| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 1350, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0022222222222222222, | |
| "grad_norm": 0.09812982380390167, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.647, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0044444444444444444, | |
| "grad_norm": 0.11622905731201172, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.6394, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.006666666666666667, | |
| "grad_norm": 0.09975595772266388, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.6547, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.008888888888888889, | |
| "grad_norm": 0.09113696217536926, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.8181, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.011111111111111112, | |
| "grad_norm": 0.10436710715293884, | |
| "learning_rate": 1.111111111111111e-05, | |
| "loss": 0.6559, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.013333333333333334, | |
| "grad_norm": 0.0823315680027008, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.668, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.015555555555555555, | |
| "grad_norm": 0.11999443173408508, | |
| "learning_rate": 1.5555555555555555e-05, | |
| "loss": 0.7814, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.017777777777777778, | |
| "grad_norm": 0.15283814072608948, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 0.7522, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.11188939958810806, | |
| "learning_rate": 1.9999999999999998e-05, | |
| "loss": 0.7548, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.022222222222222223, | |
| "grad_norm": 0.18060794472694397, | |
| "learning_rate": 2.222222222222222e-05, | |
| "loss": 0.8093, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.024444444444444446, | |
| "grad_norm": 0.12840260565280914, | |
| "learning_rate": 2.4444444444444445e-05, | |
| "loss": 0.8086, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02666666666666667, | |
| "grad_norm": 0.14665867388248444, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.7813, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.028888888888888888, | |
| "grad_norm": 0.12048947066068649, | |
| "learning_rate": 2.8888888888888885e-05, | |
| "loss": 0.6564, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03111111111111111, | |
| "grad_norm": 0.08004658669233322, | |
| "learning_rate": 3.111111111111111e-05, | |
| "loss": 0.4802, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03333333333333333, | |
| "grad_norm": 0.09866306185722351, | |
| "learning_rate": 3.333333333333333e-05, | |
| "loss": 0.6343, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.035555555555555556, | |
| "grad_norm": 0.1632370799779892, | |
| "learning_rate": 3.555555555555555e-05, | |
| "loss": 0.7072, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03777777777777778, | |
| "grad_norm": 0.17000310122966766, | |
| "learning_rate": 3.777777777777777e-05, | |
| "loss": 0.9045, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.15400268137454987, | |
| "learning_rate": 3.9999999999999996e-05, | |
| "loss": 0.6858, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.042222222222222223, | |
| "grad_norm": 0.2376849204301834, | |
| "learning_rate": 4.222222222222222e-05, | |
| "loss": 0.8061, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.044444444444444446, | |
| "grad_norm": 0.13313539326190948, | |
| "learning_rate": 4.444444444444444e-05, | |
| "loss": 0.7519, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04666666666666667, | |
| "grad_norm": 0.15802322328090668, | |
| "learning_rate": 4.6666666666666665e-05, | |
| "loss": 0.636, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04888888888888889, | |
| "grad_norm": 0.19093258678913116, | |
| "learning_rate": 4.888888888888889e-05, | |
| "loss": 0.6709, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.051111111111111114, | |
| "grad_norm": 0.11460094153881073, | |
| "learning_rate": 5.111111111111111e-05, | |
| "loss": 0.5528, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05333333333333334, | |
| "grad_norm": 0.1695665717124939, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 0.6439, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.05555555555555555, | |
| "grad_norm": 0.21789449453353882, | |
| "learning_rate": 5.5555555555555545e-05, | |
| "loss": 0.7578, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.057777777777777775, | |
| "grad_norm": 0.18585897982120514, | |
| "learning_rate": 5.777777777777777e-05, | |
| "loss": 0.7475, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.21285971999168396, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 0.6961, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.06222222222222222, | |
| "grad_norm": 0.17276039719581604, | |
| "learning_rate": 6.222222222222222e-05, | |
| "loss": 0.6964, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.06444444444444444, | |
| "grad_norm": 0.14113998413085938, | |
| "learning_rate": 6.444444444444444e-05, | |
| "loss": 0.5862, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.06666666666666667, | |
| "grad_norm": 0.16765236854553223, | |
| "learning_rate": 6.666666666666666e-05, | |
| "loss": 0.7335, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06888888888888889, | |
| "grad_norm": 0.16216206550598145, | |
| "learning_rate": 6.888888888888888e-05, | |
| "loss": 0.6828, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07111111111111111, | |
| "grad_norm": 0.14560526609420776, | |
| "learning_rate": 7.11111111111111e-05, | |
| "loss": 0.7145, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07333333333333333, | |
| "grad_norm": 0.1773616522550583, | |
| "learning_rate": 7.333333333333332e-05, | |
| "loss": 0.6111, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.07555555555555556, | |
| "grad_norm": 0.19081515073776245, | |
| "learning_rate": 7.555555555555554e-05, | |
| "loss": 0.6096, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.07777777777777778, | |
| "grad_norm": 0.2005636990070343, | |
| "learning_rate": 7.777777777777777e-05, | |
| "loss": 0.6531, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.16567005217075348, | |
| "learning_rate": 7.999999999999999e-05, | |
| "loss": 0.4763, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08222222222222222, | |
| "grad_norm": 0.20861048996448517, | |
| "learning_rate": 8.222222222222222e-05, | |
| "loss": 0.7007, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.08444444444444445, | |
| "grad_norm": 0.20962809026241302, | |
| "learning_rate": 8.444444444444444e-05, | |
| "loss": 0.5476, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.08666666666666667, | |
| "grad_norm": 0.2487533837556839, | |
| "learning_rate": 8.666666666666665e-05, | |
| "loss": 0.6924, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.08888888888888889, | |
| "grad_norm": 0.15908700227737427, | |
| "learning_rate": 8.888888888888888e-05, | |
| "loss": 0.5608, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09111111111111111, | |
| "grad_norm": 0.20576216280460358, | |
| "learning_rate": 9.11111111111111e-05, | |
| "loss": 0.6377, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.09333333333333334, | |
| "grad_norm": 0.2465807944536209, | |
| "learning_rate": 9.333333333333333e-05, | |
| "loss": 0.7257, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.09555555555555556, | |
| "grad_norm": 0.2680942118167877, | |
| "learning_rate": 9.555555555555555e-05, | |
| "loss": 0.7667, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.09777777777777778, | |
| "grad_norm": 0.2033873051404953, | |
| "learning_rate": 9.777777777777778e-05, | |
| "loss": 0.7678, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.23584719002246857, | |
| "learning_rate": 9.999999999999999e-05, | |
| "loss": 0.7, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10222222222222223, | |
| "grad_norm": 0.31025272607803345, | |
| "learning_rate": 0.00010222222222222222, | |
| "loss": 0.5652, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.10444444444444445, | |
| "grad_norm": 0.3050137758255005, | |
| "learning_rate": 0.00010444444444444444, | |
| "loss": 0.4665, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.10666666666666667, | |
| "grad_norm": 0.24884286522865295, | |
| "learning_rate": 0.00010666666666666667, | |
| "loss": 0.5984, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.10888888888888888, | |
| "grad_norm": 0.36036840081214905, | |
| "learning_rate": 0.00010888888888888889, | |
| "loss": 0.5503, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1111111111111111, | |
| "grad_norm": 0.31548580527305603, | |
| "learning_rate": 0.00011111111111111109, | |
| "loss": 0.6297, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11333333333333333, | |
| "grad_norm": 0.3017398715019226, | |
| "learning_rate": 0.00011333333333333331, | |
| "loss": 0.5088, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.11555555555555555, | |
| "grad_norm": 0.3036445379257202, | |
| "learning_rate": 0.00011555555555555554, | |
| "loss": 0.4934, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.11777777777777777, | |
| "grad_norm": 0.23346473276615143, | |
| "learning_rate": 0.00011777777777777776, | |
| "loss": 0.6604, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.18171657621860504, | |
| "learning_rate": 0.00011999999999999999, | |
| "loss": 0.5129, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12222222222222222, | |
| "grad_norm": 0.3094526529312134, | |
| "learning_rate": 0.00012222222222222221, | |
| "loss": 0.6951, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.12444444444444444, | |
| "grad_norm": 0.27897611260414124, | |
| "learning_rate": 0.00012444444444444444, | |
| "loss": 0.4351, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.12666666666666668, | |
| "grad_norm": 0.3004632592201233, | |
| "learning_rate": 0.00012666666666666666, | |
| "loss": 0.4784, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1288888888888889, | |
| "grad_norm": 0.38149529695510864, | |
| "learning_rate": 0.0001288888888888889, | |
| "loss": 0.5342, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.13111111111111112, | |
| "grad_norm": 0.2744050621986389, | |
| "learning_rate": 0.00013111111111111111, | |
| "loss": 0.4903, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.13333333333333333, | |
| "grad_norm": 0.2738029360771179, | |
| "learning_rate": 0.0001333333333333333, | |
| "loss": 0.6033, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13555555555555557, | |
| "grad_norm": 0.2923201024532318, | |
| "learning_rate": 0.00013555555555555554, | |
| "loss": 0.6404, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.13777777777777778, | |
| "grad_norm": 0.27477923035621643, | |
| "learning_rate": 0.00013777777777777776, | |
| "loss": 0.5334, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.3043128550052643, | |
| "learning_rate": 0.00014, | |
| "loss": 0.6544, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.14222222222222222, | |
| "grad_norm": 0.27908855676651, | |
| "learning_rate": 0.0001422222222222222, | |
| "loss": 0.5998, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.14444444444444443, | |
| "grad_norm": 0.2519969940185547, | |
| "learning_rate": 0.0001444444444444444, | |
| "loss": 0.5356, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.14666666666666667, | |
| "grad_norm": 0.23952998220920563, | |
| "learning_rate": 0.00014666666666666664, | |
| "loss": 0.5948, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.14888888888888888, | |
| "grad_norm": 0.3552301824092865, | |
| "learning_rate": 0.00014888888888888886, | |
| "loss": 0.6904, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.1511111111111111, | |
| "grad_norm": 0.1643562763929367, | |
| "learning_rate": 0.00015111111111111109, | |
| "loss": 0.5608, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.15333333333333332, | |
| "grad_norm": 0.2761911153793335, | |
| "learning_rate": 0.0001533333333333333, | |
| "loss": 0.6825, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.15555555555555556, | |
| "grad_norm": 0.25992175936698914, | |
| "learning_rate": 0.00015555555555555554, | |
| "loss": 0.613, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15777777777777777, | |
| "grad_norm": 0.19053922593593597, | |
| "learning_rate": 0.00015777777777777776, | |
| "loss": 0.5386, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.24203050136566162, | |
| "learning_rate": 0.00015999999999999999, | |
| "loss": 0.4925, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.1622222222222222, | |
| "grad_norm": 0.2526642978191376, | |
| "learning_rate": 0.0001622222222222222, | |
| "loss": 0.6204, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.16444444444444445, | |
| "grad_norm": 0.2451915442943573, | |
| "learning_rate": 0.00016444444444444444, | |
| "loss": 0.645, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.21893125772476196, | |
| "learning_rate": 0.00016666666666666666, | |
| "loss": 0.5017, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1688888888888889, | |
| "grad_norm": 0.4453028738498688, | |
| "learning_rate": 0.00016888888888888889, | |
| "loss": 0.6523, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1711111111111111, | |
| "grad_norm": 0.9082323908805847, | |
| "learning_rate": 0.0001711111111111111, | |
| "loss": 0.565, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.17333333333333334, | |
| "grad_norm": 0.31528621912002563, | |
| "learning_rate": 0.0001733333333333333, | |
| "loss": 0.4987, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.17555555555555555, | |
| "grad_norm": 0.2677905559539795, | |
| "learning_rate": 0.00017555555555555553, | |
| "loss": 0.6598, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.17777777777777778, | |
| "grad_norm": 0.188288152217865, | |
| "learning_rate": 0.00017777777777777776, | |
| "loss": 0.5779, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.34003448486328125, | |
| "learning_rate": 0.00017999999999999998, | |
| "loss": 0.5187, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.18222222222222223, | |
| "grad_norm": 0.20124401152133942, | |
| "learning_rate": 0.0001822222222222222, | |
| "loss": 0.4285, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.18444444444444444, | |
| "grad_norm": 0.2644328773021698, | |
| "learning_rate": 0.00018444444444444443, | |
| "loss": 0.5645, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.18666666666666668, | |
| "grad_norm": 0.2735807001590729, | |
| "learning_rate": 0.00018666666666666666, | |
| "loss": 0.5681, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.18888888888888888, | |
| "grad_norm": 0.21255911886692047, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 0.6939, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.19111111111111112, | |
| "grad_norm": 0.2228507250547409, | |
| "learning_rate": 0.0001911111111111111, | |
| "loss": 0.6011, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.19333333333333333, | |
| "grad_norm": 0.19216273725032806, | |
| "learning_rate": 0.00019333333333333333, | |
| "loss": 0.6372, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.19555555555555557, | |
| "grad_norm": 0.22664876282215118, | |
| "learning_rate": 0.00019555555555555556, | |
| "loss": 0.4512, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.19777777777777777, | |
| "grad_norm": 0.18906566500663757, | |
| "learning_rate": 0.00019777777777777776, | |
| "loss": 0.5389, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.2111695557832718, | |
| "learning_rate": 0.00019999999999999998, | |
| "loss": 0.4883, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20222222222222222, | |
| "grad_norm": 0.1915666162967682, | |
| "learning_rate": 0.0002022222222222222, | |
| "loss": 0.5814, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.20444444444444446, | |
| "grad_norm": 0.2227359414100647, | |
| "learning_rate": 0.00020444444444444443, | |
| "loss": 0.6084, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.20666666666666667, | |
| "grad_norm": 0.229144886136055, | |
| "learning_rate": 0.00020666666666666666, | |
| "loss": 0.6006, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.2088888888888889, | |
| "grad_norm": 0.5038937926292419, | |
| "learning_rate": 0.00020888888888888888, | |
| "loss": 0.5346, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2111111111111111, | |
| "grad_norm": 0.23706114292144775, | |
| "learning_rate": 0.0002111111111111111, | |
| "loss": 0.6215, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.21333333333333335, | |
| "grad_norm": 0.25473618507385254, | |
| "learning_rate": 0.00021333333333333333, | |
| "loss": 0.6144, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.21555555555555556, | |
| "grad_norm": 0.19588348269462585, | |
| "learning_rate": 0.00021555555555555556, | |
| "loss": 0.4981, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.21777777777777776, | |
| "grad_norm": 0.1835206001996994, | |
| "learning_rate": 0.00021777777777777778, | |
| "loss": 0.5044, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.17308925092220306, | |
| "learning_rate": 0.00021999999999999995, | |
| "loss": 0.5236, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.2222222222222222, | |
| "grad_norm": 0.20074211061000824, | |
| "learning_rate": 0.00022222222222222218, | |
| "loss": 0.5777, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.22444444444444445, | |
| "grad_norm": 0.24682994186878204, | |
| "learning_rate": 0.0002244444444444444, | |
| "loss": 0.6089, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.22666666666666666, | |
| "grad_norm": 0.1767152100801468, | |
| "learning_rate": 0.00022666666666666663, | |
| "loss": 0.3809, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.2288888888888889, | |
| "grad_norm": 0.24087251722812653, | |
| "learning_rate": 0.00022888888888888885, | |
| "loss": 0.4984, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2311111111111111, | |
| "grad_norm": 0.2400844693183899, | |
| "learning_rate": 0.00023111111111111108, | |
| "loss": 0.5902, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.23333333333333334, | |
| "grad_norm": 0.14252978563308716, | |
| "learning_rate": 0.0002333333333333333, | |
| "loss": 0.4849, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.23555555555555555, | |
| "grad_norm": 0.20786775648593903, | |
| "learning_rate": 0.00023555555555555553, | |
| "loss": 0.4328, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.23777777777777778, | |
| "grad_norm": 0.21549014747142792, | |
| "learning_rate": 0.00023777777777777775, | |
| "loss": 0.5847, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.2561536729335785, | |
| "learning_rate": 0.00023999999999999998, | |
| "loss": 0.6082, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.24222222222222223, | |
| "grad_norm": 0.2900320887565613, | |
| "learning_rate": 0.0002422222222222222, | |
| "loss": 0.469, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.24444444444444444, | |
| "grad_norm": 0.22536203265190125, | |
| "learning_rate": 0.00024444444444444443, | |
| "loss": 0.4642, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.24666666666666667, | |
| "grad_norm": 0.21168376505374908, | |
| "learning_rate": 0.0002466666666666666, | |
| "loss": 0.5069, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.24888888888888888, | |
| "grad_norm": 0.16381123661994934, | |
| "learning_rate": 0.0002488888888888889, | |
| "loss": 0.5384, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.2511111111111111, | |
| "grad_norm": 0.17860746383666992, | |
| "learning_rate": 0.0002511111111111111, | |
| "loss": 0.4772, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.25333333333333335, | |
| "grad_norm": 0.20924940705299377, | |
| "learning_rate": 0.00025333333333333333, | |
| "loss": 0.6553, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.25555555555555554, | |
| "grad_norm": 0.14452561736106873, | |
| "learning_rate": 0.00025555555555555553, | |
| "loss": 0.4973, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2577777777777778, | |
| "grad_norm": 0.2318609207868576, | |
| "learning_rate": 0.0002577777777777778, | |
| "loss": 0.4733, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.18076716363430023, | |
| "learning_rate": 0.00026, | |
| "loss": 0.4933, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.26222222222222225, | |
| "grad_norm": 0.21175789833068848, | |
| "learning_rate": 0.00026222222222222223, | |
| "loss": 0.5845, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2644444444444444, | |
| "grad_norm": 0.173040971159935, | |
| "learning_rate": 0.00026444444444444443, | |
| "loss": 0.6786, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.26666666666666666, | |
| "grad_norm": 0.19183841347694397, | |
| "learning_rate": 0.0002666666666666666, | |
| "loss": 0.5924, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2688888888888889, | |
| "grad_norm": 0.15684929490089417, | |
| "learning_rate": 0.0002688888888888889, | |
| "loss": 0.4669, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.27111111111111114, | |
| "grad_norm": 0.20070019364356995, | |
| "learning_rate": 0.0002711111111111111, | |
| "loss": 0.6485, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.2733333333333333, | |
| "grad_norm": 0.20072908699512482, | |
| "learning_rate": 0.00027333333333333333, | |
| "loss": 0.5249, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.27555555555555555, | |
| "grad_norm": 0.273030161857605, | |
| "learning_rate": 0.0002755555555555555, | |
| "loss": 0.5192, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 0.30652835965156555, | |
| "learning_rate": 0.0002777777777777778, | |
| "loss": 0.6819, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.2657855749130249, | |
| "learning_rate": 0.00028, | |
| "loss": 0.699, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.2822222222222222, | |
| "grad_norm": 0.1956014782190323, | |
| "learning_rate": 0.00028222222222222223, | |
| "loss": 0.505, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.28444444444444444, | |
| "grad_norm": 0.19476287066936493, | |
| "learning_rate": 0.0002844444444444444, | |
| "loss": 0.3625, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.2866666666666667, | |
| "grad_norm": 0.18950065970420837, | |
| "learning_rate": 0.0002866666666666667, | |
| "loss": 0.5267, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.28888888888888886, | |
| "grad_norm": 0.1630088835954666, | |
| "learning_rate": 0.0002888888888888888, | |
| "loss": 0.4426, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2911111111111111, | |
| "grad_norm": 0.1103912889957428, | |
| "learning_rate": 0.0002911111111111111, | |
| "loss": 0.4793, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.29333333333333333, | |
| "grad_norm": 0.2730814218521118, | |
| "learning_rate": 0.00029333333333333327, | |
| "loss": 0.6861, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.29555555555555557, | |
| "grad_norm": 0.17552460730075836, | |
| "learning_rate": 0.0002955555555555555, | |
| "loss": 0.5703, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.29777777777777775, | |
| "grad_norm": 0.27005037665367126, | |
| "learning_rate": 0.0002977777777777777, | |
| "loss": 0.4464, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.19579066336154938, | |
| "learning_rate": 0.0003, | |
| "loss": 0.4173, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3022222222222222, | |
| "grad_norm": 0.21760328114032745, | |
| "learning_rate": 0.00029975308641975307, | |
| "loss": 0.5617, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.30444444444444446, | |
| "grad_norm": 0.1697392612695694, | |
| "learning_rate": 0.00029950617283950616, | |
| "loss": 0.566, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.30666666666666664, | |
| "grad_norm": 0.28145694732666016, | |
| "learning_rate": 0.00029925925925925926, | |
| "loss": 0.564, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.3088888888888889, | |
| "grad_norm": 0.16204604506492615, | |
| "learning_rate": 0.0002990123456790123, | |
| "loss": 0.4182, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.3111111111111111, | |
| "grad_norm": 0.15137669444084167, | |
| "learning_rate": 0.0002987654320987654, | |
| "loss": 0.5234, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.31333333333333335, | |
| "grad_norm": 0.1461869329214096, | |
| "learning_rate": 0.0002985185185185185, | |
| "loss": 0.5237, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.31555555555555553, | |
| "grad_norm": 0.24307969212532043, | |
| "learning_rate": 0.0002982716049382716, | |
| "loss": 0.6029, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.31777777777777777, | |
| "grad_norm": 0.3020559549331665, | |
| "learning_rate": 0.0002980246913580247, | |
| "loss": 0.6136, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.19990377128124237, | |
| "learning_rate": 0.0002977777777777777, | |
| "loss": 0.5379, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.32222222222222224, | |
| "grad_norm": 0.22437147796154022, | |
| "learning_rate": 0.00029753086419753087, | |
| "loss": 0.4162, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.3244444444444444, | |
| "grad_norm": 0.1510290801525116, | |
| "learning_rate": 0.0002972839506172839, | |
| "loss": 0.5294, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.32666666666666666, | |
| "grad_norm": 0.43192407488822937, | |
| "learning_rate": 0.000297037037037037, | |
| "loss": 0.5305, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.3288888888888889, | |
| "grad_norm": 0.16263210773468018, | |
| "learning_rate": 0.0002967901234567901, | |
| "loss": 0.4888, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.33111111111111113, | |
| "grad_norm": 0.1575966477394104, | |
| "learning_rate": 0.0002965432098765432, | |
| "loss": 0.5763, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.17795632779598236, | |
