eval_f1_weighted": 0.9315281668222845, | |
| "eval_loss": 0.2894458770751953, | |
| "eval_precision_macro": 0.9553571428571429, | |
| "eval_precision_micro": 0.935064935064935, | |
| "eval_precision_weighted": 0.9553571428571429, | |
| "eval_recall_macro": 0.935064935064935, | |
| "eval_recall_micro": 0.935064935064935, | |
| "eval_recall_weighted": 0.935064935064935, | |
| "eval_runtime": 23.6193, | |
| "eval_samples_per_second": 3.26, | |
| "eval_steps_per_second": 0.212, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 1.910532832145691, | |
| "learning_rate": 3.6919831223628695e-05, | |
| "loss": 0.0282, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 2.4399540424346924, | |
| "learning_rate": 3.670886075949367e-05, | |
| "loss": 0.0293, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "grad_norm": 2.8237359523773193, | |
| "learning_rate": 3.649789029535865e-05, | |
| "loss": 0.0184, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "grad_norm": 5.545835971832275, | |
| "learning_rate": 3.628691983122363e-05, | |
| "loss": 0.0542, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 23.268957138061523, | |
| "learning_rate": 3.607594936708861e-05, | |
| "loss": 0.3274, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 38.25579833984375, | |
| "learning_rate": 3.586497890295359e-05, | |
| "loss": 0.8174, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 29.19462776184082, | |
| "learning_rate": 3.565400843881857e-05, | |
| "loss": 0.4695, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 15.189168930053711, | |
| "learning_rate": 3.5443037974683544e-05, | |
| "loss": 0.5005, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 13.557092666625977, | |
| "learning_rate": 3.523206751054853e-05, | |
| "loss": 0.1581, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "grad_norm": 24.019737243652344, | |
| "learning_rate": 3.50210970464135e-05, | |
| "loss": 0.1921, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 8.620636940002441, | |
| "learning_rate": 3.4810126582278487e-05, | |
| "loss": 0.1287, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 27.298948287963867, | |
| "learning_rate": 3.459915611814346e-05, | |
| "loss": 0.1861, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 2.0457346439361572, | |
| "learning_rate": 3.438818565400844e-05, | |
| "loss": 0.0309, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 0.20146967470645905, | |
| "learning_rate": 3.4177215189873416e-05, | |
| "loss": 0.0051, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "grad_norm": 5.008052825927734, | |
| "learning_rate": 3.39662447257384e-05, | |
| "loss": 0.0413, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 7.44640588760376, | |
| "learning_rate": 3.3755274261603375e-05, | |
| "loss": 0.0625, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "grad_norm": 1.3147083520889282, | |
| "learning_rate": 3.354430379746836e-05, | |
| "loss": 0.0202, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 30.254919052124023, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 0.5385, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 8.866931915283203, | |
| "learning_rate": 3.312236286919831e-05, | |
| "loss": 0.0891, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 6.436508655548096, | |
| "learning_rate": 3.291139240506329e-05, | |
| "loss": 0.0887, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 25.527908325195312, | |
| "learning_rate": 3.270042194092827e-05, | |
| "loss": 0.4717, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 20.949968338012695, | |
| "learning_rate": 3.248945147679325e-05, | |
| "loss": 0.2791, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 1.1503649950027466, | |
| "learning_rate": 3.227848101265823e-05, | |
| "loss": 0.0226, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 11.437671661376953, | |
| "learning_rate": 3.2067510548523214e-05, | |
| "loss": 0.1807, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 0.25332069396972656, | |
| "learning_rate": 3.185654008438819e-05, | |
| "loss": 0.0062, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 1.8780378103256226, | |
| "learning_rate": 3.1645569620253167e-05, | |
| "loss": 0.0284, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 6.392487049102783, | |
| "learning_rate": 3.143459915611814e-05, | |
| "loss": 0.06, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 23.06869888305664, | |
| "learning_rate": 3.1223628691983126e-05, | |
| "loss": 0.3186, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 9.919283866882324, | |
| "learning_rate": 3.10126582278481e-05, | |
| "loss": 0.0903, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 13.233536720275879, | |
| "learning_rate": 3.0801687763713086e-05, | |
| "loss": 0.1328, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "grad_norm": 18.54045867919922, | |
| "learning_rate": 3.059071729957806e-05, | |
| "loss": 0.2366, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 1.4564893245697021, | |
| "learning_rate": 3.0379746835443042e-05, | |
| "loss": 0.0253, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 0.2900470495223999, | |
| "learning_rate": 3.0168776371308015e-05, | |
| "loss": 0.0065, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 0.4513213038444519, | |
| "learning_rate": 2.9957805907172998e-05, | |
| "loss": 0.0087, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 4.35510778427124, | |
| "learning_rate": 2.9746835443037974e-05, | |
| "loss": 0.0477, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 2.07529354095459, | |
| "learning_rate": 2.9535864978902954e-05, | |
| "loss": 0.0166, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 0.21565364301204681, | |
| "learning_rate": 2.9324894514767937e-05, | |
| "loss": 0.0047, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 1.532738208770752, | |
| "learning_rate": 2.9113924050632914e-05, | |
| "loss": 0.0174, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 7.4955949783325195, | |
| "learning_rate": 2.8902953586497894e-05, | |
| "loss": 0.1436, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 27.709796905517578, | |
| "learning_rate": 2.869198312236287e-05, | |
| "loss": 0.784, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "grad_norm": 0.33867835998535156, | |
| "learning_rate": 2.848101265822785e-05, | |
| "loss": 0.0058, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 0.512848436832428, | |
| "learning_rate": 2.8270042194092826e-05, | |
| "loss": 0.0065, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "grad_norm": 18.563438415527344, | |
| "learning_rate": 2.805907172995781e-05, | |
| "loss": 0.1981, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 18.82804298400879, | |
| "learning_rate": 2.7848101265822786e-05, | |
| "loss": 0.1722, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "grad_norm": 4.428062438964844, | |
| "learning_rate": 2.7637130801687766e-05, | |
| "loss": 0.038, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 3.624861478805542, | |
| "learning_rate": 2.7426160337552742e-05, | |
| "loss": 0.0346, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 6.063285827636719, | |
| "learning_rate": 2.7215189873417722e-05, | |
| "loss": 0.0494, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "grad_norm": 12.928357124328613, | |
| "learning_rate": 2.7004219409282698e-05, | |
| "loss": 0.3157, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 1.6940982341766357, | |
| "learning_rate": 2.679324894514768e-05, | |
| "loss": 0.0192, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "grad_norm": 0.3945218622684479, | |
| "learning_rate": 2.6582278481012658e-05, | |
| "loss": 0.0058, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 0.15542316436767578, | |
| "learning_rate": 2.6371308016877638e-05, | |
| "loss": 0.0032, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 0.11174587905406952, | |
| "learning_rate": 2.616033755274262e-05, | |
| "loss": 0.0025, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 23.488004684448242, | |
| "learning_rate": 2.5949367088607597e-05, | |
| "loss": 0.8641, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "grad_norm": 0.5066362023353577, | |
| "learning_rate": 2.5738396624472577e-05, | |
| "loss": 0.0083, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "grad_norm": 32.906803131103516, | |
| "learning_rate": 2.5527426160337553e-05, | |
| "loss": 0.6371, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "grad_norm": 11.186979293823242, | |
| "learning_rate": 2.5316455696202533e-05, | |
| "loss": 0.1184, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 26.368669509887695, | |
| "learning_rate": 2.510548523206751e-05, | |
| "loss": 0.2113, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "grad_norm": 2.7630584239959717, | |
| "learning_rate": 2.4894514767932493e-05, | |
| "loss": 0.0236, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "grad_norm": 26.372888565063477, | |
| "learning_rate": 2.468354430379747e-05, | |
| "loss": 0.339, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "grad_norm": 7.727011203765869, | |
| "learning_rate": 2.447257383966245e-05, | |
| "loss": 0.091, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "grad_norm": 7.276378631591797, | |