| "learning_rate": 0.00029629629629629624, | |
| "loss": 0.5609, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33555555555555555, | |
| "grad_norm": 0.19898423552513123, | |
| "learning_rate": 0.00029604938271604933, | |
| "loss": 0.5039, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.3377777777777778, | |
| "grad_norm": 0.16012270748615265, | |
| "learning_rate": 0.0002958024691358025, | |
| "loss": 0.4414, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.1376243531703949, | |
| "learning_rate": 0.0002955555555555555, | |
| "loss": 0.5205, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.3422222222222222, | |
| "grad_norm": 0.26047253608703613, | |
| "learning_rate": 0.0002953086419753086, | |
| "loss": 0.6142, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.34444444444444444, | |
| "grad_norm": 0.12584461271762848, | |
| "learning_rate": 0.0002950617283950617, | |
| "loss": 0.505, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.3466666666666667, | |
| "grad_norm": 0.16078580915927887, | |
| "learning_rate": 0.0002948148148148148, | |
| "loss": 0.6309, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.3488888888888889, | |
| "grad_norm": 0.17269696295261383, | |
| "learning_rate": 0.00029456790123456785, | |
| "loss": 0.6352, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3511111111111111, | |
| "grad_norm": 0.16890980303287506, | |
| "learning_rate": 0.00029432098765432095, | |
| "loss": 0.4686, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.35333333333333333, | |
| "grad_norm": 0.2434924840927124, | |
| "learning_rate": 0.00029407407407407404, | |
| "loss": 0.6411, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.35555555555555557, | |
| "grad_norm": 0.17170937359333038, | |
| "learning_rate": 0.00029382716049382714, | |
| "loss": 0.5415, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.35777777777777775, | |
| "grad_norm": 0.1625782996416092, | |
| "learning_rate": 0.00029358024691358023, | |
| "loss": 0.5254, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.15762503445148468, | |
| "learning_rate": 0.00029333333333333327, | |
| "loss": 0.4622, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.3622222222222222, | |
| "grad_norm": 0.1959887444972992, | |
| "learning_rate": 0.0002930864197530864, | |
| "loss": 0.4678, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.36444444444444446, | |
| "grad_norm": 0.2749437391757965, | |
| "learning_rate": 0.00029283950617283946, | |
| "loss": 0.6137, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.36666666666666664, | |
| "grad_norm": 0.18642671406269073, | |
| "learning_rate": 0.00029259259259259256, | |
| "loss": 0.5227, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3688888888888889, | |
| "grad_norm": 0.19407618045806885, | |
| "learning_rate": 0.00029234567901234565, | |
| "loss": 0.5151, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.3711111111111111, | |
| "grad_norm": 0.17601671814918518, | |
| "learning_rate": 0.00029209876543209875, | |
| "loss": 0.616, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.37333333333333335, | |
| "grad_norm": 0.17339468002319336, | |
| "learning_rate": 0.00029185185185185184, | |
| "loss": 0.5178, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.37555555555555553, | |
| "grad_norm": 0.19407111406326294, | |
| "learning_rate": 0.0002916049382716049, | |
| "loss": 0.4918, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.37777777777777777, | |
| "grad_norm": 0.18815182149410248, | |
| "learning_rate": 0.000291358024691358, | |
| "loss": 0.588, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.1510704755783081, | |
| "learning_rate": 0.0002911111111111111, | |
| "loss": 0.5143, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.38222222222222224, | |
| "grad_norm": 0.14182798564434052, | |
| "learning_rate": 0.00029086419753086417, | |
| "loss": 0.5953, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.3844444444444444, | |
| "grad_norm": 0.21492992341518402, | |
| "learning_rate": 0.00029061728395061726, | |
| "loss": 0.4714, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.38666666666666666, | |
| "grad_norm": 0.17209042608737946, | |
| "learning_rate": 0.00029037037037037036, | |
| "loss": 0.5507, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.3888888888888889, | |
| "grad_norm": 0.16871920228004456, | |
| "learning_rate": 0.00029012345679012345, | |
| "loss": 0.5589, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.39111111111111113, | |
| "grad_norm": 0.17786094546318054, | |
| "learning_rate": 0.0002898765432098765, | |
| "loss": 0.5746, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3933333333333333, | |
| "grad_norm": 0.1881207823753357, | |
| "learning_rate": 0.0002896296296296296, | |
| "loss": 0.6293, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.39555555555555555, | |
| "grad_norm": 0.19975589215755463, | |
| "learning_rate": 0.0002893827160493827, | |
| "loss": 0.516, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.3977777777777778, | |
| "grad_norm": 0.22304534912109375, | |
| "learning_rate": 0.0002891358024691358, | |
| "loss": 0.5633, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.14426018297672272, | |
| "learning_rate": 0.0002888888888888888, | |
| "loss": 0.4885, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4022222222222222, | |
| "grad_norm": 0.13130474090576172, | |
| "learning_rate": 0.00028864197530864197, | |
| "loss": 0.4831, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.40444444444444444, | |
| "grad_norm": 0.14933516085147858, | |
| "learning_rate": 0.00028839506172839507, | |
| "loss": 0.5218, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.4066666666666667, | |
| "grad_norm": 0.15594199299812317, | |
| "learning_rate": 0.0002881481481481481, | |
| "loss": 0.5768, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.4088888888888889, | |
| "grad_norm": 0.1540926843881607, | |
| "learning_rate": 0.0002879012345679012, | |
| "loss": 0.5169, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.4111111111111111, | |
| "grad_norm": 0.18075008690357208, | |
| "learning_rate": 0.0002876543209876543, | |
| "loss": 0.5052, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.41333333333333333, | |
| "grad_norm": 0.18323589861392975, | |
| "learning_rate": 0.0002874074074074074, | |
| "loss": 0.5293, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.41555555555555557, | |
| "grad_norm": 0.17484310269355774, | |
| "learning_rate": 0.00028716049382716043, | |
| "loss": 0.5611, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.4177777777777778, | |
| "grad_norm": 0.1708841472864151, | |
| "learning_rate": 0.00028691358024691353, | |
| "loss": 0.6788, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.1502651572227478, | |
| "learning_rate": 0.0002866666666666667, | |
| "loss": 0.5619, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.4222222222222222, | |
| "grad_norm": 0.1580984890460968, | |
| "learning_rate": 0.0002864197530864197, | |
| "loss": 0.6505, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.42444444444444446, | |
| "grad_norm": 0.16270650923252106, | |
| "learning_rate": 0.0002861728395061728, | |
| "loss": 0.5653, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.4266666666666667, | |
| "grad_norm": 0.17126257717609406, | |
| "learning_rate": 0.0002859259259259259, | |
| "loss": 0.6122, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.4288888888888889, | |
| "grad_norm": 0.24494518339633942, | |
| "learning_rate": 0.000285679012345679, | |
| "loss": 0.5697, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.4311111111111111, | |
| "grad_norm": 0.21074801683425903, | |
| "learning_rate": 0.00028543209876543204, | |
| "loss": 0.5749, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.43333333333333335, | |
| "grad_norm": 0.16077393293380737, | |
| "learning_rate": 0.00028518518518518514, | |
| "loss": 0.5637, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.43555555555555553, | |
| "grad_norm": 0.1925012469291687, | |
| "learning_rate": 0.00028493827160493824, | |
| "loss": 0.5521, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.43777777777777777, | |
| "grad_norm": 0.22952796518802643, | |
| "learning_rate": 0.00028469135802469133, | |
| "loss": 0.4862, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.17839300632476807, | |
| "learning_rate": 0.0002844444444444444, | |
| "loss": 0.4975, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.44222222222222224, | |
| "grad_norm": 0.18458259105682373, | |
| "learning_rate": 0.0002841975308641975, | |
| "loss": 0.4283, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.16126206517219543, | |
| "learning_rate": 0.0002839506172839506, | |
| "loss": 0.5694, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.44666666666666666, | |
| "grad_norm": 0.2904411256313324, | |
| "learning_rate": 0.00028370370370370366, | |
| "loss": 0.3765, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.4488888888888889, | |
| "grad_norm": 0.19960430264472961, | |
| "learning_rate": 0.00028345679012345675, | |
| "loss": 0.5403, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.45111111111111113, | |
| "grad_norm": 0.23420177400112152, | |
| "learning_rate": 0.00028320987654320985, | |
| "loss": 0.6315, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.4533333333333333, | |
| "grad_norm": 0.17232675850391388, | |
| "learning_rate": 0.00028296296296296294, | |
| "loss": 0.4387, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.45555555555555555, | |
| "grad_norm": 0.43902498483657837, | |
| "learning_rate": 0.00028271604938271604, | |
| "loss": 0.5478, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.4577777777777778, | |
| "grad_norm": 0.15370631217956543, | |
| "learning_rate": 0.0002824691358024691, | |
| "loss": 0.4061, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.13001392781734467, | |
| "learning_rate": 0.00028222222222222223, | |
| "loss": 0.396, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.4622222222222222, | |
| "grad_norm": 0.18333856761455536, | |
| "learning_rate": 0.00028197530864197527, | |
| "loss": 0.538, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.46444444444444444, | |
| "grad_norm": 0.2740779519081116, | |
| "learning_rate": 0.00028172839506172836, | |
| "loss": 0.6133, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.4666666666666667, | |
| "grad_norm": 0.17451998591423035, | |
| "learning_rate": 0.00028148148148148146, | |
| "loss": 0.488, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4688888888888889, | |
| "grad_norm": 0.2760486304759979, | |
| "learning_rate": 0.00028123456790123455, | |
| "loss": 0.5366, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.4711111111111111, | |
| "grad_norm": 0.16626881062984467, | |
| "learning_rate": 0.00028098765432098765, | |
| "loss": 0.4814, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.47333333333333333, | |
| "grad_norm": 0.2446630299091339, | |
| "learning_rate": 0.0002807407407407407, | |
| "loss": 0.6241, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.47555555555555556, | |
| "grad_norm": 0.18423357605934143, | |
| "learning_rate": 0.0002804938271604938, | |
| "loss": 0.4679, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.4777777777777778, | |
| "grad_norm": 0.19439800083637238, | |
| "learning_rate": 0.0002802469135802469, | |
| "loss": 0.489, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.17421875894069672, | |
| "learning_rate": 0.00028, | |
| "loss": 0.6016, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.4822222222222222, | |
| "grad_norm": 0.1359538435935974, | |
| "learning_rate": 0.00027975308641975307, | |
| "loss": 0.4532, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.48444444444444446, | |
| "grad_norm": 0.16284197568893433, | |
| "learning_rate": 0.00027950617283950617, | |
| "loss": 0.4448, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.4866666666666667, | |
| "grad_norm": 0.12663625180721283, | |
| "learning_rate": 0.00027925925925925926, | |
| "loss": 0.5962, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.4888888888888889, | |
| "grad_norm": 0.16653625667095184, | |
| "learning_rate": 0.0002790123456790123, | |
| "loss": 0.7413, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4911111111111111, | |
| "grad_norm": 0.2262076586484909, | |
| "learning_rate": 0.0002787654320987654, | |
| "loss": 0.4672, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.49333333333333335, | |
| "grad_norm": 0.1869400292634964, | |
| "learning_rate": 0.0002785185185185185, | |
| "loss": 0.3975, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.4955555555555556, | |
| "grad_norm": 0.15228945016860962, | |
| "learning_rate": 0.0002782716049382716, | |
| "loss": 0.5648, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.49777777777777776, | |
| "grad_norm": 0.12363588809967041, | |
| "learning_rate": 0.00027802469135802463, | |
| "loss": 0.4571, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.18033596873283386, | |
| "learning_rate": 0.0002777777777777778, | |
| "loss": 0.5628, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5022222222222222, | |
| "grad_norm": 0.16061805188655853, | |
| "learning_rate": 0.00027753086419753087, | |
| "loss": 0.491, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.5044444444444445, | |
| "grad_norm": 0.18852145969867706, | |
| "learning_rate": 0.0002772839506172839, | |
| "loss": 0.6697, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.5066666666666667, | |
| "grad_norm": 0.16807378828525543, | |
| "learning_rate": 0.000277037037037037, | |
| "loss": 0.4647, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.5088888888888888, | |
| "grad_norm": 0.1589706540107727, | |
| "learning_rate": 0.0002767901234567901, | |
| "loss": 0.6322, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.5111111111111111, | |
| "grad_norm": 0.13450591266155243, | |
| "learning_rate": 0.0002765432098765432, | |
| "loss": 0.7376, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5133333333333333, | |
| "grad_norm": 0.18603169918060303, | |
| "learning_rate": 0.00027629629629629624, | |
| "loss": 0.5751, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.5155555555555555, | |
| "grad_norm": 0.3561168313026428, | |
| "learning_rate": 0.00027604938271604934, | |
| "loss": 0.6167, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.5177777777777778, | |
| "grad_norm": 0.22319789230823517, | |
| "learning_rate": 0.00027580246913580243, | |
| "loss": 0.6282, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.15226209163665771, | |
| "learning_rate": 0.0002755555555555555, | |
| "loss": 0.5736, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.5222222222222223, | |
| "grad_norm": 0.15853361785411835, | |
| "learning_rate": 0.0002753086419753086, | |
| "loss": 0.4946, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.5244444444444445, | |
| "grad_norm": 0.1475980281829834, | |
| "learning_rate": 0.0002750617283950617, | |
| "loss": 0.4983, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.5266666666666666, | |
| "grad_norm": 0.1835266798734665, | |
| "learning_rate": 0.0002748148148148148, | |
| "loss": 0.7749, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.5288888888888889, | |
| "grad_norm": 0.12562094628810883, | |
| "learning_rate": 0.00027456790123456785, | |
| "loss": 0.4855, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.5311111111111111, | |
| "grad_norm": 0.31017449498176575, | |
| "learning_rate": 0.00027432098765432095, | |
| "loss": 0.4986, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.5333333333333333, | |
| "grad_norm": 0.13299651443958282, | |
| "learning_rate": 0.00027407407407407404, | |
| "loss": 0.5675, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5355555555555556, | |
| "grad_norm": 0.16204799711704254, | |
| "learning_rate": 0.00027382716049382714, | |
| "loss": 0.3745, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.5377777777777778, | |
| "grad_norm": 0.30393674969673157, | |
| "learning_rate": 0.00027358024691358023, | |
| "loss": 0.617, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.16176091134548187, | |
| "learning_rate": 0.00027333333333333333, | |
| "loss": 0.5888, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.5422222222222223, | |
| "grad_norm": 0.19906075298786163, | |
| "learning_rate": 0.0002730864197530864, | |
| "loss": 0.8738, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.5444444444444444, | |
| "grad_norm": 0.1559479832649231, | |
| "learning_rate": 0.00027283950617283946, | |
| "loss": 0.4154, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5466666666666666, | |
| "grad_norm": 0.23664313554763794, | |
| "learning_rate": 0.00027259259259259256, | |
| "loss": 0.5207, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.5488888888888889, | |
| "grad_norm": 0.1270369291305542, | |
| "learning_rate": 0.00027234567901234565, | |
| "loss": 0.5812, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.5511111111111111, | |
| "grad_norm": 0.18237239122390747, | |
| "learning_rate": 0.00027209876543209875, | |
| "loss": 0.6277, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.5533333333333333, | |
| "grad_norm": 0.12331552058458328, | |
| "learning_rate": 0.00027185185185185184, | |
| "loss": 0.4045, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 0.1962405890226364, | |
| "learning_rate": 0.0002716049382716049, | |
| "loss": 0.4407, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5577777777777778, | |
| "grad_norm": 0.12834173440933228, | |
| "learning_rate": 0.00027135802469135803, | |
| "loss": 0.5736, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.1871601790189743, | |
| "learning_rate": 0.0002711111111111111, | |
| "loss": 0.597, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.5622222222222222, | |
| "grad_norm": 0.1498972624540329, | |
| "learning_rate": 0.00027086419753086417, | |
| "loss": 0.4776, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.5644444444444444, | |
| "grad_norm": 0.202190563082695, | |
| "learning_rate": 0.00027061728395061727, | |
| "loss": 0.5531, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.5666666666666667, | |
| "grad_norm": 0.17293180525302887, | |
| "learning_rate": 0.00027037037037037036, | |
| "loss": 0.4757, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5688888888888889, | |
| "grad_norm": 0.20435073971748352, | |
| "learning_rate": 0.00027012345679012346, | |
| "loss": 0.5344, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.5711111111111111, | |
| "grad_norm": 0.23538696765899658, | |
| "learning_rate": 0.0002698765432098765, | |
| "loss": 0.6067, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.5733333333333334, | |
| "grad_norm": 0.16673316061496735, | |
| "learning_rate": 0.0002696296296296296, | |
| "loss": 0.4677, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.5755555555555556, | |
| "grad_norm": 0.1636734902858734, | |
| "learning_rate": 0.0002693827160493827, | |
| "loss": 0.5181, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.5777777777777777, | |
| "grad_norm": 0.17571769654750824, | |
| "learning_rate": 0.0002691358024691358, | |
| "loss": 0.5128, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.21826326847076416, | |
| "learning_rate": 0.0002688888888888889, | |
| "loss": 0.4491, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.5822222222222222, | |
| "grad_norm": 0.2198294699192047, | |
| "learning_rate": 0.00026864197530864197, | |
| "loss": 0.5419, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.5844444444444444, | |
| "grad_norm": 0.22619126737117767, | |
| "learning_rate": 0.000268395061728395, | |
| "loss": 0.6673, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.5866666666666667, | |
| "grad_norm": 0.16405685245990753, | |
| "learning_rate": 0.0002681481481481481, | |
| "loss": 0.7017, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.5888888888888889, | |
| "grad_norm": 0.15559597313404083, | |
| "learning_rate": 0.0002679012345679012, | |
| "loss": 0.5741, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5911111111111111, | |
| "grad_norm": 0.2267346978187561, | |
| "learning_rate": 0.0002676543209876543, | |
| "loss": 0.4673, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.5933333333333334, | |
| "grad_norm": 0.17021146416664124, | |
| "learning_rate": 0.0002674074074074074, | |
| "loss": 0.5276, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.5955555555555555, | |
| "grad_norm": 0.16342853009700775, | |
| "learning_rate": 0.00026716049382716043, | |
| "loss": 0.6414, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.5977777777777777, | |
| "grad_norm": 0.17768199741840363, | |
| "learning_rate": 0.0002669135802469136, | |
| "loss": 0.5884, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.17032460868358612, | |
| "learning_rate": 0.0002666666666666666, | |
| "loss": 0.4814, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6022222222222222, | |
| "grad_norm": 0.1983184814453125, | |
| "learning_rate": 0.0002664197530864197, | |
| "loss": 0.4645, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.6044444444444445, | |
| "grad_norm": 0.1555837243795395, | |
| "learning_rate": 0.0002661728395061728, | |
| "loss": 0.7038, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.6066666666666667, | |
| "grad_norm": 0.14357896149158478, | |
| "learning_rate": 0.0002659259259259259, | |
| "loss": 0.5283, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.6088888888888889, | |
| "grad_norm": 0.19277998805046082, | |
| "learning_rate": 0.000265679012345679, | |
| "loss": 0.5303, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.6111111111111112, | |
| "grad_norm": 0.18067921698093414, | |
| "learning_rate": 0.00026543209876543205, | |
| "loss": 0.569, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6133333333333333, | |
| "grad_norm": 0.2134593278169632, | |
| "learning_rate": 0.00026518518518518514, | |
| "loss": 0.5626, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.6155555555555555, | |
| "grad_norm": 0.18603608012199402, | |
| "learning_rate": 0.00026493827160493824, | |
| "loss": 0.4404, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.6177777777777778, | |
| "grad_norm": 0.20176568627357483, | |
| "learning_rate": 0.00026469135802469133, | |
| "loss": 0.4687, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.2177278995513916, | |
| "learning_rate": 0.00026444444444444443, | |
| "loss": 0.4948, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.6222222222222222, | |
| "grad_norm": 0.20735816657543182, | |
| "learning_rate": 0.0002641975308641975, | |
| "loss": 0.4512, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.6244444444444445, | |
| "grad_norm": 0.19933420419692993, | |
| "learning_rate": 0.0002639506172839506, | |
| "loss": 0.5725, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.6266666666666667, | |
| "grad_norm": 0.13915471732616425, | |
| "learning_rate": 0.00026370370370370366, | |
| "loss": 0.5257, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.6288888888888889, | |
| "grad_norm": 0.12906256318092346, | |
| "learning_rate": 0.00026345679012345675, | |
| "loss": 0.5034, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.6311111111111111, | |
| "grad_norm": 0.21937648952007294, | |
| "learning_rate": 0.00026320987654320985, | |
| "loss": 0.5848, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.6333333333333333, | |
| "grad_norm": 0.13235728442668915, | |
| "learning_rate": 0.00026296296296296294, | |
| "loss": 0.5744, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.6355555555555555, | |
| "grad_norm": 0.19538620114326477, | |
| "learning_rate": 0.00026271604938271604, | |
| "loss": 0.4948, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.6377777777777778, | |
| "grad_norm": 0.1299598067998886, | |
| "learning_rate": 0.00026246913580246913, | |
| "loss": 0.4548, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.12741626799106598, | |
| "learning_rate": 0.00026222222222222223, | |
| "loss": 0.5207, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.6422222222222222, | |
| "grad_norm": 0.1801108419895172, | |
| "learning_rate": 0.00026197530864197527, | |
| "loss": 0.542, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.6444444444444445, | |
| "grad_norm": 0.16658473014831543, | |
| "learning_rate": 0.00026172839506172837, | |
| "loss": 0.5773, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6466666666666666, | |
| "grad_norm": 0.16049516201019287, | |
| "learning_rate": 0.00026148148148148146, | |
| "loss": 0.4965, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.6488888888888888, | |
| "grad_norm": 0.258699893951416, | |
| "learning_rate": 0.00026123456790123456, | |
| "loss": 0.6893, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.6511111111111111, | |
| "grad_norm": 0.2080325484275818, | |
| "learning_rate": 0.0002609876543209876, | |
| "loss": 0.6101, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.6533333333333333, | |
| "grad_norm": 0.16091585159301758, | |
| "learning_rate": 0.0002607407407407407, | |
| "loss": 0.5637, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.6555555555555556, | |
| "grad_norm": 0.270709365606308, | |
| "learning_rate": 0.00026049382716049384, | |
| "loss": 0.4344, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.6577777777777778, | |
| "grad_norm": 0.14196595549583435, | |
| "learning_rate": 0.0002602469135802469, | |
| "loss": 0.4826, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.1903015822172165, | |
| "learning_rate": 0.00026, | |
| "loss": 0.4584, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.6622222222222223, | |
| "grad_norm": 0.1537121832370758, | |
| "learning_rate": 0.00025975308641975307, | |
| "loss": 0.4276, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.6644444444444444, | |
| "grad_norm": 0.2700488865375519, | |
| "learning_rate": 0.00025950617283950617, | |
| "loss": 0.6081, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.1334371268749237, | |
| "learning_rate": 0.0002592592592592592, | |
| "loss": 0.3704, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6688888888888889, | |
| "grad_norm": 0.18560272455215454, | |
| "learning_rate": 0.0002590123456790123, | |
| "loss": 0.7141, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.6711111111111111, | |
| "grad_norm": 0.16761454939842224, | |
| "learning_rate": 0.0002587654320987654, | |
| "loss": 0.6264, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.6733333333333333, | |
| "grad_norm": 0.28938424587249756, | |
| "learning_rate": 0.0002585185185185185, | |
| "loss": 0.6142, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.6755555555555556, | |
| "grad_norm": 0.2370755970478058, | |
| "learning_rate": 0.0002582716049382716, | |
| "loss": 0.591, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.6777777777777778, | |
| "grad_norm": 0.14213621616363525, | |
| "learning_rate": 0.0002580246913580247, | |
| "loss": 0.4651, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.17966413497924805, | |
| "learning_rate": 0.0002577777777777778, | |
| "loss": 0.516, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.6822222222222222, | |
| "grad_norm": 0.1060803085565567, | |
| "learning_rate": 0.0002575308641975308, | |
| "loss": 0.468, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.6844444444444444, | |
| "grad_norm": 0.15440872311592102, | |
| "learning_rate": 0.0002572839506172839, | |
| "loss": 0.6035, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.6866666666666666, | |
| "grad_norm": 0.1453942060470581, | |
| "learning_rate": 0.000257037037037037, | |
| "loss": 0.5136, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.6888888888888889, | |
| "grad_norm": 0.18055713176727295, | |
| "learning_rate": 0.0002567901234567901, | |
| "loss": 0.5357, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6911111111111111, | |
| "grad_norm": 0.13261505961418152, | |
| "learning_rate": 0.0002565432098765432, | |
| "loss": 0.4666, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.6933333333333334, | |
| "grad_norm": 0.18729257583618164, | |
| "learning_rate": 0.00025629629629629624, | |
| "loss": 0.51, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.6955555555555556, | |
| "grad_norm": 0.2084057778120041, | |
| "learning_rate": 0.0002560493827160494, | |
| "loss": 0.6109, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.6977777777777778, | |
| "grad_norm": 0.14162084460258484, | |
| "learning_rate": 0.00025580246913580243, | |
| "loss": 0.5889, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.143287792801857, | |
| "learning_rate": 0.00025555555555555553, | |
| "loss": 0.4492, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7022222222222222, | |
| "grad_norm": 0.17587856948375702, | |
| "learning_rate": 0.0002553086419753086, | |
| "loss": 0.4669, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.7044444444444444, | |
| "grad_norm": 0.12248572707176208, | |
| "learning_rate": 0.0002550617283950617, | |
| "loss": 0.5852, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.7066666666666667, | |
| "grad_norm": 0.1443788856267929, | |
| "learning_rate": 0.0002548148148148148, | |
| "loss": 0.5865, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.7088888888888889, | |
| "grad_norm": 0.17955581843852997, | |
| "learning_rate": 0.00025456790123456785, | |
| "loss": 0.4901, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.7111111111111111, | |
| "grad_norm": 0.1374693363904953, | |
| "learning_rate": 0.00025432098765432095, | |
| "loss": 0.4654, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.7133333333333334, | |
| "grad_norm": 0.16100934147834778, | |
| "learning_rate": 0.00025407407407407404, | |
| "loss": 0.5871, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.7155555555555555, | |
| "grad_norm": 0.11821124702692032, | |
| "learning_rate": 0.00025382716049382714, | |
| "loss": 0.4874, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.7177777777777777, | |
| "grad_norm": 0.13106414675712585, | |
| "learning_rate": 0.00025358024691358023, | |
| "loss": 0.4548, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.1800754815340042, | |
| "learning_rate": 0.00025333333333333333, | |
| "loss": 0.628, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.7222222222222222, | |
| "grad_norm": 0.13882426917552948, | |
| "learning_rate": 0.0002530864197530864, | |
| "loss": 0.5583, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7244444444444444, | |
| "grad_norm": 0.12302982807159424, | |
| "learning_rate": 0.00025283950617283947, | |
| "loss": 0.487, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.7266666666666667, | |
| "grad_norm": 0.1677716076374054, | |
| "learning_rate": 0.00025259259259259256, | |
| "loss": 0.4769, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.7288888888888889, | |
| "grad_norm": 0.12374882400035858, | |
| "learning_rate": 0.00025234567901234566, | |
| "loss": 0.5071, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.7311111111111112, | |
| "grad_norm": 0.1606243997812271, | |
| "learning_rate": 0.00025209876543209875, | |
| "loss": 0.4161, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.7333333333333333, | |
| "grad_norm": 0.17513667047023773, | |
| "learning_rate": 0.0002518518518518518, | |
| "loss": 0.5917, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7355555555555555, | |
| "grad_norm": 0.1605701744556427, | |
| "learning_rate": 0.00025160493827160494, | |
| "loss": 0.6128, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.7377777777777778, | |
| "grad_norm": 0.15330223739147186, | |
| "learning_rate": 0.00025135802469135804, | |
| "loss": 0.5156, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.12888985872268677, | |
| "learning_rate": 0.0002511111111111111, | |
| "loss": 0.4776, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.7422222222222222, | |
| "grad_norm": 0.15710951387882233, | |
| "learning_rate": 0.00025086419753086417, | |
| "loss": 0.4939, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.7444444444444445, | |
| "grad_norm": 0.1283513307571411, | |
| "learning_rate": 0.00025061728395061727, | |
| "loss": 0.5321, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.7466666666666667, | |
| "grad_norm": 0.1467665582895279, | |
| "learning_rate": 0.00025037037037037036, | |
| "loss": 0.5827, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.7488888888888889, | |
| "grad_norm": 0.1438836306333542, | |
| "learning_rate": 0.0002501234567901234, | |
| "loss": 0.7135, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.7511111111111111, | |
| "grad_norm": 0.1473802775144577, | |
| "learning_rate": 0.0002498765432098765, | |
| "loss": 0.4535, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.7533333333333333, | |
| "grad_norm": 0.17046429216861725, | |
| "learning_rate": 0.00024962962962962965, | |
| "loss": 0.5923, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.7555555555555555, | |
| "grad_norm": 0.11560607701539993, | |
| "learning_rate": 0.0002493827160493827, | |
| "loss": 0.563, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7577777777777778, | |
| "grad_norm": 0.1518397182226181, | |
| "learning_rate": 0.0002491358024691358, | |
| "loss": 0.5356, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.2090211659669876, | |
| "learning_rate": 0.0002488888888888889, | |
| "loss": 0.5605, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.7622222222222222, | |
| "grad_norm": 0.13951998949050903, | |
| "learning_rate": 0.000248641975308642, | |
| "loss": 0.632, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.7644444444444445, | |
| "grad_norm": 0.15694496035575867, | |
| "learning_rate": 0.000248395061728395, | |
| "loss": 0.6715, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.7666666666666667, | |
| "grad_norm": 0.16268084943294525, | |
| "learning_rate": 0.0002481481481481481, | |
| "loss": 0.4463, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7688888888888888, | |
| "grad_norm": 0.21926285326480865, | |
| "learning_rate": 0.0002479012345679012, | |
| "loss": 0.5846, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.7711111111111111, | |
| "grad_norm": 0.2690124213695526, | |
| "learning_rate": 0.0002476543209876543, | |
| "loss": 0.5963, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.7733333333333333, | |
| "grad_norm": 0.21665515005588531, | |
| "learning_rate": 0.0002474074074074074, | |
| "loss": 0.4725, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.7755555555555556, | |
| "grad_norm": 0.10690062493085861, | |
| "learning_rate": 0.0002471604938271605, | |
| "loss": 0.5844, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.7777777777777778, | |
| "grad_norm": 0.16186372935771942, | |
| "learning_rate": 0.0002469135802469136, | |
| "loss": 0.4665, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.17135635018348694, | |
| "learning_rate": 0.0002466666666666666, | |
| "loss": 0.4256, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.7822222222222223, | |
| "grad_norm": 0.22984741628170013, | |
| "learning_rate": 0.0002464197530864197, | |
| "loss": 0.5772, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.7844444444444445, | |
| "grad_norm": 0.12628202140331268, | |
| "learning_rate": 0.0002461728395061728, | |
| "loss": 0.4654, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.7866666666666666, | |
| "grad_norm": 0.10722055286169052, | |
| "learning_rate": 0.0002459259259259259, | |
| "loss": 0.4817, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.7888888888888889, | |
| "grad_norm": 0.18734730780124664, | |
| "learning_rate": 0.000245679012345679, | |
| "loss": 0.4892, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.7911111111111111, | |
| "grad_norm": 0.1820126324892044, | |
| "learning_rate": 0.00024543209876543205, | |
| "loss": 0.4987, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.7933333333333333, | |
| "grad_norm": 0.1389104425907135, | |
| "learning_rate": 0.0002451851851851852, | |
| "loss": 0.5332, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.7955555555555556, | |
| "grad_norm": 0.1396469920873642, | |
| "learning_rate": 0.00024493827160493824, | |
| "loss": 0.4166, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.7977777777777778, | |
| "grad_norm": 0.1483020782470703, | |
| "learning_rate": 0.00024469135802469133, | |
| "loss": 0.4869, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.23203179240226746, | |
| "learning_rate": 0.00024444444444444443, | |
| "loss": 0.7642, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8022222222222222, | |
| "grad_norm": 0.19104941189289093, | |
| "learning_rate": 0.0002441975308641975, | |
| "loss": 0.5143, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.8044444444444444, | |
| "grad_norm": 0.13722693920135498, | |
| "learning_rate": 0.00024395061728395062, | |
| "loss": 0.5746, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.8066666666666666, | |
| "grad_norm": 0.13355255126953125, | |
| "learning_rate": 0.0002437037037037037, | |
| "loss": 0.4699, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.8088888888888889, | |
| "grad_norm": 0.13934768736362457, | |
| "learning_rate": 0.00024345679012345678, | |
| "loss": 0.637, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.8111111111111111, | |
| "grad_norm": 0.1919456571340561, | |
| "learning_rate": 0.00024320987654320985, | |
| "loss": 0.5514, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.8133333333333334, | |
| "grad_norm": 0.1947246789932251, | |
| "learning_rate": 0.00024296296296296295, | |
| "loss": 0.5247, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.8155555555555556, | |
| "grad_norm": 0.20515243709087372, | |
| "learning_rate": 0.00024271604938271601, | |
| "loss": 0.6269, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.8177777777777778, | |
| "grad_norm": 0.13479793071746826, | |
| "learning_rate": 0.0002424691358024691, | |
| "loss": 0.4613, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.1373935043811798, | |
| "learning_rate": 0.0002422222222222222, | |
| "loss": 0.4497, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.8222222222222222, | |
| "grad_norm": 0.18367323279380798, | |
| "learning_rate": 0.00024197530864197527, | |
| "loss": 0.4536, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.8244444444444444, | |
| "grad_norm": 0.23737306892871857, | |
| "learning_rate": 0.0002417283950617284, | |
| "loss": 0.5727, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.8266666666666667, | |
| "grad_norm": 0.1628907322883606, | |
| "learning_rate": 0.00024148148148148146, | |
| "loss": 0.4713, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.8288888888888889, | |
| "grad_norm": 0.18385647237300873, | |
| "learning_rate": 0.00024123456790123456, | |
| "loss": 0.4263, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.8311111111111111, | |
| "grad_norm": 0.16067151725292206, | |
| "learning_rate": 0.00024098765432098763, | |
| "loss": 0.5655, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.18861906230449677, | |
| "learning_rate": 0.00024074074074074072, | |
| "loss": 0.4556, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.8355555555555556, | |
| "grad_norm": 0.160345658659935, | |
| "learning_rate": 0.0002404938271604938, | |
| "loss": 0.4554, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.8377777777777777, | |
| "grad_norm": 0.21110886335372925, | |
| "learning_rate": 0.00024024691358024688, | |
| "loss": 0.4128, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.13896746933460236, | |
| "learning_rate": 0.00023999999999999998, | |
| "loss": 0.4896, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.8422222222222222, | |
| "grad_norm": 0.18256375193595886, | |
| "learning_rate": 0.00023975308641975305, | |
| "loss": 0.5534, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.8444444444444444, | |
| "grad_norm": 0.1329539567232132, | |
| "learning_rate": 0.00023950617283950617, | |
| "loss": 0.6054, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8466666666666667, | |
| "grad_norm": 0.1916404366493225, | |
| "learning_rate": 0.00023925925925925924, | |
| "loss": 0.5104, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.8488888888888889, | |
| "grad_norm": 0.18349631130695343, | |
| "learning_rate": 0.00023901234567901233, | |
| "loss": 0.5076, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.8511111111111112, | |
| "grad_norm": 0.19793935120105743, | |
| "learning_rate": 0.0002387654320987654, | |
| "loss": 0.6826, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.8533333333333334, | |
| "grad_norm": 0.1868397742509842, | |
| "learning_rate": 0.0002385185185185185, | |
| "loss": 0.5886, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.8555555555555555, | |
| "grad_norm": 0.15153376758098602, | |
| "learning_rate": 0.0002382716049382716, | |
| "loss": 0.3526, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.8577777777777778, | |
| "grad_norm": 0.13153988122940063, | |
| "learning_rate": 0.00023802469135802466, | |
| "loss": 0.4554, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.12699386477470398, | |
| "learning_rate": 0.00023777777777777775, | |
| "loss": 0.6665, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.8622222222222222, | |
| "grad_norm": 0.2641882598400116, | |
| "learning_rate": 0.00023753086419753082, | |
| "loss": 0.5676, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.8644444444444445, | |
| "grad_norm": 0.12574434280395508, | |
| "learning_rate": 0.00023728395061728394, | |
| "loss": 0.5497, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.8666666666666667, | |
| "grad_norm": 0.171394944190979, | |
| "learning_rate": 0.000237037037037037, | |
| "loss": 0.5884, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8688888888888889, | |
| "grad_norm": 0.22201262414455414, | |
| "learning_rate": 0.0002367901234567901, | |
| "loss": 0.5841, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.8711111111111111, | |
| "grad_norm": 0.15717816352844238, | |
| "learning_rate": 0.0002365432098765432, | |
| "loss": 0.421, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.8733333333333333, | |
| "grad_norm": 0.16818246245384216, | |
| "learning_rate": 0.00023629629629629627, | |
| "loss": 0.4381, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.8755555555555555, | |
| "grad_norm": 0.22220592200756073, | |
| "learning_rate": 0.00023604938271604937, | |
| "loss": 0.5435, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.8777777777777778, | |
| "grad_norm": 0.1777927428483963, | |
| "learning_rate": 0.00023580246913580243, | |
| "loss": 0.4472, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.2609533965587616, | |
| "learning_rate": 0.00023555555555555553, | |
| "loss": 0.4603, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.8822222222222222, | |
| "grad_norm": 0.2210220843553543, | |
| "learning_rate": 0.0002353086419753086, | |
| "loss": 0.5801, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.8844444444444445, | |
| "grad_norm": 0.20401951670646667, | |
| "learning_rate": 0.00023506172839506172, | |
| "loss": 0.594, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.8866666666666667, | |
| "grad_norm": 0.15360064804553986, | |
| "learning_rate": 0.0002348148148148148, | |
| "loss": 0.518, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 0.15452717244625092, | |
| "learning_rate": 0.00023456790123456788, | |
| "loss": 0.4914, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8911111111111111, | |
| "grad_norm": 0.13852454721927643, | |
| "learning_rate": 0.00023432098765432098, | |
| "loss": 0.3475, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.8933333333333333, | |
| "grad_norm": 0.17671892046928406, | |
| "learning_rate": 0.00023407407407407405, | |
| "loss": 0.5573, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.8955555555555555, | |
| "grad_norm": 0.31494978070259094, | |
| "learning_rate": 0.00023382716049382714, | |
| "loss": 0.5598, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.8977777777777778, | |
| "grad_norm": 0.13132472336292267, | |
| "learning_rate": 0.0002335802469135802, | |
| "loss": 0.4545, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.7265273332595825, | |
| "learning_rate": 0.0002333333333333333, | |
| "loss": 0.6252, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9022222222222223, | |
| "grad_norm": 0.20422174036502838, | |
| "learning_rate": 0.00023308641975308637, | |
| "loss": 0.5854, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.9044444444444445, | |
| "grad_norm": 0.14237628877162933, | |
| "learning_rate": 0.0002328395061728395, | |
| "loss": 0.3958, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.9066666666666666, | |
| "grad_norm": 0.19874413311481476, | |
| "learning_rate": 0.0002325925925925926, | |
| "loss": 0.6152, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.9088888888888889, | |
| "grad_norm": 0.1612745225429535, | |
| "learning_rate": 0.00023234567901234566, | |
| "loss": 0.5447, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.9111111111111111, | |
| "grad_norm": 0.14159739017486572, | |
| "learning_rate": 0.00023209876543209875, | |
| "loss": 0.6015, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.9133333333333333, | |
| "grad_norm": 0.13011768460273743, | |
| "learning_rate": 0.00023185185185185182, | |
| "loss": 0.525, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.9155555555555556, | |
| "grad_norm": 0.19322142004966736, | |
| "learning_rate": 0.00023160493827160492, | |
| "loss": 0.5399, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.9177777777777778, | |
| "grad_norm": 0.6533721089363098, | |
| "learning_rate": 0.00023135802469135798, | |
| "loss": 0.3829, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.17026424407958984, | |
| "learning_rate": 0.00023111111111111108, | |
| "loss": 0.6095, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.9222222222222223, | |
| "grad_norm": 0.20249877870082855, | |
| "learning_rate": 0.0002308641975308642, | |
| "loss": 0.5871, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.9244444444444444, | |
| "grad_norm": 0.14929679036140442, | |
| "learning_rate": 0.00023061728395061727, | |
| "loss": 0.5386, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.9266666666666666, | |
| "grad_norm": 0.10986840724945068, | |
| "learning_rate": 0.00023037037037037036, | |
| "loss": 0.4834, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.9288888888888889, | |
| "grad_norm": 0.14803917706012726, | |
| "learning_rate": 0.00023012345679012343, | |
| "loss": 0.4813, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.9311111111111111, | |
| "grad_norm": 0.16032618284225464, | |
| "learning_rate": 0.00022987654320987653, | |
| "loss": 0.4703, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.9333333333333333, | |
| "grad_norm": 0.1804179549217224, | |
| "learning_rate": 0.0002296296296296296, | |
| "loss": 0.5443, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9355555555555556, | |
| "grad_norm": 0.2294391244649887, | |
| "learning_rate": 0.0002293827160493827, | |
| "loss": 0.5518, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.9377777777777778, | |
| "grad_norm": 0.13405844569206238, | |
| "learning_rate": 0.00022913580246913579, | |
| "loss": 0.5114, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.3711307942867279, | |
| "learning_rate": 0.00022888888888888885, | |
| "loss": 0.6694, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.9422222222222222, | |
| "grad_norm": 0.17901724576950073, | |
| "learning_rate": 0.00022864197530864198, | |
| "loss": 0.4346, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.9444444444444444, | |
| "grad_norm": 0.22841103374958038, | |
| "learning_rate": 0.00022839506172839504, | |
| "loss": 0.5202, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9466666666666667, | |
| "grad_norm": 0.26436498761177063, | |
| "learning_rate": 0.00022814814814814814, | |
| "loss": 0.5507, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.9488888888888889, | |
| "grad_norm": 0.1576514095067978, | |
| "learning_rate": 0.0002279012345679012, | |
| "loss": 0.3687, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.9511111111111111, | |
| "grad_norm": 0.16450360417366028, | |
| "learning_rate": 0.0002276543209876543, | |
| "loss": 0.533, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.9533333333333334, | |
| "grad_norm": 0.12933999300003052, | |
| "learning_rate": 0.00022740740740740737, | |
| "loss": 0.53, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.9555555555555556, | |
| "grad_norm": 0.2541653513908386, | |
| "learning_rate": 0.00022716049382716047, | |
| "loss": 0.4178, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9577777777777777, | |
| "grad_norm": 0.21268242597579956, | |
| "learning_rate": 0.00022691358024691356, | |
| "loss": 0.4404, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.18136733770370483, | |
| "learning_rate": 0.00022666666666666663, | |
| "loss": 0.6026, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.9622222222222222, | |
| "grad_norm": 0.18260523676872253, | |
| "learning_rate": 0.00022641975308641975, | |
| "loss": 0.6733, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.9644444444444444, | |
| "grad_norm": 0.30254852771759033, | |
| "learning_rate": 0.00022617283950617282, | |
| "loss": 0.6309, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.9666666666666667, | |
| "grad_norm": 0.18107043206691742, | |
| "learning_rate": 0.00022592592592592591, | |
| "loss": 0.7921, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.9688888888888889, | |
| "grad_norm": 0.16891856491565704, | |
| "learning_rate": 0.00022567901234567898, | |
| "loss": 0.5121, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.9711111111111111, | |
| "grad_norm": 0.14732027053833008, | |
| "learning_rate": 0.00022543209876543208, | |
| "loss": 0.4805, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.9733333333333334, | |
| "grad_norm": 0.19609785079956055, | |
| "learning_rate": 0.00022518518518518517, | |
| "loss": 0.5566, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.9755555555555555, | |
| "grad_norm": 0.11922861635684967, | |
| "learning_rate": 0.00022493827160493824, | |
| "loss": 0.4757, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.9777777777777777, | |
| "grad_norm": 0.18207553029060364, | |
| "learning_rate": 0.00022469135802469134, | |
| "loss": 0.6347, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 0.1859586536884308, | |
| "learning_rate": 0.0002244444444444444, | |
| "loss": 0.3895, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.9822222222222222, | |
| "grad_norm": 0.1757041960954666, | |
| "learning_rate": 0.00022419753086419753, | |
| "loss": 0.5824, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.9844444444444445, | |
| "grad_norm": 0.24221178889274597, | |
| "learning_rate": 0.0002239506172839506, | |
| "loss": 0.5045, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.9866666666666667, | |
| "grad_norm": 0.16379328072071075, | |
| "learning_rate": 0.0002237037037037037, | |
| "loss": 0.5463, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.9888888888888889, | |
| "grad_norm": 0.13701243698596954, | |
| "learning_rate": 0.00022345679012345678, | |
| "loss": 0.5069, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.9911111111111112, | |
| "grad_norm": 0.14070475101470947, | |
| "learning_rate": 0.00022320987654320985, | |
| "loss": 0.5496, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.9933333333333333, | |
| "grad_norm": 0.20788441598415375, | |
| "learning_rate": 0.00022296296296296295, | |
| "loss": 0.6053, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.9955555555555555, | |
| "grad_norm": 0.1296846717596054, | |
| "learning_rate": 0.00022271604938271602, | |
| "loss": 0.4171, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.9977777777777778, | |
| "grad_norm": 0.2014620155096054, | |
| "learning_rate": 0.0002224691358024691, | |