| "learning_rate": 2.426160337552743e-05, | |
| "loss": 0.0345, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "grad_norm": 1.4434276819229126, | |
| "learning_rate": 2.4050632911392405e-05, | |
| "loss": 0.0191, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 1.5443305969238281, | |
| "learning_rate": 2.3839662447257385e-05, | |
| "loss": 0.0139, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "grad_norm": 7.824341297149658, | |
| "learning_rate": 2.3628691983122365e-05, | |
| "loss": 0.1015, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "grad_norm": 0.20900146663188934, | |
| "learning_rate": 2.341772151898734e-05, | |
| "loss": 0.0033, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "grad_norm": 0.32141363620758057, | |
| "learning_rate": 2.320675105485232e-05, | |
| "loss": 0.0059, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 21.58867073059082, | |
| "learning_rate": 2.29957805907173e-05, | |
| "loss": 0.5102, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "grad_norm": 1.1995337009429932, | |
| "learning_rate": 2.278481012658228e-05, | |
| "loss": 0.0088, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 8.670762062072754, | |
| "learning_rate": 2.2573839662447257e-05, | |
| "loss": 0.0449, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 43.51858139038086, | |
| "learning_rate": 2.2362869198312237e-05, | |
| "loss": 0.7478, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "grad_norm": 21.07337760925293, | |
| "learning_rate": 2.2151898734177217e-05, | |
| "loss": 0.4485, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 0.5225343704223633, | |
| "learning_rate": 2.1940928270042196e-05, | |
| "loss": 0.0063, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "grad_norm": 0.6471118330955505, | |
| "learning_rate": 2.1729957805907176e-05, | |
| "loss": 0.0052, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 5.603309154510498, | |
| "learning_rate": 2.1518987341772153e-05, | |
| "loss": 0.0676, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "grad_norm": 18.191370010375977, | |
| "learning_rate": 2.1308016877637132e-05, | |
| "loss": 0.1428, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "grad_norm": 15.37728214263916, | |
| "learning_rate": 2.1097046413502112e-05, | |
| "loss": 0.4738, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.964952290058136, | |
| "learning_rate": 2.088607594936709e-05, | |
| "loss": 0.0135, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 20.822654724121094, | |
| "learning_rate": 2.067510548523207e-05, | |
| "loss": 0.359, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "grad_norm": 15.292733192443848, | |
| "learning_rate": 2.0464135021097048e-05, | |
| "loss": 0.3184, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "grad_norm": 1.5272413492202759, | |
| "learning_rate": 2.0253164556962025e-05, | |
| "loss": 0.0186, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 17.016498565673828, | |
| "learning_rate": 2.0042194092827004e-05, | |
| "loss": 0.2154, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "grad_norm": 1.4779866933822632, | |
| "learning_rate": 1.9831223628691984e-05, | |
| "loss": 0.0165, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 0.8727405071258545, | |
| "learning_rate": 1.962025316455696e-05, | |
| "loss": 0.0102, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "grad_norm": 13.600626945495605, | |
| "learning_rate": 1.940928270042194e-05, | |
| "loss": 0.1604, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "grad_norm": 0.77298504114151, | |
| "learning_rate": 1.919831223628692e-05, | |
| "loss": 0.0094, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "grad_norm": 8.220309257507324, | |
| "learning_rate": 1.89873417721519e-05, | |
| "loss": 0.0887, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "grad_norm": 15.369383811950684, | |
| "learning_rate": 1.877637130801688e-05, | |
| "loss": 0.3962, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 7.943267822265625, | |
| "learning_rate": 1.856540084388186e-05, | |
| "loss": 0.0342, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.961038961038961, | |
| "eval_f1_macro": 0.9603007518796992, | |
| "eval_f1_micro": 0.961038961038961, | |
| "eval_f1_weighted": 0.9603007518796992, | |
| "eval_loss": 0.05716231092810631, | |
| "eval_precision_macro": 0.9693877551020408, | |
| "eval_precision_micro": 0.961038961038961, | |
| "eval_precision_weighted": 0.9693877551020408, | |
| "eval_recall_macro": 0.961038961038961, | |
| "eval_recall_micro": 0.961038961038961, | |
| "eval_recall_weighted": 0.961038961038961, | |
| "eval_runtime": 23.4765, | |
| "eval_samples_per_second": 3.28, | |
| "eval_steps_per_second": 0.213, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "grad_norm": 0.05514498054981232, | |
| "learning_rate": 1.8354430379746836e-05, | |
| "loss": 0.0014, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 2.1552014350891113, | |
| "learning_rate": 1.8143459915611816e-05, | |
| "loss": 0.0202, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "grad_norm": 0.025016020983457565, | |
| "learning_rate": 1.7932489451476795e-05, | |
| "loss": 0.0007, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 15.627620697021484, | |
| "learning_rate": 1.7721518987341772e-05, | |
| "loss": 0.1231, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "grad_norm": 0.46118250489234924, | |
| "learning_rate": 1.751054852320675e-05, | |
| "loss": 0.0063, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "grad_norm": 15.023947715759277, | |
| "learning_rate": 1.729957805907173e-05, | |
| "loss": 0.1147, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 11.93247127532959, | |
| "learning_rate": 1.7088607594936708e-05, | |
| "loss": 0.0811, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "grad_norm": 2.5019609928131104, | |
| "learning_rate": 1.6877637130801688e-05, | |
| "loss": 0.0119, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 0.09159235656261444, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.0014, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "grad_norm": 26.242231369018555, | |
| "learning_rate": 1.6455696202531644e-05, | |
| "loss": 0.236, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 2.3096747398376465, | |
| "learning_rate": 1.6244725738396624e-05, | |
| "loss": 0.0093, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 0.5707495212554932, | |
| "learning_rate": 1.6033755274261607e-05, | |
| "loss": 0.0059, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "grad_norm": 6.667356014251709, | |
| "learning_rate": 1.5822784810126583e-05, | |
| "loss": 0.0563, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 0.10668758302927017, | |
| "learning_rate": 1.5611814345991563e-05, | |
| "loss": 0.0023, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "grad_norm": 4.116157531738281, | |
| "learning_rate": 1.5400843881856543e-05, | |
| "loss": 0.0335, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 26.253101348876953, | |
| "learning_rate": 1.5189873417721521e-05, | |
| "loss": 0.1706, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "grad_norm": 1.0219850540161133, | |
| "learning_rate": 1.4978902953586499e-05, | |
| "loss": 0.0066, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 41.47359848022461, | |
| "learning_rate": 1.4767932489451477e-05, | |
| "loss": 0.6927, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 9.33047866821289, | |
| "learning_rate": 1.4556962025316457e-05, | |
| "loss": 0.0786, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "grad_norm": 0.5574650168418884, | |
| "learning_rate": 1.4345991561181435e-05, | |
| "loss": 0.0057, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 0.4510033428668976, | |
| "learning_rate": 1.4135021097046413e-05, | |
| "loss": 0.0043, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "grad_norm": 0.07699360698461533, | |
| "learning_rate": 1.3924050632911393e-05, | |
| "loss": 0.0015, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "grad_norm": 12.339035034179688, | |
| "learning_rate": 1.3713080168776371e-05, | |
| "loss": 0.1513, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 29.081424713134766, | |
| "learning_rate": 1.3502109704641349e-05, | |
| "loss": 0.2272, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "grad_norm": 1.5039241313934326, | |
| "learning_rate": 1.3291139240506329e-05, | |
| "loss": 0.0113, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "grad_norm": 1.7569365501403809, | |
| "learning_rate": 1.308016877637131e-05, | |
| "loss": 0.0143, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "grad_norm": 4.633851051330566, | |
| "learning_rate": 1.2869198312236289e-05, | |
| "loss": 0.0273, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 9.019872665405273, | |
| "learning_rate": 1.2658227848101267e-05, | |
| "loss": 0.1055, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "grad_norm": 0.14426438510417938, | |
| "learning_rate": 1.2447257383966246e-05, | |
| "loss": 0.0023, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "grad_norm": 0.1260414719581604, | |
| "learning_rate": 1.2236286919831224e-05, | |
| "loss": 0.0023, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "grad_norm": 0.14926879107952118, | |