| "loss": 0.4917, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.1412200927734375, | |
| "learning_rate": 0.00022222222222222218, | |
| "loss": 0.4675, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.0022222222222221, | |
| "grad_norm": 0.15296180546283722, | |
| "learning_rate": 0.0002219753086419753, | |
| "loss": 0.4816, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.0044444444444445, | |
| "grad_norm": 0.1596982330083847, | |
| "learning_rate": 0.0002217283950617284, | |
| "loss": 0.5158, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.0066666666666666, | |
| "grad_norm": 0.20045016705989838, | |
| "learning_rate": 0.00022148148148148146, | |
| "loss": 0.4579, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.008888888888889, | |
| "grad_norm": 0.21337562799453735, | |
| "learning_rate": 0.00022123456790123456, | |
| "loss": 0.577, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.011111111111111, | |
| "grad_norm": 0.14974337816238403, | |
| "learning_rate": 0.00022098765432098763, | |
| "loss": 0.61, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.0133333333333334, | |
| "grad_norm": 0.1343904733657837, | |
| "learning_rate": 0.00022074074074074072, | |
| "loss": 0.5267, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.0155555555555555, | |
| "grad_norm": 0.1901727318763733, | |
| "learning_rate": 0.0002204938271604938, | |
| "loss": 0.3962, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.0177777777777777, | |
| "grad_norm": 0.14229881763458252, | |
| "learning_rate": 0.00022024691358024689, | |
| "loss": 0.4632, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.21129313111305237, | |
| "learning_rate": 0.00021999999999999995, | |
| "loss": 0.5091, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.0222222222222221, | |
| "grad_norm": 0.20234957337379456, | |
| "learning_rate": 0.00021975308641975308, | |
| "loss": 0.6278, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.0244444444444445, | |
| "grad_norm": 0.14051935076713562, | |
| "learning_rate": 0.00021950617283950617, | |
| "loss": 0.5114, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.0266666666666666, | |
| "grad_norm": 0.1425868421792984, | |
| "learning_rate": 0.00021925925925925924, | |
| "loss": 0.461, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.028888888888889, | |
| "grad_norm": 0.3667563796043396, | |
| "learning_rate": 0.00021901234567901233, | |
| "loss": 0.4416, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.031111111111111, | |
| "grad_norm": 0.22489885985851288, | |
| "learning_rate": 0.0002187654320987654, | |
| "loss": 0.6011, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.0333333333333334, | |
| "grad_norm": 0.21429428458213806, | |
| "learning_rate": 0.0002185185185185185, | |
| "loss": 0.5969, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.0355555555555556, | |
| "grad_norm": 0.17646625638008118, | |
| "learning_rate": 0.00021827160493827157, | |
| "loss": 0.4538, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.0377777777777777, | |
| "grad_norm": 0.17430634796619415, | |
| "learning_rate": 0.00021802469135802466, | |
| "loss": 0.6081, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.23083315789699554, | |
| "learning_rate": 0.00021777777777777778, | |
| "loss": 0.4757, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.0422222222222222, | |
| "grad_norm": 0.21813291311264038, | |
| "learning_rate": 0.00021753086419753085, | |
| "loss": 0.5772, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.0444444444444445, | |
| "grad_norm": 0.17318132519721985, | |
| "learning_rate": 0.00021728395061728395, | |
| "loss": 0.4643, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0466666666666666, | |
| "grad_norm": 0.2051350325345993, | |
| "learning_rate": 0.00021703703703703701, | |
| "loss": 0.5271, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.048888888888889, | |
| "grad_norm": 0.15854017436504364, | |
| "learning_rate": 0.0002167901234567901, | |
| "loss": 0.5484, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.051111111111111, | |
| "grad_norm": 0.21494866907596588, | |
| "learning_rate": 0.00021654320987654318, | |
| "loss": 0.463, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.0533333333333332, | |
| "grad_norm": 0.15565277636051178, | |
| "learning_rate": 0.00021629629629629627, | |
| "loss": 0.4496, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.0555555555555556, | |
| "grad_norm": 0.2026614099740982, | |
| "learning_rate": 0.00021604938271604937, | |
| "loss": 0.5573, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0577777777777777, | |
| "grad_norm": 0.2115136682987213, | |
| "learning_rate": 0.00021580246913580244, | |
| "loss": 0.5417, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 0.17910204827785492, | |
| "learning_rate": 0.00021555555555555556, | |
| "loss": 0.4107, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.0622222222222222, | |
| "grad_norm": 0.16029846668243408, | |
| "learning_rate": 0.00021530864197530863, | |
| "loss": 0.4197, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.0644444444444445, | |
| "grad_norm": 0.13962920010089874, | |
| "learning_rate": 0.00021506172839506172, | |
| "loss": 0.5397, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.0666666666666667, | |
| "grad_norm": 0.19910384714603424, | |
| "learning_rate": 0.0002148148148148148, | |
| "loss": 0.5013, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.068888888888889, | |
| "grad_norm": 0.18806932866573334, | |
| "learning_rate": 0.00021456790123456788, | |
| "loss": 0.5681, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.0711111111111111, | |
| "grad_norm": 0.16994863748550415, | |
| "learning_rate": 0.00021432098765432095, | |
| "loss": 0.4911, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.0733333333333333, | |
| "grad_norm": 0.1846705973148346, | |
| "learning_rate": 0.00021407407407407405, | |
| "loss": 0.5247, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.0755555555555556, | |
| "grad_norm": 0.1335802674293518, | |
| "learning_rate": 0.00021382716049382714, | |
| "loss": 0.5972, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.0777777777777777, | |
| "grad_norm": 0.15970145165920258, | |
| "learning_rate": 0.0002135802469135802, | |
| "loss": 0.4708, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 0.18331503868103027, | |
| "learning_rate": 0.00021333333333333333, | |
| "loss": 0.5471, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.0822222222222222, | |
| "grad_norm": 0.1520104557275772, | |
| "learning_rate": 0.0002130864197530864, | |
| "loss": 0.4853, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.0844444444444445, | |
| "grad_norm": 0.16329914331436157, | |
| "learning_rate": 0.0002128395061728395, | |
| "loss": 0.4402, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.0866666666666667, | |
| "grad_norm": 0.15783226490020752, | |
| "learning_rate": 0.00021259259259259256, | |
| "loss": 0.4405, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.0888888888888888, | |
| "grad_norm": 0.1584683209657669, | |
| "learning_rate": 0.00021234567901234566, | |
| "loss": 0.3924, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.0911111111111111, | |
| "grad_norm": 0.16735412180423737, | |
| "learning_rate": 0.00021209876543209875, | |
| "loss": 0.5651, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.0933333333333333, | |
| "grad_norm": 0.20828117430210114, | |
| "learning_rate": 0.00021185185185185182, | |
| "loss": 0.524, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.0955555555555556, | |
| "grad_norm": 0.14907875657081604, | |
| "learning_rate": 0.00021160493827160492, | |
| "loss": 0.6376, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.0977777777777777, | |
| "grad_norm": 0.2043023705482483, | |
| "learning_rate": 0.00021135802469135799, | |
| "loss": 0.4262, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 0.14947763085365295, | |
| "learning_rate": 0.0002111111111111111, | |
| "loss": 0.4732, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.1022222222222222, | |
| "grad_norm": 0.1837744265794754, | |
| "learning_rate": 0.00021086419753086418, | |
| "loss": 0.5766, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.1044444444444443, | |
| "grad_norm": 0.1411743313074112, | |
| "learning_rate": 0.00021061728395061727, | |
| "loss": 0.4675, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.1066666666666667, | |
| "grad_norm": 0.14611735939979553, | |
| "learning_rate": 0.00021037037037037037, | |
| "loss": 0.4986, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.1088888888888888, | |
| "grad_norm": 0.143850177526474, | |
| "learning_rate": 0.00021012345679012343, | |
| "loss": 0.441, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 0.14008420705795288, | |
| "learning_rate": 0.00020987654320987653, | |
| "loss": 0.6538, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.1133333333333333, | |
| "grad_norm": 0.1524328887462616, | |
| "learning_rate": 0.0002096296296296296, | |
| "loss": 0.4386, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.1155555555555556, | |
| "grad_norm": 0.1267358660697937, | |
| "learning_rate": 0.0002093827160493827, | |
| "loss": 0.4863, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.1177777777777778, | |
| "grad_norm": 0.1797122359275818, | |
| "learning_rate": 0.00020913580246913576, | |
| "loss": 0.4287, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.2134157121181488, | |
| "learning_rate": 0.00020888888888888888, | |
| "loss": 0.4902, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.1222222222222222, | |
| "grad_norm": 0.20062175393104553, | |
| "learning_rate": 0.00020864197530864198, | |
| "loss": 0.5243, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.1244444444444444, | |
| "grad_norm": 0.19181513786315918, | |
| "learning_rate": 0.00020839506172839505, | |
| "loss": 0.4947, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.1266666666666667, | |
| "grad_norm": 0.17851033806800842, | |
| "learning_rate": 0.00020814814814814814, | |
| "loss": 0.5148, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.1288888888888888, | |
| "grad_norm": 0.19144859910011292, | |
| "learning_rate": 0.0002079012345679012, | |
| "loss": 0.5808, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.1311111111111112, | |
| "grad_norm": 0.17474748194217682, | |
| "learning_rate": 0.0002076543209876543, | |
| "loss": 0.5784, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.1333333333333333, | |
| "grad_norm": 0.575061023235321, | |
| "learning_rate": 0.00020740740740740737, | |
| "loss": 0.621, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.1355555555555557, | |
| "grad_norm": 0.24973663687705994, | |
| "learning_rate": 0.00020716049382716047, | |
| "loss": 0.5519, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.1377777777777778, | |
| "grad_norm": 0.17157737910747528, | |
| "learning_rate": 0.00020691358024691354, | |
| "loss": 0.4849, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.1400000000000001, | |
| "grad_norm": 0.23158717155456543, | |
| "learning_rate": 0.00020666666666666666, | |
| "loss": 0.448, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.1422222222222222, | |
| "grad_norm": 0.14639237523078918, | |
| "learning_rate": 0.00020641975308641975, | |
| "loss": 0.4809, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.1444444444444444, | |
| "grad_norm": 0.23363938927650452, | |
| "learning_rate": 0.00020617283950617282, | |
| "loss": 0.4925, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.1466666666666667, | |
| "grad_norm": 0.32770803570747375, | |
| "learning_rate": 0.00020592592592592592, | |
| "loss": 0.5383, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.1488888888888888, | |
| "grad_norm": 0.14956258237361908, | |
| "learning_rate": 0.00020567901234567898, | |
| "loss": 0.4385, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.1511111111111112, | |
| "grad_norm": 0.1975625902414322, | |
| "learning_rate": 0.00020543209876543208, | |
| "loss": 0.4663, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.1533333333333333, | |
| "grad_norm": 0.12537258863449097, | |
| "learning_rate": 0.00020518518518518515, | |
| "loss": 0.4136, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.1555555555555554, | |
| "grad_norm": 0.15473900735378265, | |
| "learning_rate": 0.00020493827160493824, | |
| "loss": 0.5035, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.1577777777777778, | |
| "grad_norm": 0.2243552953004837, | |
| "learning_rate": 0.00020469135802469136, | |
| "loss": 0.5115, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 0.13791115581989288, | |
| "learning_rate": 0.00020444444444444443, | |
| "loss": 0.5034, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.1622222222222223, | |
| "grad_norm": 0.17481450736522675, | |
| "learning_rate": 0.00020419753086419753, | |
| "loss": 0.5384, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.1644444444444444, | |
| "grad_norm": 0.1647374927997589, | |
| "learning_rate": 0.0002039506172839506, | |
| "loss": 0.4774, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.1666666666666667, | |
| "grad_norm": 0.17841963469982147, | |
| "learning_rate": 0.0002037037037037037, | |
| "loss": 0.5758, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.1688888888888889, | |
| "grad_norm": 0.19152382016181946, | |
| "learning_rate": 0.00020345679012345676, | |
| "loss": 0.4626, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.1711111111111112, | |
| "grad_norm": 0.17557677626609802, | |
| "learning_rate": 0.00020320987654320985, | |
| "loss": 0.5215, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.1733333333333333, | |
| "grad_norm": 0.19200007617473602, | |
| "learning_rate": 0.00020296296296296295, | |
| "loss": 0.5668, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.1755555555555555, | |
| "grad_norm": 0.16062043607234955, | |
| "learning_rate": 0.00020271604938271602, | |
| "loss": 0.4757, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.1777777777777778, | |
| "grad_norm": 0.18774579465389252, | |
| "learning_rate": 0.00020246913580246914, | |
| "loss": 0.624, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 0.18138159811496735, | |
| "learning_rate": 0.0002022222222222222, | |
| "loss": 0.4251, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.1822222222222223, | |
| "grad_norm": 0.25393691658973694, | |
| "learning_rate": 0.0002019753086419753, | |
| "loss": 0.5845, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.1844444444444444, | |
| "grad_norm": 0.18442478775978088, | |
| "learning_rate": 0.00020172839506172837, | |
| "loss": 0.4026, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.1866666666666668, | |
| "grad_norm": 0.1802750676870346, | |
| "learning_rate": 0.00020148148148148147, | |
| "loss": 0.4954, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.1888888888888889, | |
| "grad_norm": 0.1883816123008728, | |
| "learning_rate": 0.00020123456790123456, | |
| "loss": 0.7166, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.1911111111111112, | |
| "grad_norm": 0.2110777497291565, | |
| "learning_rate": 0.00020098765432098763, | |
| "loss": 0.5168, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.1933333333333334, | |
| "grad_norm": 0.1972721815109253, | |
| "learning_rate": 0.00020074074074074072, | |
| "loss": 0.427, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.1955555555555555, | |
| "grad_norm": 0.1931055188179016, | |
| "learning_rate": 0.0002004938271604938, | |
| "loss": 0.4154, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.1977777777777778, | |
| "grad_norm": 0.20624488592147827, | |
| "learning_rate": 0.00020024691358024691, | |
| "loss": 0.4764, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.19730456173419952, | |
| "learning_rate": 0.00019999999999999998, | |
| "loss": 0.5672, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.2022222222222223, | |
| "grad_norm": 0.19493892788887024, | |
| "learning_rate": 0.00019975308641975308, | |
| "loss": 0.488, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.2044444444444444, | |
| "grad_norm": 0.18839912116527557, | |
| "learning_rate": 0.00019950617283950615, | |
| "loss": 0.5617, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.2066666666666666, | |
| "grad_norm": 0.23850169777870178, | |
| "learning_rate": 0.00019925925925925924, | |
| "loss": 0.614, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.208888888888889, | |
| "grad_norm": 0.18365950882434845, | |
| "learning_rate": 0.00019901234567901234, | |
| "loss": 0.4778, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.211111111111111, | |
| "grad_norm": 0.17773182690143585, | |
| "learning_rate": 0.0001987654320987654, | |
| "loss": 0.512, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.2133333333333334, | |
| "grad_norm": 0.21308952569961548, | |
| "learning_rate": 0.0001985185185185185, | |
| "loss": 0.6476, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.2155555555555555, | |
| "grad_norm": 0.1732819527387619, | |
| "learning_rate": 0.00019827160493827157, | |
| "loss": 0.4544, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.2177777777777778, | |
| "grad_norm": 0.15607161819934845, | |
| "learning_rate": 0.0001980246913580247, | |
| "loss": 0.5787, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 0.16454316675662994, | |
| "learning_rate": 0.00019777777777777776, | |
| "loss": 0.4792, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.2222222222222223, | |
| "grad_norm": 0.16066400706768036, | |
| "learning_rate": 0.00019753086419753085, | |
| "loss": 0.4656, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.2244444444444444, | |
| "grad_norm": 0.15481796860694885, | |
| "learning_rate": 0.00019728395061728395, | |
| "loss": 0.47, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.2266666666666666, | |
| "grad_norm": 0.21378105878829956, | |
| "learning_rate": 0.00019703703703703702, | |
| "loss": 0.4756, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.228888888888889, | |
| "grad_norm": 0.20864234864711761, | |
| "learning_rate": 0.0001967901234567901, | |
| "loss": 0.4278, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.231111111111111, | |
| "grad_norm": 0.16556324064731598, | |
| "learning_rate": 0.00019654320987654318, | |
| "loss": 0.4792, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.2333333333333334, | |
| "grad_norm": 0.19661420583724976, | |
| "learning_rate": 0.00019629629629629627, | |
| "loss": 0.5862, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.2355555555555555, | |
| "grad_norm": 0.2666536271572113, | |
| "learning_rate": 0.00019604938271604934, | |
| "loss": 0.4847, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.2377777777777779, | |
| "grad_norm": 0.1885252147912979, | |
| "learning_rate": 0.00019580246913580246, | |
| "loss": 0.4694, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 0.19279441237449646, | |
| "learning_rate": 0.00019555555555555556, | |
| "loss": 0.567, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.2422222222222223, | |
| "grad_norm": 0.24274209141731262, | |
| "learning_rate": 0.00019530864197530863, | |
| "loss": 0.4831, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.2444444444444445, | |
| "grad_norm": 0.1520586609840393, | |
| "learning_rate": 0.00019506172839506172, | |
| "loss": 0.4659, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.2466666666666666, | |
| "grad_norm": 0.4970318675041199, | |
| "learning_rate": 0.0001948148148148148, | |
| "loss": 0.5206, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.248888888888889, | |
| "grad_norm": 0.18483081459999084, | |
| "learning_rate": 0.00019456790123456789, | |
| "loss": 0.4844, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.251111111111111, | |
| "grad_norm": 0.17052306234836578, | |
| "learning_rate": 0.00019432098765432095, | |
| "loss": 0.4774, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.2533333333333334, | |
| "grad_norm": 0.30775099992752075, | |
| "learning_rate": 0.00019407407407407405, | |
| "loss": 0.4726, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.2555555555555555, | |
| "grad_norm": 0.18412767350673676, | |
| "learning_rate": 0.00019382716049382712, | |
| "loss": 0.616, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.2577777777777777, | |
| "grad_norm": 0.24758651852607727, | |
| "learning_rate": 0.00019358024691358024, | |
| "loss": 0.4505, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 0.15375128388404846, | |
| "learning_rate": 0.00019333333333333333, | |
| "loss": 0.4015, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.2622222222222224, | |
| "grad_norm": 0.17346544563770294, | |
| "learning_rate": 0.0001930864197530864, | |
| "loss": 0.6291, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.2644444444444445, | |
| "grad_norm": 0.17779450118541718, | |
| "learning_rate": 0.0001928395061728395, | |
| "loss": 0.5275, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.2666666666666666, | |
| "grad_norm": 0.2299584001302719, | |
| "learning_rate": 0.00019259259259259257, | |
| "loss": 0.479, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.268888888888889, | |
| "grad_norm": 0.19577407836914062, | |
| "learning_rate": 0.00019234567901234566, | |
| "loss": 0.5587, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.271111111111111, | |
| "grad_norm": 0.2726730406284332, | |
| "learning_rate": 0.00019209876543209873, | |
| "loss": 0.4941, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.2733333333333334, | |
| "grad_norm": 0.18272462487220764, | |
| "learning_rate": 0.00019185185185185182, | |
| "loss": 0.4434, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.2755555555555556, | |
| "grad_norm": 0.2076050490140915, | |
| "learning_rate": 0.00019160493827160495, | |
| "loss": 0.5415, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.2777777777777777, | |
| "grad_norm": 0.18124337494373322, | |
| "learning_rate": 0.00019135802469135801, | |
| "loss": 0.4524, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 0.21853291988372803, | |
| "learning_rate": 0.0001911111111111111, | |
| "loss": 0.4859, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.2822222222222222, | |
| "grad_norm": 0.20311839878559113, | |
| "learning_rate": 0.00019086419753086418, | |
| "loss": 0.3825, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.2844444444444445, | |
| "grad_norm": 0.2179584801197052, | |
| "learning_rate": 0.00019061728395061727, | |
| "loss": 0.5508, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.2866666666666666, | |
| "grad_norm": 0.18006446957588196, | |
| "learning_rate": 0.00019037037037037034, | |
| "loss": 0.5068, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.2888888888888888, | |
| "grad_norm": 0.3354756832122803, | |
| "learning_rate": 0.00019012345679012344, | |
| "loss": 0.5284, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.291111111111111, | |
| "grad_norm": 0.20429521799087524, | |
| "learning_rate": 0.00018987654320987653, | |
| "loss": 0.42, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.2933333333333334, | |
| "grad_norm": 0.2871708869934082, | |
| "learning_rate": 0.0001896296296296296, | |
| "loss": 0.6303, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.2955555555555556, | |
| "grad_norm": 0.19127105176448822, | |
| "learning_rate": 0.00018938271604938272, | |
| "loss": 0.4727, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.2977777777777777, | |
| "grad_norm": 0.22170588374137878, | |
| "learning_rate": 0.0001891358024691358, | |
| "loss": 0.5222, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 0.2183113992214203, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 0.5839, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.3022222222222222, | |
| "grad_norm": 0.18902623653411865, | |
| "learning_rate": 0.00018864197530864195, | |
| "loss": 0.487, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.3044444444444445, | |
| "grad_norm": 0.19840282201766968, | |
| "learning_rate": 0.00018839506172839505, | |
| "loss": 0.6223, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.3066666666666666, | |
| "grad_norm": 0.19085773825645447, | |
| "learning_rate": 0.00018814814814814814, | |
| "loss": 0.714, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.3088888888888888, | |
| "grad_norm": 0.21159861981868744, | |
| "learning_rate": 0.0001879012345679012, | |
| "loss": 0.6117, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.3111111111111111, | |
| "grad_norm": 0.1644948571920395, | |
| "learning_rate": 0.0001876543209876543, | |
| "loss": 0.43, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.3133333333333335, | |
| "grad_norm": 0.1644945591688156, | |
| "learning_rate": 0.00018740740740740737, | |
| "loss": 0.522, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.3155555555555556, | |
| "grad_norm": 0.16714589297771454, | |
| "learning_rate": 0.0001871604938271605, | |
| "loss": 0.4637, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.3177777777777777, | |
| "grad_norm": 0.17184139788150787, | |
| "learning_rate": 0.00018691358024691356, | |
| "loss": 0.5053, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 0.22288395464420319, | |
| "learning_rate": 0.00018666666666666666, | |
| "loss": 0.4802, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.3222222222222222, | |
| "grad_norm": 0.22077727317810059, | |
| "learning_rate": 0.00018641975308641973, | |
| "loss": 0.5234, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.3244444444444445, | |
| "grad_norm": 0.24655035138130188, | |
| "learning_rate": 0.00018617283950617282, | |
| "loss": 0.605, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.3266666666666667, | |
| "grad_norm": 0.1810985654592514, | |
| "learning_rate": 0.00018592592592592592, | |
| "loss": 0.4612, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.3288888888888888, | |
| "grad_norm": 0.22254349291324615, | |
| "learning_rate": 0.00018567901234567899, | |
| "loss": 0.5434, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.3311111111111111, | |
| "grad_norm": 0.2537665367126465, | |
| "learning_rate": 0.00018543209876543208, | |
| "loss": 0.5439, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.19868597388267517, | |
| "learning_rate": 0.00018518518518518515, | |
| "loss": 0.5914, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.3355555555555556, | |
| "grad_norm": 0.17404809594154358, | |
| "learning_rate": 0.00018493827160493827, | |
| "loss": 0.5246, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.3377777777777777, | |
| "grad_norm": 0.2385467290878296, | |
| "learning_rate": 0.00018469135802469134, | |
| "loss": 0.511, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 0.1850982904434204, | |
| "learning_rate": 0.00018444444444444443, | |
| "loss": 0.5489, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.3422222222222222, | |
| "grad_norm": 0.22791941463947296, | |
| "learning_rate": 0.00018419753086419753, | |
| "loss": 0.4131, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.3444444444444446, | |
| "grad_norm": 0.18677116930484772, | |
| "learning_rate": 0.0001839506172839506, | |
| "loss": 0.4958, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.3466666666666667, | |
| "grad_norm": 0.200926274061203, | |
| "learning_rate": 0.0001837037037037037, | |
| "loss": 0.4987, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.3488888888888888, | |
| "grad_norm": 0.1807965189218521, | |
| "learning_rate": 0.00018345679012345676, | |
| "loss": 0.4659, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.3511111111111112, | |
| "grad_norm": 0.24230799078941345, | |
| "learning_rate": 0.00018320987654320986, | |
| "loss": 0.5353, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.3533333333333333, | |
| "grad_norm": 0.18938079476356506, | |
| "learning_rate": 0.00018296296296296292, | |
| "loss": 0.5847, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.3555555555555556, | |
| "grad_norm": 0.2192828208208084, | |
| "learning_rate": 0.00018271604938271605, | |
| "loss": 0.5538, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.3577777777777778, | |
| "grad_norm": 0.17889611423015594, | |
| "learning_rate": 0.00018246913580246914, | |
| "loss": 0.4422, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.3599999999999999, | |
| "grad_norm": 0.17387720942497253, | |
| "learning_rate": 0.0001822222222222222, | |
| "loss": 0.5156, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.3622222222222222, | |
| "grad_norm": 0.2507362961769104, | |
| "learning_rate": 0.0001819753086419753, | |
| "loss": 0.4544, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.3644444444444446, | |
| "grad_norm": 0.1838897168636322, | |
| "learning_rate": 0.00018172839506172837, | |
| "loss": 0.6998, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.3666666666666667, | |
| "grad_norm": 0.18592320382595062, | |
| "learning_rate": 0.00018148148148148147, | |
| "loss": 0.498, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.3688888888888888, | |
| "grad_norm": 0.20797160267829895, | |
| "learning_rate": 0.00018123456790123454, | |
| "loss": 0.4756, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.3711111111111112, | |
| "grad_norm": 0.19926156103610992, | |
| "learning_rate": 0.00018098765432098763, | |
| "loss": 0.5444, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.3733333333333333, | |
| "grad_norm": 0.34967702627182007, | |
| "learning_rate": 0.00018074074074074075, | |
| "loss": 0.692, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.3755555555555556, | |
| "grad_norm": 0.2074728012084961, | |
| "learning_rate": 0.00018049382716049382, | |
| "loss": 0.4972, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.3777777777777778, | |
| "grad_norm": 0.15851154923439026, | |
| "learning_rate": 0.00018024691358024692, | |
| "loss": 0.4242, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 0.1698676347732544, | |
| "learning_rate": 0.00017999999999999998, | |
| "loss": 0.4802, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.3822222222222222, | |
| "grad_norm": 0.18896076083183289, | |
| "learning_rate": 0.00017975308641975308, | |
| "loss": 0.4807, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.3844444444444444, | |
| "grad_norm": 0.17187227308750153, | |
| "learning_rate": 0.00017950617283950615, | |
| "loss": 0.4615, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.3866666666666667, | |
| "grad_norm": 0.2240261435508728, | |
| "learning_rate": 0.00017925925925925924, | |
| "loss": 0.4862, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 0.26844900846481323, | |
| "learning_rate": 0.0001790123456790123, | |
| "loss": 0.4504, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.3911111111111112, | |
| "grad_norm": 0.2594693601131439, | |
| "learning_rate": 0.0001787654320987654, | |
| "loss": 0.4518, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.3933333333333333, | |
| "grad_norm": 0.1672193855047226, | |
| "learning_rate": 0.00017851851851851853, | |
| "loss": 0.6343, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.3955555555555557, | |
| "grad_norm": 0.22965764999389648, | |
| "learning_rate": 0.0001782716049382716, | |
| "loss": 0.5414, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.3977777777777778, | |
| "grad_norm": 0.19246816635131836, | |
| "learning_rate": 0.0001780246913580247, | |
| "loss": 0.4507, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 0.2101546674966812, | |
| "learning_rate": 0.00017777777777777776, | |
| "loss": 0.4744, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.4022222222222223, | |
| "grad_norm": 0.2266424000263214, | |
| "learning_rate": 0.00017753086419753085, | |
| "loss": 0.461, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.4044444444444444, | |
| "grad_norm": 0.19170035421848297, | |
| "learning_rate": 0.00017728395061728392, | |
| "loss": 0.4783, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.4066666666666667, | |
| "grad_norm": 0.14152956008911133, | |
| "learning_rate": 0.00017703703703703702, | |
| "loss": 0.4776, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.4088888888888889, | |
| "grad_norm": 0.19685135781764984, | |
| "learning_rate": 0.0001767901234567901, | |
| "loss": 0.6729, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.411111111111111, | |
| "grad_norm": 0.16328732669353485, | |
| "learning_rate": 0.00017654320987654318, | |
| "loss": 0.3997, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.4133333333333333, | |
| "grad_norm": 0.15920418500900269, | |
| "learning_rate": 0.0001762962962962963, | |
| "loss": 0.5095, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.4155555555555557, | |
| "grad_norm": 0.18346062302589417, | |
| "learning_rate": 0.00017604938271604937, | |
| "loss": 0.5207, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.4177777777777778, | |
| "grad_norm": 0.21879293024539948, | |
| "learning_rate": 0.00017580246913580247, | |
| "loss": 0.6211, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 0.1826939582824707, | |
| "learning_rate": 0.00017555555555555553, | |
| "loss": 0.6162, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.4222222222222223, | |
| "grad_norm": 0.20309215784072876, | |
| "learning_rate": 0.00017530864197530863, | |
| "loss": 0.5145, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.4244444444444444, | |
| "grad_norm": 0.16178536415100098, | |
| "learning_rate": 0.00017506172839506172, | |
| "loss": 0.4698, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.4266666666666667, | |
| "grad_norm": 0.4259434640407562, | |
| "learning_rate": 0.0001748148148148148, | |
| "loss": 0.5742, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.4288888888888889, | |
| "grad_norm": 0.20084147155284882, | |
| "learning_rate": 0.0001745679012345679, | |
| "loss": 0.3503, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.431111111111111, | |
| "grad_norm": 0.1782788336277008, | |
| "learning_rate": 0.00017432098765432096, | |
| "loss": 0.4734, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.4333333333333333, | |
| "grad_norm": 0.19422651827335358, | |
| "learning_rate": 0.00017407407407407408, | |
| "loss": 0.5768, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.4355555555555555, | |
| "grad_norm": 0.2092157006263733, | |
| "learning_rate": 0.00017382716049382715, | |
| "loss": 0.4841, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.4377777777777778, | |
| "grad_norm": 0.6713825464248657, | |
| "learning_rate": 0.00017358024691358024, | |
| "loss": 0.5932, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 0.21312807500362396, | |
| "learning_rate": 0.0001733333333333333, | |
| "loss": 0.4797, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.4422222222222223, | |
| "grad_norm": 0.17761079967021942, | |
| "learning_rate": 0.0001730864197530864, | |
| "loss": 0.5327, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.4444444444444444, | |
| "grad_norm": 0.16172362864017487, | |
| "learning_rate": 0.0001728395061728395, | |
| "loss": 0.6512, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.4466666666666668, | |
| "grad_norm": 0.17499731481075287, | |
| "learning_rate": 0.00017259259259259257, | |
| "loss": 0.4079, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.448888888888889, | |
| "grad_norm": 0.19369076192378998, | |
| "learning_rate": 0.00017234567901234566, | |
| "loss": 0.4968, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.451111111111111, | |
| "grad_norm": 0.20914605259895325, | |
| "learning_rate": 0.00017209876543209873, | |
| "loss": 0.5744, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.4533333333333334, | |
| "grad_norm": 0.17183391749858856, | |
| "learning_rate": 0.00017185185185185185, | |
| "loss": 0.4904, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.4555555555555555, | |
| "grad_norm": 0.25038477778434753, | |
| "learning_rate": 0.00017160493827160492, | |
| "loss": 0.4505, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.4577777777777778, | |
| "grad_norm": 0.17621774971485138, | |
| "learning_rate": 0.00017135802469135802, | |
| "loss": 0.439, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "grad_norm": 0.1552869975566864, | |
| "learning_rate": 0.0001711111111111111, | |
| "loss": 0.4068, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.462222222222222, | |
| "grad_norm": 0.15896369516849518, | |
| "learning_rate": 0.00017086419753086418, | |
| "loss": 0.415, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.4644444444444444, | |
| "grad_norm": 0.22240282595157623, | |
| "learning_rate": 0.00017061728395061727, | |
| "loss": 0.4637, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.4666666666666668, | |
| "grad_norm": 0.1821317970752716, | |
| "learning_rate": 0.00017037037037037034, | |
| "loss": 0.5947, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.468888888888889, | |
| "grad_norm": 0.21780848503112793, | |
| "learning_rate": 0.00017012345679012344, | |
| "loss": 0.5874, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.471111111111111, | |
| "grad_norm": 0.2360154539346695, | |
| "learning_rate": 0.0001698765432098765, | |
| "loss": 0.5563, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.4733333333333334, | |
| "grad_norm": 0.22776220738887787, | |
| "learning_rate": 0.00016962962962962963, | |
| "loss": 0.4085, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.4755555555555555, | |
| "grad_norm": 0.1937866508960724, | |
| "learning_rate": 0.00016938271604938272, | |
| "loss": 0.6489, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.4777777777777779, | |
| "grad_norm": 0.2154289335012436, | |
| "learning_rate": 0.0001691358024691358, | |
| "loss": 0.4344, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 0.24307891726493835, | |
| "learning_rate": 0.00016888888888888889, | |
| "loss": 0.4118, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.482222222222222, | |
| "grad_norm": 0.2223687618970871, | |
| "learning_rate": 0.00016864197530864195, | |
| "loss": 0.4854, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.4844444444444445, | |
| "grad_norm": 0.48969876766204834, | |
| "learning_rate": 0.00016839506172839505, | |
| "loss": 0.5636, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.4866666666666668, | |
| "grad_norm": 0.19435164332389832, | |
| "learning_rate": 0.00016814814814814812, | |
| "loss": 0.5319, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.488888888888889, | |
| "grad_norm": 0.22977349162101746, | |
| "learning_rate": 0.0001679012345679012, | |
| "loss": 0.5222, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.491111111111111, | |
| "grad_norm": 0.17664363980293274, | |
| "learning_rate": 0.00016765432098765433, | |
| "loss": 0.4165, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.4933333333333334, | |
| "grad_norm": 0.24705873429775238, | |
| "learning_rate": 0.0001674074074074074, | |
| "loss": 0.4729, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.4955555555555555, | |
| "grad_norm": 0.23530292510986328, | |
| "learning_rate": 0.0001671604938271605, | |
| "loss": 0.5862, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.4977777777777779, | |
| "grad_norm": 0.187184676527977, | |
| "learning_rate": 0.00016691358024691357, | |
| "loss": 0.5, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 0.19930264353752136, | |
| "learning_rate": 0.00016666666666666666, | |
| "loss": 0.4601, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.5022222222222221, | |
| "grad_norm": 0.35129958391189575, | |
| "learning_rate": 0.00016641975308641973, | |
| "loss": 0.4551, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.5044444444444445, | |
| "grad_norm": 0.23577232658863068, | |
| "learning_rate": 0.00016617283950617282, | |
| "loss": 0.5754, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.5066666666666668, | |
| "grad_norm": 0.2713683545589447, | |
| "learning_rate": 0.0001659259259259259, | |
| "loss": 0.5194, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.508888888888889, | |
| "grad_norm": 0.20194876194000244, | |
| "learning_rate": 0.000165679012345679, | |
| "loss": 0.6008, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.511111111111111, | |
| "grad_norm": 0.2077755331993103, | |
| "learning_rate": 0.0001654320987654321, | |
| "loss": 0.5293, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.5133333333333332, | |
| "grad_norm": 0.2098294347524643, | |
| "learning_rate": 0.00016518518518518518, | |
| "loss": 0.6053, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.5155555555555555, | |
| "grad_norm": 0.177157923579216, | |
| "learning_rate": 0.00016493827160493827, | |
| "loss": 0.4967, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.517777777777778, | |
| "grad_norm": 0.17908713221549988, | |
| "learning_rate": 0.00016469135802469134, | |
| "loss": 0.4675, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 0.2160455584526062, | |
| "learning_rate": 0.00016444444444444444, | |
| "loss": 0.5882, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.5222222222222221, | |
| "grad_norm": 0.2013448178768158, | |
| "learning_rate": 0.0001641975308641975, | |
| "loss": 0.5368, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.5244444444444445, | |
| "grad_norm": 0.2998005747795105, | |
| "learning_rate": 0.0001639506172839506, | |
| "loss": 0.4577, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.5266666666666666, | |
| "grad_norm": 0.2456233948469162, | |
| "learning_rate": 0.0001637037037037037, | |
| "loss": 0.5114, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.528888888888889, | |
| "grad_norm": 0.19369514286518097, | |
| "learning_rate": 0.00016345679012345676, | |
| "loss": 0.4345, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.531111111111111, | |
| "grad_norm": 0.17276284098625183, | |
| "learning_rate": 0.00016320987654320988, | |
| "loss": 0.4565, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.5333333333333332, | |
| "grad_norm": 0.18681663274765015, | |
| "learning_rate": 0.00016296296296296295, | |
| "loss": 0.484, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.5355555555555556, | |
| "grad_norm": 0.17169655859470367, | |
| "learning_rate": 0.00016271604938271605, | |
| "loss": 0.4054, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.537777777777778, | |
| "grad_norm": 0.20770080387592316, | |
| "learning_rate": 0.00016246913580246912, | |
| "loss": 0.5247, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.22684913873672485, | |
| "learning_rate": 0.0001622222222222222, | |
| "loss": 0.4636, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.5422222222222222, | |
| "grad_norm": 0.19948816299438477, | |
| "learning_rate": 0.0001619753086419753, | |
| "loss": 0.5112, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.5444444444444443, | |
| "grad_norm": 0.16264313459396362, | |
| "learning_rate": 0.00016172839506172837, | |
| "loss": 0.4769, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.5466666666666666, | |
| "grad_norm": 0.15728119015693665, | |
| "learning_rate": 0.00016148148148148147, | |
| "loss": 0.4527, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.548888888888889, | |
| "grad_norm": 0.17388519644737244, | |
| "learning_rate": 0.00016123456790123454, | |
| "loss": 0.5092, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.551111111111111, | |
| "grad_norm": 0.23285317420959473, | |
| "learning_rate": 0.00016098765432098766, | |
| "loss": 0.6197, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.5533333333333332, | |
| "grad_norm": 0.1925249993801117, | |
| "learning_rate": 0.00016074074074074073, | |
| "loss": 0.6325, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.5555555555555556, | |
| "grad_norm": 0.17383582890033722, | |
| "learning_rate": 0.00016049382716049382, | |
| "loss": 0.4196, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.557777777777778, | |
| "grad_norm": 0.16742292046546936, | |
| "learning_rate": 0.00016024691358024692, | |
| "loss": 0.4626, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 0.1801236867904663, | |
| "learning_rate": 0.00015999999999999999, | |
| "loss": 0.4156, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.5622222222222222, | |
| "grad_norm": 0.23592707514762878, | |
| "learning_rate": 0.00015975308641975308, | |
| "loss": 0.5288, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.5644444444444443, | |
| "grad_norm": 0.2852647304534912, | |
| "learning_rate": 0.00015950617283950615, | |
| "loss": 0.4533, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.5666666666666667, | |
| "grad_norm": 0.20041382312774658, | |
| "learning_rate": 0.00015925925925925924, | |
| "loss": 0.5854, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.568888888888889, | |
| "grad_norm": 0.21541517972946167, | |
| "learning_rate": 0.0001590123456790123, | |
| "loss": 0.5857, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.5711111111111111, | |
| "grad_norm": 0.20702879130840302, | |
| "learning_rate": 0.00015876543209876543, | |
| "loss": 0.5015, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.5733333333333333, | |
| "grad_norm": 0.19715112447738647, | |
| "learning_rate": 0.0001585185185185185, | |
| "loss": 0.6239, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.5755555555555556, | |
| "grad_norm": 0.185335174202919, | |
| "learning_rate": 0.0001582716049382716, | |
| "loss": 0.4359, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.5777777777777777, | |
| "grad_norm": 0.2287055402994156, | |
| "learning_rate": 0.0001580246913580247, | |
| "loss": 0.5241, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 0.430734783411026, | |
| "learning_rate": 0.00015777777777777776, | |
| "loss": 0.5139, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.5822222222222222, | |
| "grad_norm": 0.5821022391319275, | |
| "learning_rate": 0.00015753086419753086, | |
| "loss": 0.4922, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.5844444444444443, | |
| "grad_norm": 0.3152739107608795, | |
| "learning_rate": 0.00015728395061728392, | |
| "loss": 0.5786, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.5866666666666667, | |
| "grad_norm": 0.1754438877105713, | |
| "learning_rate": 0.00015703703703703702, | |
| "loss": 0.455, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.588888888888889, | |
| "grad_norm": 0.19354331493377686, | |
| "learning_rate": 0.0001567901234567901, | |
| "loss": 0.4787, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.5911111111111111, | |
| "grad_norm": 0.2226487696170807, | |
| "learning_rate": 0.0001565432098765432, | |
| "loss": 0.5323, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.5933333333333333, | |
| "grad_norm": 0.2240685522556305, | |
| "learning_rate": 0.0001562962962962963, | |
| "loss": 0.6267, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.5955555555555554, | |
| "grad_norm": 0.19735385477542877, | |
| "learning_rate": 0.00015604938271604937, | |
| "loss": 0.4954, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.5977777777777777, | |
| "grad_norm": 0.1885635256767273, | |
| "learning_rate": 0.00015580246913580247, | |
| "loss": 0.4692, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 0.18214702606201172, | |
| "learning_rate": 0.00015555555555555554, | |
| "loss": 0.5039, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.6022222222222222, | |
| "grad_norm": 0.15900494158267975, | |
| "learning_rate": 0.00015530864197530863, | |
| "loss": 0.4295, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.6044444444444443, | |
| "grad_norm": 0.17227846384048462, | |
| "learning_rate": 0.0001550617283950617, | |
| "loss": 0.5682, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.6066666666666667, | |
| "grad_norm": 0.18904544413089752, | |
| "learning_rate": 0.0001548148148148148, | |
| "loss": 0.5475, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.608888888888889, | |
| "grad_norm": 0.17980565130710602, | |
| "learning_rate": 0.00015456790123456792, | |
| "loss": 0.5123, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.6111111111111112, | |
| "grad_norm": 0.19602270424365997, | |
| "learning_rate": 0.00015432098765432098, | |
| "loss": 0.5176, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.6133333333333333, | |
| "grad_norm": 0.2298801839351654, | |
| "learning_rate": 0.00015407407407407408, | |
| "loss": 0.5991, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.6155555555555554, | |
| "grad_norm": 0.1984095722436905, | |
| "learning_rate": 0.00015382716049382715, | |
| "loss": 0.518, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.6177777777777778, | |
| "grad_norm": 0.18328291177749634, | |
| "learning_rate": 0.00015358024691358024, | |
| "loss": 0.496, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "grad_norm": 0.3043038547039032, | |
| "learning_rate": 0.0001533333333333333, | |
| "loss": 0.4729, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.6222222222222222, | |
| "grad_norm": 0.19305519759655, | |
| "learning_rate": 0.0001530864197530864, | |
| "loss": 0.5184, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.6244444444444444, | |
| "grad_norm": 0.22729025781154633, | |
| "learning_rate": 0.00015283950617283947, | |
| "loss": 0.3421, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.6266666666666667, | |
| "grad_norm": 0.21574439108371735, | |
| "learning_rate": 0.00015259259259259257, | |
| "loss": 0.5728, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.628888888888889, | |
| "grad_norm": 0.22035643458366394, | |
| "learning_rate": 0.0001523456790123457, | |
| "loss": 0.499, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.6311111111111112, | |
| "grad_norm": 0.18493935465812683, | |
| "learning_rate": 0.00015209876543209873, | |
| "loss": 0.5382, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.6333333333333333, | |
| "grad_norm": 0.2573682963848114, | |
| "learning_rate": 0.00015185185185185185, | |
| "loss": 0.5011, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.6355555555555554, | |
| "grad_norm": 0.20061856508255005, | |
| "learning_rate": 0.00015160493827160492, | |
| "loss": 0.4955, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.6377777777777778, | |
| "grad_norm": 0.19330166280269623, | |
| "learning_rate": 0.00015135802469135802, | |
| "loss": 0.4857, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.6400000000000001, | |
| "grad_norm": 0.18491677939891815, | |
| "learning_rate": 0.00015111111111111109, | |
| "loss": 0.5352, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.6422222222222222, | |
| "grad_norm": 0.15709461271762848, | |
| "learning_rate": 0.00015086419753086418, | |
| "loss": 0.6668, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.6444444444444444, | |
| "grad_norm": 0.24437206983566284, | |
| "learning_rate": 0.00015061728395061728, | |
| "loss": 0.5671, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.6466666666666665, | |
| "grad_norm": 0.2835857570171356, | |
| "learning_rate": 0.00015037037037037034, | |
| "loss": 0.6399, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.6488888888888888, | |
| "grad_norm": 0.21790558099746704, | |
| "learning_rate": 0.00015012345679012347, | |
| "loss": 0.5442, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.6511111111111112, | |
| "grad_norm": 0.2051643580198288, | |
| "learning_rate": 0.00014987654320987653, | |
| "loss": 0.4625, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.6533333333333333, | |
| "grad_norm": 0.15350790321826935, | |
| "learning_rate": 0.00014962962962962963, | |
| "loss": 0.329, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.6555555555555554, | |
| "grad_norm": 0.1980409175157547, | |
| "learning_rate": 0.0001493827160493827, | |
| "loss": 0.5567, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.6577777777777778, | |
| "grad_norm": 0.20699834823608398, | |
| "learning_rate": 0.0001491358024691358, | |
| "loss": 0.3992, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.6600000000000001, | |
| "grad_norm": 0.1647353172302246, | |
| "learning_rate": 0.00014888888888888886, | |
| "loss": 0.3794, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.6622222222222223, | |
| "grad_norm": 0.6979120373725891, | |
| "learning_rate": 0.00014864197530864196, | |
| "loss": 0.428, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.6644444444444444, | |
| "grad_norm": 0.26914483308792114, | |
| "learning_rate": 0.00014839506172839505, | |
| "loss": 0.508, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 0.19674748182296753, | |
| "learning_rate": 0.00014814814814814812, | |
| "loss": 0.5914, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.6688888888888889, | |
| "grad_norm": 0.18304680287837982, | |
| "learning_rate": 0.00014790123456790124, | |
| "loss": 0.4089, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.6711111111111112, | |
| "grad_norm": 0.15116271376609802, | |
| "learning_rate": 0.0001476543209876543, | |
| "loss": 0.4115, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.6733333333333333, | |
| "grad_norm": 0.20839999616146088, | |
| "learning_rate": 0.0001474074074074074, | |
| "loss": 0.4455, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.6755555555555555, | |
| "grad_norm": 0.1945442259311676, | |
| "learning_rate": 0.00014716049382716047, | |
| "loss": 0.5677, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.6777777777777778, | |
| "grad_norm": 0.19826267659664154, | |
| "learning_rate": 0.00014691358024691357, | |
| "loss": 0.6126, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.6800000000000002, | |
| "grad_norm": 0.19688139855861664, | |
| "learning_rate": 0.00014666666666666664, | |
| "loss": 0.4731, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.6822222222222223, | |
| "grad_norm": 0.28439095616340637, | |
| "learning_rate": 0.00014641975308641973, | |
| "loss": 0.4422, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 1.6844444444444444, | |
| "grad_norm": 0.32857492566108704, | |
| "learning_rate": 0.00014617283950617283, | |
| "loss": 0.5467, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.6866666666666665, | |
| "grad_norm": 0.4552757441997528, | |
| "learning_rate": 0.00014592592592592592, | |
| "loss": 0.5614, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 1.6888888888888889, | |
| "grad_norm": 0.20224252343177795, | |
| "learning_rate": 0.000145679012345679, | |
| "loss": 0.4798, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.6911111111111112, | |
| "grad_norm": 0.17604376375675201, | |
| "learning_rate": 0.00014543209876543208, | |
| "loss": 0.5737, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 1.6933333333333334, | |
| "grad_norm": 0.1598304808139801, | |
| "learning_rate": 0.00014518518518518518, | |
| "loss": 0.5929, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 1.6955555555555555, | |
| "grad_norm": 0.17656809091567993, | |
| "learning_rate": 0.00014493827160493825, | |
| "loss": 0.4156, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 1.6977777777777778, | |
| "grad_norm": 0.31885015964508057, | |
| "learning_rate": 0.00014469135802469134, | |
| "loss": 0.4468, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "grad_norm": 0.17707641422748566, | |
| "learning_rate": 0.0001444444444444444, | |
| "loss": 0.6977, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.7022222222222223, | |
| "grad_norm": 0.19356077909469604, | |
| "learning_rate": 0.00014419753086419753, | |
| "loss": 0.413, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 1.7044444444444444, | |
| "grad_norm": 0.21356233954429626, | |
| "learning_rate": 0.0001439506172839506, | |
| "loss": 0.5174, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 1.7066666666666666, | |
| "grad_norm": 0.22634132206439972, | |
| "learning_rate": 0.0001437037037037037, | |
| "loss": 0.4468, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 1.708888888888889, | |
| "grad_norm": 0.18362516164779663, | |
| "learning_rate": 0.00014345679012345676, | |
| "loss": 0.3897, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 1.7111111111111112, | |
| "grad_norm": 0.18277226388454437, | |
| "learning_rate": 0.00014320987654320986, | |
| "loss": 0.4883, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.7133333333333334, | |
| "grad_norm": 0.17673806846141815, | |
| "learning_rate": 0.00014296296296296295, | |
| "loss": 0.4351, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 1.7155555555555555, | |
| "grad_norm": 0.22419191896915436, | |
| "learning_rate": 0.00014271604938271602, | |
| "loss": 0.5539, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.7177777777777776, | |
| "grad_norm": 0.1831635683774948, | |
| "learning_rate": 0.00014246913580246912, | |
| "loss": 0.5799, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 0.1664152443408966, | |
| "learning_rate": 0.0001422222222222222, | |
| "loss": 0.5222, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.7222222222222223, | |
| "grad_norm": 0.25792166590690613, | |
| "learning_rate": 0.0001419753086419753, | |
| "loss": 0.5003, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.7244444444444444, | |
| "grad_norm": 0.20047883689403534, | |
| "learning_rate": 0.00014172839506172838, | |
| "loss": 0.5296, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.7266666666666666, | |
| "grad_norm": 0.22562238574028015, | |
| "learning_rate": 0.00014148148148148147, | |
| "loss": 0.4975, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 1.728888888888889, | |
| "grad_norm": 0.24731186032295227, | |
| "learning_rate": 0.00014123456790123454, | |
| "loss": 0.6267, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.7311111111111113, | |
| "grad_norm": 0.20758940279483795, | |
| "learning_rate": 0.00014098765432098763, | |
| "loss": 0.5358, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 1.7333333333333334, | |
| "grad_norm": 0.23659007251262665, | |
| "learning_rate": 0.00014074074074074073, | |
| "loss": 0.4894, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.7355555555555555, | |
| "grad_norm": 0.20344805717468262, | |
| "learning_rate": 0.00014049382716049382, | |
| "loss": 0.3921, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 1.7377777777777776, | |
| "grad_norm": 0.17698940634727478, | |
| "learning_rate": 0.0001402469135802469, | |
| "loss": 0.4805, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "grad_norm": 0.19029003381729126, | |
| "learning_rate": 0.00014, | |
| "loss": 0.4149, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 1.7422222222222223, | |
| "grad_norm": 0.2106352597475052, | |
| "learning_rate": 0.00013975308641975308, | |
| "loss": 0.6197, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.7444444444444445, | |
| "grad_norm": 0.18349190056324005, | |
| "learning_rate": 0.00013950617283950615, | |
| "loss": 0.5057, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.7466666666666666, | |
| "grad_norm": 0.1870691180229187, | |
| "learning_rate": 0.00013925925925925925, | |
| "loss": 0.4372, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.748888888888889, | |
| "grad_norm": 0.17990504205226898, | |
| "learning_rate": 0.00013901234567901231, | |
| "loss": 0.5753, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 1.751111111111111, | |
| "grad_norm": 0.17211493849754333, | |
| "learning_rate": 0.00013876543209876544, | |
| "loss": 0.5469, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.7533333333333334, | |
| "grad_norm": 0.25620609521865845, | |
| "learning_rate": 0.0001385185185185185, | |
| "loss": 0.5231, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.7555555555555555, | |
| "grad_norm": 0.22142796218395233, | |
| "learning_rate": 0.0001382716049382716, | |
| "loss": 0.6096, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.7577777777777777, | |
| "grad_norm": 0.2566666305065155, | |
| "learning_rate": 0.00013802469135802467, | |
| "loss": 0.4896, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 0.18952696025371552, | |
| "learning_rate": 0.00013777777777777776, | |
| "loss": 0.4553, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.7622222222222224, | |
| "grad_norm": 0.5168548822402954, | |
| "learning_rate": 0.00013753086419753086, | |
| "loss": 0.5118, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 1.7644444444444445, | |
| "grad_norm": 0.25155314803123474, | |
| "learning_rate": 0.00013728395061728393, | |
| "loss": 0.5459, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.7666666666666666, | |
| "grad_norm": 0.21072307229042053, | |
| "learning_rate": 0.00013703703703703702, | |
| "loss": 0.5674, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.7688888888888887, | |
| "grad_norm": 0.22808969020843506, | |
| "learning_rate": 0.00013679012345679012, | |
| "loss": 0.3832, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 1.771111111111111, | |
| "grad_norm": 0.19215697050094604, | |
| "learning_rate": 0.0001365432098765432, | |
| "loss": 0.5255, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 1.7733333333333334, | |
| "grad_norm": 0.26651448011398315, | |
| "learning_rate": 0.00013629629629629628, | |
| "loss": 0.5604, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.7755555555555556, | |
| "grad_norm": 0.17132802307605743, | |
| "learning_rate": 0.00013604938271604937, | |
| "loss": 0.4772, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.20619603991508484, | |
| "learning_rate": 0.00013580246913580244, | |
| "loss": 0.5187, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 0.18786154687404633, | |
| "learning_rate": 0.00013555555555555554, | |
| "loss": 0.4297, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 1.7822222222222224, | |
| "grad_norm": 0.1857113093137741, | |
| "learning_rate": 0.00013530864197530863, | |
| "loss": 0.4228, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 1.7844444444444445, | |
| "grad_norm": 0.3219771981239319, | |
| "learning_rate": 0.00013506172839506173, | |
| "loss": 0.42, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 1.7866666666666666, | |
| "grad_norm": 0.2078295797109604, | |
| "learning_rate": 0.0001348148148148148, | |
| "loss": 0.4053, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 1.7888888888888888, | |
| "grad_norm": 0.2320524901151657, | |
| "learning_rate": 0.0001345679012345679, | |
| "loss": 0.5345, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.791111111111111, | |
| "grad_norm": 0.2078818827867508, | |
| "learning_rate": 0.00013432098765432099, | |
| "loss": 0.4233, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 1.7933333333333334, | |
| "grad_norm": 0.19511982798576355, | |
| "learning_rate": 0.00013407407407407405, | |
| "loss": 0.4398, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 1.7955555555555556, | |
| "grad_norm": 0.19873318076133728, | |
| "learning_rate": 0.00013382716049382715, | |
| "loss": 0.5205, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 1.7977777777777777, | |
| "grad_norm": 0.16824252903461456, | |
| "learning_rate": 0.00013358024691358022, | |
| "loss": 0.4738, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 0.18699440360069275, | |
| "learning_rate": 0.0001333333333333333, | |
| "loss": 0.5153, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.8022222222222222, | |
| "grad_norm": 0.1715898960828781, | |
| "learning_rate": 0.0001330864197530864, | |
| "loss": 0.5829, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 1.8044444444444445, | |
| "grad_norm": 0.20615488290786743, | |
| "learning_rate": 0.0001328395061728395, | |
| "loss": 0.3645, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 1.8066666666666666, | |
| "grad_norm": 0.2258550077676773, | |
| "learning_rate": 0.00013259259259259257, | |
| "loss": 0.5682, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 1.8088888888888888, | |
| "grad_norm": 0.21636167168617249, | |
| "learning_rate": 0.00013234567901234567, | |
| "loss": 0.4189, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 1.8111111111111111, | |
| "grad_norm": 0.1962098926305771, | |
| "learning_rate": 0.00013209876543209876, | |
| "loss": 0.4858, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.8133333333333335, | |
| "grad_norm": 0.36040857434272766, | |
| "learning_rate": 0.00013185185185185183, | |
| "loss": 0.406, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 1.8155555555555556, | |
| "grad_norm": 0.2069418877363205, | |
| "learning_rate": 0.00013160493827160492, | |
| "loss": 0.5318, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 1.8177777777777777, | |
| "grad_norm": 0.16201527416706085, | |
| "learning_rate": 0.00013135802469135802, | |
| "loss": 0.5466, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 1.8199999999999998, | |
| "grad_norm": 0.21278874576091766, | |
| "learning_rate": 0.00013111111111111111, | |
| "loss": 0.5144, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 1.8222222222222222, | |
| "grad_norm": 0.16780561208724976, | |
| "learning_rate": 0.00013086419753086418, | |
| "loss": 0.427, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.8244444444444445, | |
| "grad_norm": 0.2174573689699173, | |
| "learning_rate": 0.00013061728395061728, | |
| "loss": 0.428, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 1.8266666666666667, | |
| "grad_norm": 0.1837967336177826, | |
| "learning_rate": 0.00013037037037037035, | |
| "loss": 0.5745, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 1.8288888888888888, | |
| "grad_norm": 0.2510828673839569, | |
| "learning_rate": 0.00013012345679012344, | |
| "loss": 0.3721, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 1.8311111111111111, | |
| "grad_norm": 0.19989460706710815, | |
| "learning_rate": 0.00012987654320987654, | |
| "loss": 0.4975, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 1.8333333333333335, | |
| "grad_norm": 0.19319596886634827, | |
| "learning_rate": 0.0001296296296296296, | |
| "loss": 0.43, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.8355555555555556, | |
| "grad_norm": 0.210468128323555, | |
| "learning_rate": 0.0001293827160493827, | |
| "loss": 0.4715, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 1.8377777777777777, | |
| "grad_norm": 0.1974964439868927, | |
| "learning_rate": 0.0001291358024691358, | |
| "loss": 0.5927, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 1.8399999999999999, | |
| "grad_norm": 0.1610316038131714, | |
| "learning_rate": 0.0001288888888888889, | |
| "loss": 0.4647, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 1.8422222222222222, | |
| "grad_norm": 0.19050978124141693, | |
| "learning_rate": 0.00012864197530864196, | |
| "loss": 0.5599, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 1.8444444444444446, | |
| "grad_norm": 0.25903093814849854, | |
| "learning_rate": 0.00012839506172839505, | |
| "loss": 0.46, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.8466666666666667, | |
| "grad_norm": 0.2022525817155838, | |
| "learning_rate": 0.00012814814814814812, | |
| "loss": 0.3934, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 1.8488888888888888, | |
| "grad_norm": 0.2194070965051651, | |
| "learning_rate": 0.00012790123456790122, | |
| "loss": 0.4065, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 1.8511111111111112, | |
| "grad_norm": 0.14326930046081543, | |
| "learning_rate": 0.0001276543209876543, | |
| "loss": 0.4014, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 1.8533333333333335, | |
| "grad_norm": 0.19791755080223083, | |
| "learning_rate": 0.0001274074074074074, | |
| "loss": 0.4356, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 1.8555555555555556, | |
| "grad_norm": 0.17042705416679382, | |
| "learning_rate": 0.00012716049382716047, | |
| "loss": 0.5047, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.8577777777777778, | |
| "grad_norm": 0.18079350888729095, | |
| "learning_rate": 0.00012691358024691357, | |
| "loss": 0.5037, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 1.8599999999999999, | |
| "grad_norm": 0.19158481061458588, | |
| "learning_rate": 0.00012666666666666666, | |
| "loss": 0.5508, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 1.8622222222222222, | |
| "grad_norm": 0.2053975760936737, | |
| "learning_rate": 0.00012641975308641973, | |
| "loss": 0.4761, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 1.8644444444444446, | |
| "grad_norm": 0.2540335953235626, | |
| "learning_rate": 0.00012617283950617283, | |
| "loss": 0.4546, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 1.8666666666666667, | |
| "grad_norm": 0.23568333685398102, | |
| "learning_rate": 0.0001259259259259259, | |
| "loss": 0.4206, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.8688888888888888, | |
| "grad_norm": 0.2795317769050598, | |
| "learning_rate": 0.00012567901234567902, | |
| "loss": 0.6551, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 1.871111111111111, | |
| "grad_norm": 0.2189972698688507, | |
| "learning_rate": 0.00012543209876543209, | |
| "loss": 0.686, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 1.8733333333333333, | |
| "grad_norm": 0.23649589717388153, | |
| "learning_rate": 0.00012518518518518518, | |
| "loss": 0.5333, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 1.8755555555555556, | |
| "grad_norm": 0.19060775637626648, | |
| "learning_rate": 0.00012493827160493825, | |
| "loss": 0.4808, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 1.8777777777777778, | |
| "grad_norm": 0.315738707780838, | |
| "learning_rate": 0.00012469135802469134, | |
| "loss": 0.7281, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.20539192855358124, | |
| "learning_rate": 0.00012444444444444444, | |
| "loss": 0.4933, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 1.8822222222222222, | |
| "grad_norm": 0.18650329113006592, | |
| "learning_rate": 0.0001241975308641975, | |
| "loss": 0.4899, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 1.8844444444444446, | |
| "grad_norm": 0.18989071249961853, | |
| "learning_rate": 0.0001239506172839506, | |
| "loss": 0.5128, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 1.8866666666666667, | |
| "grad_norm": 0.23419564962387085, | |
| "learning_rate": 0.0001237037037037037, | |
| "loss": 0.5095, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 1.8888888888888888, | |
| "grad_norm": 0.19001886248588562, | |
| "learning_rate": 0.0001234567901234568, | |
| "loss": 0.5382, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.891111111111111, | |
| "grad_norm": 0.2222389578819275, | |
| "learning_rate": 0.00012320987654320986, | |
| "loss": 0.4132, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 1.8933333333333333, | |
| "grad_norm": 0.21193592250347137, | |
| "learning_rate": 0.00012296296296296296, | |
| "loss": 0.5554, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 1.8955555555555557, | |
| "grad_norm": 0.26315727829933167, | |
| "learning_rate": 0.00012271604938271602, | |
| "loss": 0.5161, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 1.8977777777777778, | |
| "grad_norm": 0.18496252596378326, | |
| "learning_rate": 0.00012246913580246912, | |
| "loss": 0.4143, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "grad_norm": 0.1588750034570694, | |
| "learning_rate": 0.00012222222222222221, | |
| "loss": 0.5138, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.9022222222222223, | |
| "grad_norm": 0.19370391964912415, | |
| "learning_rate": 0.00012197530864197531, | |
| "loss": 0.5304, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 1.9044444444444446, | |
| "grad_norm": 0.23615752160549164, | |
| "learning_rate": 0.00012172839506172839, | |
| "loss": 0.4368, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 1.9066666666666667, | |
| "grad_norm": 0.17469893395900726, | |
| "learning_rate": 0.00012148148148148147, | |
| "loss": 0.4681, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 1.9088888888888889, | |
| "grad_norm": 0.1781235933303833, | |
| "learning_rate": 0.00012123456790123455, | |
| "loss": 0.4829, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 1.911111111111111, | |
| "grad_norm": 0.18554018437862396, | |
| "learning_rate": 0.00012098765432098764, | |
| "loss": 0.4468, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.9133333333333333, | |
| "grad_norm": 0.22353385388851166, | |
| "learning_rate": 0.00012074074074074073, | |
| "loss": 0.5095, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 1.9155555555555557, | |
| "grad_norm": 0.2333441525697708, | |
| "learning_rate": 0.00012049382716049381, | |
| "loss": 0.4968, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 1.9177777777777778, | |
| "grad_norm": 0.20324921607971191, | |
| "learning_rate": 0.0001202469135802469, | |
| "loss": 0.5194, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 0.16060517728328705, | |
| "learning_rate": 0.00011999999999999999, | |
| "loss": 0.4701, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 1.9222222222222223, | |
| "grad_norm": 0.22599969804286957, | |
| "learning_rate": 0.00011975308641975308, | |
| "loss": 0.4869, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.9244444444444444, | |
| "grad_norm": 0.1759766787290573, | |
| "learning_rate": 0.00011950617283950617, | |
| "loss": 0.5143, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 1.9266666666666667, | |
| "grad_norm": 0.17240673303604126, | |
| "learning_rate": 0.00011925925925925925, | |
| "loss": 0.4249, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 1.9288888888888889, | |
| "grad_norm": 0.18993405997753143, | |
| "learning_rate": 0.00011901234567901233, | |
| "loss": 0.4549, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 1.931111111111111, | |
| "grad_norm": 0.23247094452381134, | |
| "learning_rate": 0.00011876543209876541, | |
| "loss": 0.4951, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 1.9333333333333333, | |
| "grad_norm": 0.1876572072505951, | |
| "learning_rate": 0.0001185185185185185, | |
| "loss": 0.4515, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.9355555555555557, | |
| "grad_norm": 0.18355309963226318, | |
| "learning_rate": 0.0001182716049382716, | |
| "loss": 0.4013, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 1.9377777777777778, | |
| "grad_norm": 0.14535872638225555, | |
| "learning_rate": 0.00011802469135802468, | |
| "loss": 0.5282, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 0.20540978014469147, | |
| "learning_rate": 0.00011777777777777776, | |
| "loss": 0.4743, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 1.942222222222222, | |
| "grad_norm": 0.21471437811851501, | |
| "learning_rate": 0.00011753086419753086, | |
| "loss": 0.5149, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 1.9444444444444444, | |
| "grad_norm": 0.17426244914531708, | |
| "learning_rate": 0.00011728395061728394, | |
| "loss": 0.4874, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.9466666666666668, | |
| "grad_norm": 0.18514637649059296, | |
| "learning_rate": 0.00011703703703703702, | |
| "loss": 0.4725, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 1.948888888888889, | |
| "grad_norm": 0.18219704926013947, | |
| "learning_rate": 0.0001167901234567901, | |
| "loss": 0.4573, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 1.951111111111111, | |
| "grad_norm": 0.1740552932024002, | |
| "learning_rate": 0.00011654320987654319, | |
| "loss": 0.4537, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 1.9533333333333334, | |
| "grad_norm": 0.20307578146457672, | |
| "learning_rate": 0.0001162962962962963, | |
| "loss": 0.5837, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 1.9555555555555557, | |
| "grad_norm": 0.21196050941944122, | |
| "learning_rate": 0.00011604938271604938, | |
| "loss": 0.4758, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.9577777777777778, | |
| "grad_norm": 0.18597066402435303, | |
| "learning_rate": 0.00011580246913580246, | |
| "loss": 0.5342, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 0.19975154101848602, | |
| "learning_rate": 0.00011555555555555554, | |
| "loss": 0.5093, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 1.962222222222222, | |
| "grad_norm": 0.16883786022663116, | |
| "learning_rate": 0.00011530864197530863, | |
| "loss": 0.5046, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 1.9644444444444444, | |
| "grad_norm": 0.19684910774230957, | |
| "learning_rate": 0.00011506172839506172, | |
| "loss": 0.5143, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 1.9666666666666668, | |
| "grad_norm": 0.17863331735134125, | |
| "learning_rate": 0.0001148148148148148, | |
| "loss": 0.4796, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.968888888888889, | |
| "grad_norm": 0.252346396446228, | |
| "learning_rate": 0.00011456790123456789, | |
| "loss": 0.4926, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 1.971111111111111, | |
| "grad_norm": 0.17805711925029755, | |
| "learning_rate": 0.00011432098765432099, | |
| "loss": 0.5864, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 1.9733333333333334, | |
| "grad_norm": 0.2277565598487854, | |
| "learning_rate": 0.00011407407407407407, | |
| "loss": 0.582, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 1.9755555555555555, | |
| "grad_norm": 0.23421330749988556, | |
| "learning_rate": 0.00011382716049382715, | |
| "loss": 0.6081, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 1.9777777777777779, | |
| "grad_norm": 0.23326914012432098, | |
| "learning_rate": 0.00011358024691358023, | |
| "loss": 0.5049, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "grad_norm": 0.17574244737625122, | |
| "learning_rate": 0.00011333333333333331, | |
| "loss": 0.5323, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 1.982222222222222, | |
| "grad_norm": 0.22068263590335846, | |
| "learning_rate": 0.00011308641975308641, | |
| "loss": 0.4082, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 1.9844444444444445, | |
| "grad_norm": 0.19230858981609344, | |
| "learning_rate": 0.00011283950617283949, | |
| "loss": 0.5426, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 1.9866666666666668, | |
| "grad_norm": 0.2325981706380844, | |
| "learning_rate": 0.00011259259259259259, | |
| "loss": 0.6252, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 1.988888888888889, | |
| "grad_norm": 0.22357137501239777, | |
| "learning_rate": 0.00011234567901234567, | |
| "loss": 0.4211, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.991111111111111, | |
| "grad_norm": 0.2307082563638687, | |
| "learning_rate": 0.00011209876543209876, | |
| "loss": 0.5304, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 1.9933333333333332, | |
| "grad_norm": 0.2556259334087372, | |
| "learning_rate": 0.00011185185185185184, | |
| "loss": 0.5676, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 1.9955555555555555, | |
| "grad_norm": 0.18439683318138123, | |
| "learning_rate": 0.00011160493827160493, | |
| "loss": 0.4426, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 1.9977777777777779, | |
| "grad_norm": 0.2009860873222351, | |
| "learning_rate": 0.00011135802469135801, | |
| "loss": 0.5774, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.14363819360733032, | |
| "learning_rate": 0.00011111111111111109, | |
| "loss": 0.4286, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.002222222222222, | |
| "grad_norm": 0.161085844039917, | |
| "learning_rate": 0.0001108641975308642, | |
| "loss": 0.4121, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 2.0044444444444443, | |
| "grad_norm": 0.15874294936656952, | |
| "learning_rate": 0.00011061728395061728, | |
| "loss": 0.4661, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.006666666666667, | |
| "grad_norm": 0.17735889554023743, | |
| "learning_rate": 0.00011037037037037036, | |
| "loss": 0.5782, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 2.008888888888889, | |
| "grad_norm": 0.17721158266067505, | |
| "learning_rate": 0.00011012345679012344, | |
| "loss": 0.4644, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.011111111111111, | |
| "grad_norm": 0.16406133770942688, | |
| "learning_rate": 0.00010987654320987654, | |
| "loss": 0.422, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.013333333333333, | |
| "grad_norm": 0.2074761539697647, | |
| "learning_rate": 0.00010962962962962962, | |
| "loss": 0.4846, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.0155555555555558, | |
| "grad_norm": 0.22118505835533142, | |
| "learning_rate": 0.0001093827160493827, | |
| "loss": 0.5174, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 2.017777777777778, | |
| "grad_norm": 0.1525031179189682, | |
| "learning_rate": 0.00010913580246913578, | |
| "loss": 0.519, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 0.17472517490386963, | |
| "learning_rate": 0.00010888888888888889, | |
| "loss": 0.416, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 2.022222222222222, | |
| "grad_norm": 0.18598146736621857, | |
| "learning_rate": 0.00010864197530864197, | |
| "loss": 0.6634, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.0244444444444443, | |
| "grad_norm": 0.15725669264793396, | |
| "learning_rate": 0.00010839506172839505, | |
| "loss": 0.4202, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 2.026666666666667, | |
| "grad_norm": 0.17347405850887299, | |
| "learning_rate": 0.00010814814814814814, | |
| "loss": 0.4054, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.028888888888889, | |
| "grad_norm": 0.1937365084886551, | |
| "learning_rate": 0.00010790123456790122, | |
| "loss": 0.5662, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 2.031111111111111, | |
| "grad_norm": 0.17436783015727997, | |
| "learning_rate": 0.00010765432098765431, | |
| "loss": 0.4425, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.033333333333333, | |
| "grad_norm": 0.2667844593524933, | |
| "learning_rate": 0.0001074074074074074, | |
| "loss": 0.5084, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.0355555555555553, | |
| "grad_norm": 0.17061366140842438, | |
| "learning_rate": 0.00010716049382716048, | |
| "loss": 0.5198, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.037777777777778, | |
| "grad_norm": 0.19045591354370117, | |
| "learning_rate": 0.00010691358024691357, | |
| "loss": 0.455, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 0.15703974664211273, | |
| "learning_rate": 0.00010666666666666667, | |
| "loss": 0.4895, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.042222222222222, | |
| "grad_norm": 0.304848849773407, | |
| "learning_rate": 0.00010641975308641975, | |
| "loss": 0.5071, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 2.0444444444444443, | |
| "grad_norm": 0.1720246821641922, | |
| "learning_rate": 0.00010617283950617283, | |
| "loss": 0.4152, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.046666666666667, | |
| "grad_norm": 0.1733669489622116, | |
| "learning_rate": 0.00010592592592592591, | |
| "loss": 0.4133, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 2.048888888888889, | |
| "grad_norm": 0.18830733001232147, | |
| "learning_rate": 0.00010567901234567899, | |
| "loss": 0.3884, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.051111111111111, | |
| "grad_norm": 0.29931190609931946, | |
| "learning_rate": 0.00010543209876543209, | |
| "loss": 0.6018, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 2.0533333333333332, | |
| "grad_norm": 0.18850049376487732, | |
| "learning_rate": 0.00010518518518518518, | |
| "loss": 0.4683, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.0555555555555554, | |
| "grad_norm": 0.1913931965827942, | |
| "learning_rate": 0.00010493827160493826, | |
| "loss": 0.4012, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.057777777777778, | |
| "grad_norm": 0.21711386740207672, | |
| "learning_rate": 0.00010469135802469135, | |
| "loss": 0.4996, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "grad_norm": 0.2386695295572281, | |
| "learning_rate": 0.00010444444444444444, | |
| "loss": 0.4909, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 2.062222222222222, | |
| "grad_norm": 0.17317090928554535, | |
| "learning_rate": 0.00010419753086419752, | |
| "loss": 0.4022, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.0644444444444443, | |
| "grad_norm": 0.1801401972770691, | |
| "learning_rate": 0.0001039506172839506, | |
| "loss": 0.5695, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 2.066666666666667, | |
| "grad_norm": 0.25198373198509216, | |
| "learning_rate": 0.00010370370370370369, | |
| "loss": 0.5501, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.068888888888889, | |
| "grad_norm": 0.3184848427772522, | |
| "learning_rate": 0.00010345679012345677, | |
| "loss": 0.5924, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 2.071111111111111, | |
| "grad_norm": 0.20114697515964508, | |
| "learning_rate": 0.00010320987654320988, | |
| "loss": 0.5352, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.0733333333333333, | |
| "grad_norm": 0.18714340031147003, | |
| "learning_rate": 0.00010296296296296296, | |
| "loss": 0.4235, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 2.0755555555555554, | |
| "grad_norm": 0.23173195123672485, | |
| "learning_rate": 0.00010271604938271604, | |
| "loss": 0.3959, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.077777777777778, | |
| "grad_norm": 0.21912041306495667, | |
| "learning_rate": 0.00010246913580246912, | |
| "loss": 0.56, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 0.17358708381652832, | |
| "learning_rate": 0.00010222222222222222, | |
| "loss": 0.403, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.082222222222222, | |
| "grad_norm": 0.18150022625923157, | |
| "learning_rate": 0.0001019753086419753, | |
| "loss": 0.4763, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 2.0844444444444443, | |
| "grad_norm": 0.24984142184257507, | |
| "learning_rate": 0.00010172839506172838, | |
| "loss": 0.5316, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.086666666666667, | |
| "grad_norm": 0.18395419418811798, | |
| "learning_rate": 0.00010148148148148147, | |
| "loss": 0.4465, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 2.088888888888889, | |
| "grad_norm": 0.2704440951347351, | |
| "learning_rate": 0.00010123456790123457, | |
| "loss": 0.4491, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.091111111111111, | |
| "grad_norm": 0.2826973497867584, | |
| "learning_rate": 0.00010098765432098765, | |
| "loss": 0.3804, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 2.0933333333333333, | |
| "grad_norm": 0.27621716260910034, | |
| "learning_rate": 0.00010074074074074073, | |
| "loss": 0.3984, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.0955555555555554, | |
| "grad_norm": 0.2091452181339264, | |
| "learning_rate": 0.00010049382716049381, | |
| "loss": 0.3938, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 2.097777777777778, | |
| "grad_norm": 0.2692224979400635, | |
| "learning_rate": 0.0001002469135802469, | |
| "loss": 0.5258, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 0.1720598340034485, | |
| "learning_rate": 9.999999999999999e-05, | |
| "loss": 0.421, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.102222222222222, | |
| "grad_norm": 0.23547446727752686, | |
| "learning_rate": 9.975308641975307e-05, | |
| "loss": 0.5601, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.1044444444444443, | |
| "grad_norm": 0.20717960596084595, | |
| "learning_rate": 9.950617283950617e-05, | |
| "loss": 0.4045, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 2.1066666666666665, | |
| "grad_norm": 0.21972191333770752, | |
| "learning_rate": 9.925925925925925e-05, | |
| "loss": 0.6487, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.108888888888889, | |
| "grad_norm": 0.20680764317512512, | |
| "learning_rate": 9.901234567901234e-05, | |
| "loss": 0.4511, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 2.111111111111111, | |
| "grad_norm": 0.16888514161109924, | |
| "learning_rate": 9.876543209876543e-05, | |
| "loss": 0.3545, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.1133333333333333, | |
| "grad_norm": 0.23663461208343506, | |
| "learning_rate": 9.851851851851851e-05, | |
| "loss": 0.3841, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 2.1155555555555554, | |
| "grad_norm": 0.22071480751037598, | |
| "learning_rate": 9.827160493827159e-05, | |
| "loss": 0.4487, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.117777777777778, | |
| "grad_norm": 0.21643605828285217, | |
| "learning_rate": 9.802469135802467e-05, | |
| "loss": 0.616, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 0.25067830085754395, | |
| "learning_rate": 9.777777777777778e-05, | |
| "loss": 0.4673, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.1222222222222222, | |
| "grad_norm": 0.3261134624481201, | |
| "learning_rate": 9.753086419753086e-05, | |
| "loss": 0.5496, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.1244444444444444, | |
| "grad_norm": 0.30298173427581787, | |
| "learning_rate": 9.728395061728394e-05, | |
| "loss": 0.5964, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.1266666666666665, | |
| "grad_norm": 0.2082533985376358, | |
| "learning_rate": 9.703703703703702e-05, | |
| "loss": 0.4617, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 2.128888888888889, | |
| "grad_norm": 0.2543969750404358, | |
| "learning_rate": 9.679012345679012e-05, | |
| "loss": 0.5521, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.131111111111111, | |
| "grad_norm": 0.227843776345253, | |
| "learning_rate": 9.65432098765432e-05, | |
| "loss": 0.3569, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 2.1333333333333333, | |
| "grad_norm": 0.29863792657852173, | |
| "learning_rate": 9.629629629629628e-05, | |
| "loss": 0.5587, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.1355555555555554, | |
| "grad_norm": 0.22359107434749603, | |
| "learning_rate": 9.604938271604936e-05, | |
| "loss": 0.3963, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 2.137777777777778, | |
| "grad_norm": 0.20687472820281982, | |
| "learning_rate": 9.580246913580247e-05, | |
| "loss": 0.4206, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 0.21215292811393738, | |
| "learning_rate": 9.555555555555555e-05, | |
| "loss": 0.5411, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 2.1422222222222222, | |
| "grad_norm": 0.23551730811595917, | |
| "learning_rate": 9.530864197530864e-05, | |
| "loss": 0.4963, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.1444444444444444, | |
| "grad_norm": 0.24306513369083405, | |
| "learning_rate": 9.506172839506172e-05, | |
| "loss": 0.5604, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.1466666666666665, | |
| "grad_norm": 0.18296387791633606, | |
| "learning_rate": 9.48148148148148e-05, | |
| "loss": 0.3915, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.148888888888889, | |
| "grad_norm": 0.17923778295516968, | |
| "learning_rate": 9.45679012345679e-05, | |
| "loss": 0.4397, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 2.151111111111111, | |
| "grad_norm": 0.27194756269454956, | |
| "learning_rate": 9.432098765432098e-05, | |
| "loss": 0.464, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.1533333333333333, | |
| "grad_norm": 0.2555384039878845, | |
| "learning_rate": 9.407407407407407e-05, | |
| "loss": 0.487, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 2.1555555555555554, | |
| "grad_norm": 0.19348587095737457, | |
| "learning_rate": 9.382716049382715e-05, | |
| "loss": 0.447, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.1577777777777776, | |
| "grad_norm": 0.34053802490234375, | |
| "learning_rate": 9.358024691358025e-05, | |
| "loss": 0.4894, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 0.20440658926963806, | |
| "learning_rate": 9.333333333333333e-05, | |
| "loss": 0.4568, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.1622222222222223, | |
| "grad_norm": 0.2978667616844177, | |
| "learning_rate": 9.308641975308641e-05, | |
| "loss": 0.5573, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 2.1644444444444444, | |
| "grad_norm": 0.21186195313930511, | |
| "learning_rate": 9.283950617283949e-05, | |
| "loss": 0.4004, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 2.1666666666666665, | |
| "grad_norm": 0.22105668485164642, | |
| "learning_rate": 9.259259259259257e-05, | |
| "loss": 0.4684, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.168888888888889, | |
| "grad_norm": 0.2607153058052063, | |
| "learning_rate": 9.234567901234567e-05, | |
| "loss": 0.4474, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 2.171111111111111, | |
| "grad_norm": 0.17750707268714905, | |
| "learning_rate": 9.209876543209876e-05, | |
| "loss": 0.4148, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 2.1733333333333333, | |
| "grad_norm": 0.22568461298942566, | |
| "learning_rate": 9.185185185185185e-05, | |
| "loss": 0.4474, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 2.1755555555555555, | |
| "grad_norm": 0.20598728954792023, | |
| "learning_rate": 9.160493827160493e-05, | |
| "loss": 0.3738, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 2.1777777777777776, | |
| "grad_norm": 0.27934280037879944, | |
| "learning_rate": 9.135802469135802e-05, | |
| "loss": 0.4491, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 0.20362699031829834, | |
| "learning_rate": 9.11111111111111e-05, | |
| "loss": 0.4158, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 2.1822222222222223, | |
| "grad_norm": 0.3080417215824127, | |
| "learning_rate": 9.086419753086419e-05, | |
| "loss": 0.5944, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 2.1844444444444444, | |
| "grad_norm": 0.24677591025829315, | |
| "learning_rate": 9.061728395061727e-05, | |
| "loss": 0.4922, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 2.1866666666666665, | |
| "grad_norm": 0.2841777205467224, | |
| "learning_rate": 9.037037037037038e-05, | |
| "loss": 0.4315, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 2.188888888888889, | |
| "grad_norm": 0.3089623749256134, | |
| "learning_rate": 9.012345679012346e-05, | |
| "loss": 0.5598, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.1911111111111112, | |
| "grad_norm": 0.24291516840457916, | |
| "learning_rate": 8.987654320987654e-05, | |
| "loss": 0.3336, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 2.1933333333333334, | |
| "grad_norm": 0.20366284251213074, | |
| "learning_rate": 8.962962962962962e-05, | |
| "loss": 0.4383, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 2.1955555555555555, | |
| "grad_norm": 0.23388588428497314, | |
| "learning_rate": 8.93827160493827e-05, | |
| "loss": 0.3058, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 2.1977777777777776, | |
| "grad_norm": 0.2930527329444885, | |
| "learning_rate": 8.91358024691358e-05, | |
| "loss": 0.4341, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 0.2246478796005249, | |
| "learning_rate": 8.888888888888888e-05, | |
| "loss": 0.4823, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.2022222222222223, | |
| "grad_norm": 0.3022997975349426, | |
| "learning_rate": 8.864197530864196e-05, | |
| "loss": 0.5476, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 2.2044444444444444, | |
| "grad_norm": 0.2257053703069687, | |
| "learning_rate": 8.839506172839506e-05, | |
| "loss": 0.3951, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 2.2066666666666666, | |
| "grad_norm": 0.5187708139419556, | |
| "learning_rate": 8.814814814814815e-05, | |
| "loss": 0.5601, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 2.2088888888888887, | |
| "grad_norm": 0.26322057843208313, | |
| "learning_rate": 8.790123456790123e-05, | |
| "loss": 0.4814, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 2.2111111111111112, | |
| "grad_norm": 0.26765871047973633, | |
| "learning_rate": 8.765432098765431e-05, | |
| "loss": 0.4855, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.2133333333333334, | |
| "grad_norm": 0.2329784631729126, | |
| "learning_rate": 8.74074074074074e-05, | |
| "loss": 0.3676, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 2.2155555555555555, | |
| "grad_norm": 0.22700552642345428, | |
| "learning_rate": 8.716049382716048e-05, | |
| "loss": 0.4376, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 2.2177777777777776, | |
| "grad_norm": 0.25210490822792053, | |
| "learning_rate": 8.691358024691357e-05, | |
| "loss": 0.4916, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.23448699712753296, | |
| "learning_rate": 8.666666666666665e-05, | |
| "loss": 0.4765, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 0.23905183374881744, | |
| "learning_rate": 8.641975308641975e-05, | |
| "loss": 0.4787, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.2244444444444444, | |
| "grad_norm": 0.24907182157039642, | |
| "learning_rate": 8.617283950617283e-05, | |
| "loss": 0.3769, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 2.2266666666666666, | |
| "grad_norm": 0.1960950791835785, | |
| "learning_rate": 8.592592592592593e-05, | |
| "loss": 0.4282, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 2.2288888888888887, | |
| "grad_norm": 0.2484542727470398, | |
| "learning_rate": 8.567901234567901e-05, | |
| "loss": 0.4188, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 2.2311111111111113, | |
| "grad_norm": 0.22238413989543915, | |
| "learning_rate": 8.543209876543209e-05, | |
| "loss": 0.5108, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 2.2333333333333334, | |
| "grad_norm": 0.20457886159420013, | |
| "learning_rate": 8.518518518518517e-05, | |
| "loss": 0.4057, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.2355555555555555, | |
| "grad_norm": 0.2082635909318924, | |
| "learning_rate": 8.493827160493825e-05, | |
| "loss": 0.4503, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 2.2377777777777776, | |
| "grad_norm": 0.2857392430305481, | |
| "learning_rate": 8.469135802469136e-05, | |
| "loss": 0.5822, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 0.22663135826587677, | |
| "learning_rate": 8.444444444444444e-05, | |
| "loss": 0.3647, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 2.2422222222222223, | |
| "grad_norm": 0.22252953052520752, | |
| "learning_rate": 8.419753086419752e-05, | |
| "loss": 0.4524, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 2.2444444444444445, | |
| "grad_norm": 0.22072190046310425, | |
| "learning_rate": 8.39506172839506e-05, | |
| "loss": 0.4237, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.2466666666666666, | |
| "grad_norm": 0.23063765466213226, | |
| "learning_rate": 8.37037037037037e-05, | |
| "loss": 0.3466, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 2.2488888888888887, | |
| "grad_norm": 0.23797385394573212, | |
| "learning_rate": 8.345679012345678e-05, | |
| "loss": 0.3931, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 2.2511111111111113, | |
| "grad_norm": 0.2844562530517578, | |
| "learning_rate": 8.320987654320986e-05, | |
| "loss": 0.4543, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 2.2533333333333334, | |
| "grad_norm": 0.3510430157184601, | |
| "learning_rate": 8.296296296296295e-05, | |
| "loss": 0.4259, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 2.2555555555555555, | |
| "grad_norm": 0.21694698929786682, | |
| "learning_rate": 8.271604938271605e-05, | |
| "loss": 0.3798, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 2.2577777777777777, | |
| "grad_norm": 0.24299605190753937, | |
| "learning_rate": 8.246913580246914e-05, | |
| "loss": 0.6, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "grad_norm": 0.2540102005004883, | |
| "learning_rate": 8.222222222222222e-05, | |
| "loss": 0.4493, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 2.2622222222222224, | |
| "grad_norm": 0.27778634428977966, | |
| "learning_rate": 8.19753086419753e-05, | |
| "loss": 0.5292, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 2.2644444444444445, | |
| "grad_norm": 0.24594548344612122, | |
| "learning_rate": 8.172839506172838e-05, | |
| "loss": 0.4239, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 2.2666666666666666, | |
| "grad_norm": 0.27180740237236023, | |
| "learning_rate": 8.148148148148148e-05, | |
| "loss": 0.5964, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.2688888888888887, | |
| "grad_norm": 0.2825404703617096, | |
| "learning_rate": 8.123456790123456e-05, | |
| "loss": 0.5642, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 2.2711111111111113, | |
| "grad_norm": 0.24407710134983063, | |
| "learning_rate": 8.098765432098765e-05, | |
| "loss": 0.5249, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 2.2733333333333334, | |
| "grad_norm": 0.25856301188468933, | |
| "learning_rate": 8.074074074074073e-05, | |
| "loss": 0.5028, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 2.2755555555555556, | |
| "grad_norm": 0.2732691168785095, | |
| "learning_rate": 8.049382716049383e-05, | |
| "loss": 0.4674, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 2.2777777777777777, | |
| "grad_norm": 0.2448354810476303, | |
| "learning_rate": 8.024691358024691e-05, | |
| "loss": 0.3741, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.2800000000000002, | |
| "grad_norm": 0.2410421073436737, | |
| "learning_rate": 7.999999999999999e-05, | |
| "loss": 0.3923, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 2.2822222222222224, | |
| "grad_norm": 0.23940524458885193, | |
| "learning_rate": 7.975308641975307e-05, | |
| "loss": 0.4602, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 2.2844444444444445, | |
| "grad_norm": 0.2394387274980545, | |
| "learning_rate": 7.950617283950616e-05, | |
| "loss": 0.5353, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 2.2866666666666666, | |
| "grad_norm": 0.35966625809669495, | |
| "learning_rate": 7.925925925925925e-05, | |
| "loss": 0.5002, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 2.2888888888888888, | |
| "grad_norm": 0.23209857940673828, | |
| "learning_rate": 7.901234567901235e-05, | |
| "loss": 0.4076, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.2911111111111113, | |
| "grad_norm": 0.28689658641815186, | |
| "learning_rate": 7.876543209876543e-05, | |
| "loss": 0.5033, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 2.2933333333333334, | |
| "grad_norm": 0.2399093359708786, | |
| "learning_rate": 7.851851851851851e-05, | |
| "loss": 0.5123, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 2.2955555555555556, | |
| "grad_norm": 0.23205643892288208, | |
| "learning_rate": 7.82716049382716e-05, | |
| "loss": 0.4899, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 2.2977777777777777, | |
| "grad_norm": 0.25627297163009644, | |
| "learning_rate": 7.802469135802469e-05, | |
| "loss": 0.4907, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "grad_norm": 0.19083629548549652, | |
| "learning_rate": 7.777777777777777e-05, | |
| "loss": 0.4588, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 2.3022222222222224, | |
| "grad_norm": 0.270696759223938, | |
| "learning_rate": 7.753086419753085e-05, | |
| "loss": 0.4741, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 2.3044444444444445, | |
| "grad_norm": 0.2945954203605652, | |
| "learning_rate": 7.728395061728396e-05, | |
| "loss": 0.5246, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 2.3066666666666666, | |
| "grad_norm": 0.2649891972541809, | |
| "learning_rate": 7.703703703703704e-05, | |
| "loss": 0.3741, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 2.3088888888888888, | |
| "grad_norm": 0.208853617310524, | |
| "learning_rate": 7.679012345679012e-05, | |
| "loss": 0.4936, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 2.311111111111111, | |
| "grad_norm": 0.23785865306854248, | |
| "learning_rate": 7.65432098765432e-05, | |
| "loss": 0.518, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.3133333333333335, | |
| "grad_norm": 0.24360238015651703, | |
| "learning_rate": 7.629629629629628e-05, | |
| "loss": 0.5891, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 2.3155555555555556, | |
| "grad_norm": 0.2558804452419281, | |
| "learning_rate": 7.604938271604937e-05, | |
| "loss": 0.4153, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 2.3177777777777777, | |
| "grad_norm": 0.20021018385887146, | |
| "learning_rate": 7.580246913580246e-05, | |
| "loss": 0.4733, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 0.2404988557100296, | |
| "learning_rate": 7.555555555555554e-05, | |
| "loss": 0.4339, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 2.3222222222222224, | |
| "grad_norm": 0.31835708022117615, | |
| "learning_rate": 7.530864197530864e-05, | |
| "loss": 0.4806, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 2.3244444444444445, | |
| "grad_norm": 0.24206705391407013, | |
| "learning_rate": 7.506172839506173e-05, | |
| "loss": 0.4553, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 2.3266666666666667, | |
| "grad_norm": 0.2239079475402832, | |
| "learning_rate": 7.481481481481481e-05, | |
| "loss": 0.338, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 2.328888888888889, | |
| "grad_norm": 0.23047927021980286, | |
| "learning_rate": 7.45679012345679e-05, | |
| "loss": 0.4626, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 2.3311111111111114, | |
| "grad_norm": 0.22697985172271729, | |
| "learning_rate": 7.432098765432098e-05, | |
| "loss": 0.4999, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 2.3333333333333335, | |
| "grad_norm": 0.21413327753543854, | |
| "learning_rate": 7.407407407407406e-05, | |
| "loss": 0.4192, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.3355555555555556, | |
| "grad_norm": 0.3700278401374817, | |
| "learning_rate": 7.382716049382715e-05, | |
| "loss": 0.5033, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 2.3377777777777777, | |
| "grad_norm": 0.24023550748825073, | |
| "learning_rate": 7.358024691358024e-05, | |
| "loss": 0.4166, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "grad_norm": 0.2508622109889984, | |
| "learning_rate": 7.333333333333332e-05, | |
| "loss": 0.3588, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 2.3422222222222224, | |
| "grad_norm": 0.23471470177173615, | |
| "learning_rate": 7.308641975308641e-05, | |
| "loss": 0.4104, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 2.3444444444444446, | |
| "grad_norm": 0.22947955131530762, | |
| "learning_rate": 7.28395061728395e-05, | |
| "loss": 0.4542, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 2.3466666666666667, | |
| "grad_norm": 0.24685367941856384, | |
| "learning_rate": 7.259259259259259e-05, | |
| "loss": 0.3989, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 2.348888888888889, | |
| "grad_norm": 0.2971036434173584, | |
| "learning_rate": 7.234567901234567e-05, | |
| "loss": 0.3552, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 2.351111111111111, | |
| "grad_norm": 0.22053156793117523, | |
| "learning_rate": 7.209876543209877e-05, | |
| "loss": 0.3645, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 2.3533333333333335, | |
| "grad_norm": 0.21114762127399445, | |
| "learning_rate": 7.185185185185185e-05, | |
| "loss": 0.475, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 2.3555555555555556, | |
| "grad_norm": 0.2340921312570572, | |
| "learning_rate": 7.160493827160493e-05, | |
| "loss": 0.4075, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.3577777777777778, | |
| "grad_norm": 0.20541034638881683, | |
| "learning_rate": 7.135802469135801e-05, | |
| "loss": 0.4127, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 0.2008386254310608, | |
| "learning_rate": 7.11111111111111e-05, | |
| "loss": 0.4776, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 2.362222222222222, | |
| "grad_norm": 0.23569215834140778, | |
| "learning_rate": 7.086419753086419e-05, | |
| "loss": 0.4918, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 2.3644444444444446, | |
| "grad_norm": 0.2960083484649658, | |
| "learning_rate": 7.061728395061727e-05, | |
| "loss": 0.4651, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 2.3666666666666667, | |
| "grad_norm": 0.23241014778614044, | |
| "learning_rate": 7.037037037037036e-05, | |
| "loss": 0.434, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 2.368888888888889, | |
| "grad_norm": 0.23682597279548645, | |
| "learning_rate": 7.012345679012345e-05, | |
| "loss": 0.5068, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 2.371111111111111, | |
| "grad_norm": 0.25459370017051697, | |
| "learning_rate": 6.987654320987654e-05, | |
| "loss": 0.4596, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 2.3733333333333335, | |
| "grad_norm": 0.27613893151283264, | |
| "learning_rate": 6.962962962962962e-05, | |
| "loss": 0.5186, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 2.3755555555555556, | |
| "grad_norm": 0.2263878583908081, | |
| "learning_rate": 6.938271604938272e-05, | |
| "loss": 0.3455, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 2.3777777777777778, | |
| "grad_norm": 0.2196212112903595, | |
| "learning_rate": 6.91358024691358e-05, | |
| "loss": 0.3736, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "grad_norm": 0.2080947309732437, | |
| "learning_rate": 6.888888888888888e-05, | |
| "loss": 0.4641, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 2.3822222222222225, | |
| "grad_norm": 0.30166974663734436, | |
| "learning_rate": 6.864197530864196e-05, | |
| "loss": 0.4638, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 2.3844444444444446, | |
| "grad_norm": 0.26328030228614807, | |
| "learning_rate": 6.839506172839506e-05, | |
| "loss": 0.4308, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 2.3866666666666667, | |
| "grad_norm": 0.3249216675758362, | |
| "learning_rate": 6.814814814814814e-05, | |
| "loss": 0.464, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 2.388888888888889, | |
| "grad_norm": 0.2710109353065491, | |
| "learning_rate": 6.790123456790122e-05, | |
| "loss": 0.4386, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.391111111111111, | |
| "grad_norm": 0.23786287009716034, | |
| "learning_rate": 6.765432098765432e-05, | |
| "loss": 0.4537, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 2.3933333333333335, | |
| "grad_norm": 0.34233614802360535, | |
| "learning_rate": 6.74074074074074e-05, | |
| "loss": 0.6262, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 2.3955555555555557, | |
| "grad_norm": 0.2548952102661133, | |
| "learning_rate": 6.716049382716049e-05, | |
| "loss": 0.327, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 2.397777777777778, | |
| "grad_norm": 0.3713803291320801, | |
| "learning_rate": 6.691358024691357e-05, | |
| "loss": 0.5425, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 0.21818594634532928, | |
| "learning_rate": 6.666666666666666e-05, | |
| "loss": 0.4152, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.402222222222222, | |
| "grad_norm": 0.23173841834068298, | |
| "learning_rate": 6.641975308641975e-05, | |
| "loss": 0.4905, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 2.4044444444444446, | |
| "grad_norm": 0.2621210515499115, | |
| "learning_rate": 6.617283950617283e-05, | |
| "loss": 0.5208, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 2.4066666666666667, | |
| "grad_norm": 0.3177223801612854, | |
| "learning_rate": 6.592592592592591e-05, | |
| "loss": 0.489, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 2.408888888888889, | |
| "grad_norm": 0.2511189579963684, | |
| "learning_rate": 6.567901234567901e-05, | |
| "loss": 0.4022, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 2.411111111111111, | |
| "grad_norm": 0.20931217074394226, | |
| "learning_rate": 6.543209876543209e-05, | |
| "loss": 0.4887, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.413333333333333, | |
| "grad_norm": 0.2349865883588791, | |
| "learning_rate": 6.518518518518517e-05, | |
| "loss": 0.3361, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 2.4155555555555557, | |
| "grad_norm": 0.2636266052722931, | |
| "learning_rate": 6.493827160493827e-05, | |
| "loss": 0.397, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 2.417777777777778, | |
| "grad_norm": 0.23005585372447968, | |
| "learning_rate": 6.469135802469135e-05, | |
| "loss": 0.4901, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 0.22328707575798035, | |
| "learning_rate": 6.444444444444444e-05, | |
| "loss": 0.4861, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 2.422222222222222, | |
| "grad_norm": 0.26818522810935974, | |
| "learning_rate": 6.419753086419753e-05, | |
| "loss": 0.4856, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.4244444444444446, | |
| "grad_norm": 0.25317078828811646, | |
| "learning_rate": 6.395061728395061e-05, | |
| "loss": 0.5472, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 2.4266666666666667, | |
| "grad_norm": 0.23745451867580414, | |
| "learning_rate": 6.37037037037037e-05, | |
| "loss": 0.4735, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 2.428888888888889, | |
| "grad_norm": 0.282027006149292, | |
| "learning_rate": 6.345679012345678e-05, | |
| "loss": 0.4989, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 2.431111111111111, | |
| "grad_norm": 0.25692251324653625, | |
| "learning_rate": 6.320987654320987e-05, | |
| "loss": 0.3712, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 2.4333333333333336, | |
| "grad_norm": 0.29228779673576355, | |
| "learning_rate": 6.296296296296295e-05, | |
| "loss": 0.4607, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.4355555555555557, | |
| "grad_norm": 0.281011700630188, | |
| "learning_rate": 6.271604938271604e-05, | |
| "loss": 0.5867, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 2.437777777777778, | |
| "grad_norm": 0.2752225399017334, | |
| "learning_rate": 6.246913580246912e-05, | |
| "loss": 0.6769, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "grad_norm": 0.2541247010231018, | |
| "learning_rate": 6.222222222222222e-05, | |
| "loss": 0.4304, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 2.442222222222222, | |
| "grad_norm": 0.335610032081604, | |
| "learning_rate": 6.19753086419753e-05, | |
| "loss": 0.581, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 2.4444444444444446, | |
| "grad_norm": 0.24239715933799744, | |
| "learning_rate": 6.17283950617284e-05, | |
| "loss": 0.6073, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.4466666666666668, | |
| "grad_norm": 0.21203555166721344, | |
| "learning_rate": 6.148148148148148e-05, | |
| "loss": 0.3798, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 2.448888888888889, | |
| "grad_norm": 0.24926935136318207, | |
| "learning_rate": 6.123456790123456e-05, | |
| "loss": 0.4079, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 2.451111111111111, | |
| "grad_norm": 0.2610824704170227, | |
| "learning_rate": 6.0987654320987655e-05, | |
| "loss": 0.4797, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 2.453333333333333, | |
| "grad_norm": 0.32437193393707275, | |
| "learning_rate": 6.0740740740740736e-05, | |
| "loss": 0.603, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 2.4555555555555557, | |
| "grad_norm": 0.26519477367401123, | |
| "learning_rate": 6.049382716049382e-05, | |
| "loss": 0.4826, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.457777777777778, | |
| "grad_norm": 0.20365409553050995, | |
| "learning_rate": 6.0246913580246906e-05, | |
| "loss": 0.561, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "grad_norm": 0.28684762120246887, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 0.3947, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 2.462222222222222, | |
| "grad_norm": 0.307925283908844, | |
| "learning_rate": 5.975308641975308e-05, | |
| "loss": 0.439, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 2.464444444444444, | |
| "grad_norm": 0.544221818447113, | |
| "learning_rate": 5.9506172839506165e-05, | |
| "loss": 0.653, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 2.466666666666667, | |
| "grad_norm": 0.23246723413467407, | |
| "learning_rate": 5.925925925925925e-05, | |
| "loss": 0.456, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.468888888888889, | |
| "grad_norm": 0.2787330746650696, | |
| "learning_rate": 5.901234567901234e-05, | |
| "loss": 0.5098, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 2.471111111111111, | |
| "grad_norm": 0.28684481978416443, | |
| "learning_rate": 5.876543209876543e-05, | |
| "loss": 0.4113, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 2.473333333333333, | |
| "grad_norm": 0.23598532378673553, | |
| "learning_rate": 5.851851851851851e-05, | |
| "loss": 0.5585, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 2.4755555555555557, | |
| "grad_norm": 0.2952817380428314, | |
| "learning_rate": 5.827160493827159e-05, | |
| "loss": 0.3541, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 2.477777777777778, | |
| "grad_norm": 0.32788899540901184, | |
| "learning_rate": 5.802469135802469e-05, | |
| "loss": 0.4761, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 0.3333885371685028, | |
| "learning_rate": 5.777777777777777e-05, | |
| "loss": 0.4581, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 2.482222222222222, | |
| "grad_norm": 0.26001429557800293, | |
| "learning_rate": 5.753086419753086e-05, | |
| "loss": 0.4135, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 2.4844444444444447, | |
| "grad_norm": 0.23974114656448364, | |
| "learning_rate": 5.7283950617283946e-05, | |
| "loss": 0.453, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 2.486666666666667, | |
| "grad_norm": 0.22656071186065674, | |
| "learning_rate": 5.7037037037037035e-05, | |
| "loss": 0.4823, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 2.488888888888889, | |
| "grad_norm": 0.23035010695457458, | |
| "learning_rate": 5.6790123456790116e-05, | |
| "loss": 0.4934, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.491111111111111, | |
| "grad_norm": 0.2547858953475952, | |
| "learning_rate": 5.6543209876543205e-05, | |
| "loss": 0.4803, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 2.493333333333333, | |
| "grad_norm": 0.2811505198478699, | |
| "learning_rate": 5.629629629629629e-05, | |
| "loss": 0.4191, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 2.4955555555555557, | |
| "grad_norm": 0.21262358129024506, | |
| "learning_rate": 5.604938271604938e-05, | |
| "loss": 0.3076, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 2.497777777777778, | |
| "grad_norm": 0.25532275438308716, | |
| "learning_rate": 5.580246913580246e-05, | |
| "loss": 0.483, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.35907822847366333, | |
| "learning_rate": 5.5555555555555545e-05, | |
| "loss": 0.4565, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.502222222222222, | |
| "grad_norm": 0.29786059260368347, | |
| "learning_rate": 5.530864197530864e-05, | |
| "loss": 0.5644, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 2.5044444444444443, | |
| "grad_norm": 0.5043703317642212, | |
| "learning_rate": 5.506172839506172e-05, | |
| "loss": 0.4025, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 2.506666666666667, | |
| "grad_norm": 0.28630611300468445, | |
| "learning_rate": 5.481481481481481e-05, | |
| "loss": 0.4528, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 2.508888888888889, | |
| "grad_norm": 0.23132245242595673, | |
| "learning_rate": 5.456790123456789e-05, | |
| "loss": 0.5073, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 2.511111111111111, | |
| "grad_norm": 0.2920343279838562, | |
| "learning_rate": 5.4320987654320986e-05, | |
| "loss": 0.4471, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.513333333333333, | |
| "grad_norm": 0.22276970744132996, | |
| "learning_rate": 5.407407407407407e-05, | |
| "loss": 0.4888, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 2.5155555555555553, | |
| "grad_norm": 0.21300071477890015, | |
| "learning_rate": 5.3827160493827156e-05, | |
| "loss": 0.4099, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 2.517777777777778, | |
| "grad_norm": 0.2129749357700348, | |
| "learning_rate": 5.358024691358024e-05, | |
| "loss": 0.4451, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "grad_norm": 0.26365357637405396, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 0.3659, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 2.522222222222222, | |
| "grad_norm": 0.2541384994983673, | |
| "learning_rate": 5.3086419753086415e-05, | |
| "loss": 0.3353, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.5244444444444447, | |
| "grad_norm": 0.2970863878726959, | |
| "learning_rate": 5.2839506172839496e-05, | |
| "loss": 0.4897, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 2.5266666666666664, | |
| "grad_norm": 0.24692222476005554, | |
| "learning_rate": 5.259259259259259e-05, | |
| "loss": 0.392, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 2.528888888888889, | |
| "grad_norm": 0.29498088359832764, | |
| "learning_rate": 5.234567901234567e-05, | |
| "loss": 0.5025, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 2.531111111111111, | |
| "grad_norm": 0.2868523895740509, | |
| "learning_rate": 5.209876543209876e-05, | |
| "loss": 0.402, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 2.533333333333333, | |
| "grad_norm": 0.2951558828353882, | |
| "learning_rate": 5.185185185185184e-05, | |
| "loss": 0.464, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.535555555555556, | |
| "grad_norm": 0.24855990707874298, | |
| "learning_rate": 5.160493827160494e-05, | |
| "loss": 0.4004, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 2.537777777777778, | |
| "grad_norm": 0.2685883939266205, | |
| "learning_rate": 5.135802469135802e-05, | |
| "loss": 0.3987, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "grad_norm": 0.2569471001625061, | |
| "learning_rate": 5.111111111111111e-05, | |
| "loss": 0.5368, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 2.542222222222222, | |
| "grad_norm": 0.2672557830810547, | |
| "learning_rate": 5.086419753086419e-05, | |
| "loss": 0.4106, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 2.5444444444444443, | |
| "grad_norm": 0.2457388937473297, | |
| "learning_rate": 5.0617283950617285e-05, | |
| "loss": 0.4327, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.546666666666667, | |
| "grad_norm": 0.26166629791259766, | |
| "learning_rate": 5.0370370370370366e-05, | |
| "loss": 0.4788, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 2.548888888888889, | |
| "grad_norm": 0.25557467341423035, | |
| "learning_rate": 5.012345679012345e-05, | |
| "loss": 0.4385, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 2.551111111111111, | |
| "grad_norm": 0.29093366861343384, | |
| "learning_rate": 4.9876543209876536e-05, | |
| "loss": 0.4746, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 2.5533333333333332, | |
| "grad_norm": 0.2758121192455292, | |
| "learning_rate": 4.9629629629629625e-05, | |
| "loss": 0.4468, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 2.5555555555555554, | |
| "grad_norm": 0.23853829503059387, | |
| "learning_rate": 4.938271604938271e-05, | |
| "loss": 0.5304, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.557777777777778, | |
| "grad_norm": 0.27525413036346436, | |
| "learning_rate": 4.9135802469135795e-05, | |
| "loss": 0.464, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.22877416014671326, | |
| "learning_rate": 4.888888888888889e-05, | |
| "loss": 0.489, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 2.562222222222222, | |
| "grad_norm": 0.2996022403240204, | |
| "learning_rate": 4.864197530864197e-05, | |
| "loss": 0.4299, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 2.5644444444444443, | |
| "grad_norm": 0.3242427706718445, | |
| "learning_rate": 4.839506172839506e-05, | |
| "loss": 0.4783, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 2.5666666666666664, | |
| "grad_norm": 0.26138418912887573, | |
| "learning_rate": 4.814814814814814e-05, | |
| "loss": 0.402, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.568888888888889, | |
| "grad_norm": 0.2434043437242508, | |
| "learning_rate": 4.7901234567901237e-05, | |
| "loss": 0.4271, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 2.571111111111111, | |
| "grad_norm": 0.31792590022087097, | |
| "learning_rate": 4.765432098765432e-05, | |
| "loss": 0.5864, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 2.5733333333333333, | |
| "grad_norm": 0.2542998790740967, | |
| "learning_rate": 4.74074074074074e-05, | |
| "loss": 0.4059, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 2.575555555555556, | |
| "grad_norm": 0.2536453604698181, | |
| "learning_rate": 4.716049382716049e-05, | |
| "loss": 0.5285, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 2.5777777777777775, | |
| "grad_norm": 0.2503916025161743, | |
| "learning_rate": 4.6913580246913576e-05, | |
| "loss": 0.3913, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "grad_norm": 0.279352068901062, | |
| "learning_rate": 4.6666666666666665e-05, | |
| "loss": 0.5062, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 2.582222222222222, | |
| "grad_norm": 0.20147022604942322, | |
| "learning_rate": 4.6419753086419746e-05, | |
| "loss": 0.4216, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 2.5844444444444443, | |
| "grad_norm": 0.2574199438095093, | |
| "learning_rate": 4.6172839506172835e-05, | |
| "loss": 0.5535, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 2.586666666666667, | |
| "grad_norm": 0.255137175321579, | |
| "learning_rate": 4.592592592592592e-05, | |
| "loss": 0.4969, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 2.588888888888889, | |
| "grad_norm": 0.32849064469337463, | |
| "learning_rate": 4.567901234567901e-05, | |
| "loss": 0.5328, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.591111111111111, | |
| "grad_norm": 0.28691333532333374, | |
| "learning_rate": 4.543209876543209e-05, | |
| "loss": 0.5298, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 2.5933333333333333, | |
| "grad_norm": 0.2443169355392456, | |
| "learning_rate": 4.518518518518519e-05, | |
| "loss": 0.7071, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 2.5955555555555554, | |
| "grad_norm": 0.3008064925670624, | |
| "learning_rate": 4.493827160493827e-05, | |
| "loss": 0.4355, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 2.597777777777778, | |
| "grad_norm": 0.3086701035499573, | |
| "learning_rate": 4.469135802469135e-05, | |
| "loss": 0.4457, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 0.2844820022583008, | |
| "learning_rate": 4.444444444444444e-05, | |
| "loss": 0.667, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.602222222222222, | |
| "grad_norm": 0.22900676727294922, | |
| "learning_rate": 4.419753086419753e-05, | |
| "loss": 0.4843, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 2.6044444444444443, | |
| "grad_norm": 0.26003748178482056, | |
| "learning_rate": 4.3950617283950617e-05, | |
| "loss": 0.426, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 2.6066666666666665, | |
| "grad_norm": 0.2188059687614441, | |
| "learning_rate": 4.37037037037037e-05, | |
| "loss": 0.4291, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 2.608888888888889, | |
| "grad_norm": 0.3383055627346039, | |
| "learning_rate": 4.3456790123456786e-05, | |
| "loss": 0.4128, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 2.611111111111111, | |
| "grad_norm": 0.28404900431632996, | |
| "learning_rate": 4.3209876543209875e-05, | |
| "loss": 0.4426, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.6133333333333333, | |
| "grad_norm": 0.35723522305488586, | |
| "learning_rate": 4.296296296296296e-05, | |
| "loss": 0.4899, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 2.6155555555555554, | |
| "grad_norm": 0.29344046115875244, | |
| "learning_rate": 4.2716049382716045e-05, | |
| "loss": 0.4731, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 2.6177777777777775, | |
| "grad_norm": 0.2865029275417328, | |
| "learning_rate": 4.2469135802469126e-05, | |
| "loss": 0.3514, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "grad_norm": 0.24901816248893738, | |
| "learning_rate": 4.222222222222222e-05, | |
| "loss": 0.4674, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 2.6222222222222222, | |
| "grad_norm": 0.31357455253601074, | |
| "learning_rate": 4.19753086419753e-05, | |
| "loss": 0.5031, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.6244444444444444, | |
| "grad_norm": 0.3531203866004944, | |
| "learning_rate": 4.172839506172839e-05, | |
| "loss": 0.5071, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 2.626666666666667, | |
| "grad_norm": 0.2576693594455719, | |
| "learning_rate": 4.148148148148147e-05, | |
| "loss": 0.5263, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 2.628888888888889, | |
| "grad_norm": 0.2875922918319702, | |
| "learning_rate": 4.123456790123457e-05, | |
| "loss": 0.4736, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 2.631111111111111, | |
| "grad_norm": 0.2582704722881317, | |
| "learning_rate": 4.098765432098765e-05, | |
| "loss": 0.5004, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 2.6333333333333333, | |
| "grad_norm": 0.28706392645835876, | |
| "learning_rate": 4.074074074074074e-05, | |
| "loss": 0.3561, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.6355555555555554, | |
| "grad_norm": 0.23754924535751343, | |
| "learning_rate": 4.0493827160493827e-05, | |
| "loss": 0.414, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 2.637777777777778, | |
| "grad_norm": 0.2732125222682953, | |
| "learning_rate": 4.0246913580246915e-05, | |
| "loss": 0.4151, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 0.23749655485153198, | |
| "learning_rate": 3.9999999999999996e-05, | |
| "loss": 0.4638, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 2.6422222222222222, | |
| "grad_norm": 0.21151944994926453, | |
| "learning_rate": 3.975308641975308e-05, | |
| "loss": 0.3444, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 2.6444444444444444, | |
| "grad_norm": 0.2611617147922516, | |
| "learning_rate": 3.950617283950617e-05, | |
| "loss": 0.5069, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.6466666666666665, | |
| "