| "learning_rate": 1.2025316455696203e-05, | |
| "loss": 0.0018, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 0.3730136752128601, | |
| "learning_rate": 1.1814345991561182e-05, | |
| "loss": 0.0058, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "grad_norm": 4.101214408874512, | |
| "learning_rate": 1.160337552742616e-05, | |
| "loss": 0.0279, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 2.671245574951172, | |
| "learning_rate": 1.139240506329114e-05, | |
| "loss": 0.023, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 6.172547817230225, | |
| "learning_rate": 1.1181434599156118e-05, | |
| "loss": 0.0333, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "grad_norm": 18.09273910522461, | |
| "learning_rate": 1.0970464135021098e-05, | |
| "loss": 0.3601, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 9.900489807128906, | |
| "learning_rate": 1.0759493670886076e-05, | |
| "loss": 0.5438, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "grad_norm": 27.930665969848633, | |
| "learning_rate": 1.0548523206751056e-05, | |
| "loss": 0.2069, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "grad_norm": 0.493826687335968, | |
| "learning_rate": 1.0337552742616034e-05, | |
| "loss": 0.0039, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "grad_norm": 0.08800631016492844, | |
| "learning_rate": 1.0126582278481012e-05, | |
| "loss": 0.0015, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "grad_norm": 1.6704317331314087, | |
| "learning_rate": 9.915611814345992e-06, | |
| "loss": 0.0121, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 0.060208290815353394, | |
| "learning_rate": 9.70464135021097e-06, | |
| "loss": 0.001, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "grad_norm": 0.42390766739845276, | |
| "learning_rate": 9.49367088607595e-06, | |
| "loss": 0.0043, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 5.121891021728516, | |
| "learning_rate": 9.28270042194093e-06, | |
| "loss": 0.0373, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "grad_norm": 0.2980020344257355, | |
| "learning_rate": 9.071729957805908e-06, | |
| "loss": 0.0019, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "grad_norm": 1.821443796157837, | |
| "learning_rate": 8.860759493670886e-06, | |
| "loss": 0.0096, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 0.2338918149471283, | |
| "learning_rate": 8.649789029535866e-06, | |
| "loss": 0.0033, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 0.03316354379057884, | |
| "learning_rate": 8.438818565400844e-06, | |
| "loss": 0.001, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 44.416473388671875, | |
| "learning_rate": 8.227848101265822e-06, | |
| "loss": 0.2124, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "grad_norm": 0.2957440912723541, | |
| "learning_rate": 8.016877637130803e-06, | |
| "loss": 0.0041, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "grad_norm": 18.5585994720459, | |
| "learning_rate": 7.805907172995782e-06, | |
| "loss": 0.0596, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "grad_norm": 5.535640716552734, | |
| "learning_rate": 7.5949367088607605e-06, | |
| "loss": 0.0476, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 38.173789978027344, | |
| "learning_rate": 7.3839662447257386e-06, | |
| "loss": 0.6888, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "grad_norm": 18.576967239379883, | |
| "learning_rate": 7.1729957805907175e-06, | |
| "loss": 0.3636, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "grad_norm": 3.596296548843384, | |
| "learning_rate": 6.9620253164556965e-06, | |
| "loss": 0.0224, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 0.09587596356868744, | |
| "learning_rate": 6.7510548523206746e-06, | |
| "loss": 0.0022, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "grad_norm": 0.15251006186008453, | |
| "learning_rate": 6.540084388185655e-06, | |
| "loss": 0.0026, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 0.9377198219299316, | |
| "learning_rate": 6.329113924050633e-06, | |
| "loss": 0.0081, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "grad_norm": 11.142963409423828, | |
| "learning_rate": 6.118143459915612e-06, | |
| "loss": 0.1642, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 31.128833770751953, | |
| "learning_rate": 5.907172995780591e-06, | |
| "loss": 0.3947, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "grad_norm": 0.6137996315956116, | |
| "learning_rate": 5.69620253164557e-06, | |
| "loss": 0.0033, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "grad_norm": 0.1146240383386612, | |
| "learning_rate": 5.485232067510549e-06, | |
| "loss": 0.0018, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "grad_norm": 14.48742389678955, | |
| "learning_rate": 5.274261603375528e-06, | |
| "loss": 0.2166, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 1.7509044408798218, | |
| "learning_rate": 5.063291139240506e-06, | |
| "loss": 0.0174, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 0.0605483315885067, | |
| "learning_rate": 4.852320675105485e-06, | |
| "loss": 0.001, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 0.1251741498708725, | |
| "learning_rate": 4.641350210970465e-06, | |
| "loss": 0.0018, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 0.15386824309825897, | |
| "learning_rate": 4.430379746835443e-06, | |
| "loss": 0.0025, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "grad_norm": 6.340598106384277, | |
| "learning_rate": 4.219409282700422e-06, | |
| "loss": 0.0342, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "grad_norm": 0.18014036118984222, | |
| "learning_rate": 4.008438818565402e-06, | |
| "loss": 0.0022, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 17.62770652770996, | |
| "learning_rate": 3.7974683544303802e-06, | |
| "loss": 0.1143, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "grad_norm": 0.17265360057353973, | |
| "learning_rate": 3.5864978902953588e-06, | |
| "loss": 0.0026, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 6.071741580963135, | |
| "learning_rate": 3.3755274261603373e-06, | |
| "loss": 0.0373, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "grad_norm": 1.1394612789154053, | |
| "learning_rate": 3.1645569620253167e-06, | |
| "loss": 0.0098, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 0.09031672030687332, | |
| "learning_rate": 2.9535864978902956e-06, | |
| "loss": 0.0017, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "grad_norm": 23.36667251586914, | |
| "learning_rate": 2.7426160337552745e-06, | |
| "loss": 0.7473, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "grad_norm": 0.17166578769683838, | |
| "learning_rate": 2.531645569620253e-06, | |
| "loss": 0.0015, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 0.7906249165534973, | |
| "learning_rate": 2.3206751054852324e-06, | |
| "loss": 0.0071, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "grad_norm": 13.26647663116455, | |
| "learning_rate": 2.109704641350211e-06, | |
| "loss": 0.1543, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.22246617078781128, | |
| "learning_rate": 1.8987341772151901e-06, | |
| "loss": 0.002, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "grad_norm": 5.0251784324646, | |
| "learning_rate": 1.6877637130801686e-06, | |
| "loss": 0.0597, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 0.6754881143569946, | |
| "learning_rate": 1.4767932489451478e-06, | |
| "loss": 0.0062, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "grad_norm": 0.03976951539516449, | |
| "learning_rate": 1.2658227848101265e-06, | |
| "loss": 0.0009, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "grad_norm": 0.01055199932307005, | |
| "learning_rate": 1.0548523206751055e-06, | |
| "loss": 0.0003, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "grad_norm": 20.83010482788086, | |
| "learning_rate": 8.438818565400843e-07, | |
| "loss": 0.371, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "grad_norm": 0.4786561131477356, | |
| "learning_rate": 6.329113924050633e-07, | |
| "loss": 0.0055, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 11.4153413772583, | |
| "learning_rate": 4.2194092827004216e-07, | |
| "loss": 0.0949, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "grad_norm": 0.1931201070547104, | |
| "learning_rate": 2.1097046413502108e-07, | |
| "loss": 0.0028, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 6.672695636749268, | |
| "learning_rate": 0.0, | |
| "loss": 0.0624, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 1.0, | |
| "eval_f1_macro": 1.0, | |
| "eval_f1_micro": 1.0, | |
| "eval_f1_weighted": 1.0, | |
| "eval_loss": 0.011418873444199562, | |
| "eval_precision_macro": 1.0, | |
| "eval_precision_micro": 1.0, | |
| "eval_precision_weighted": 1.0, | |
| "eval_recall_macro": 1.0, | |
| "eval_recall_micro": 1.0, | |
| "eval_recall_weighted": 1.0, | |
| "eval_runtime": 23.0544, | |
| "eval_samples_per_second": 3.34, | |
| "eval_steps_per_second": 0.217, | |
| "step": 264 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 264, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 1.626709167691776e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |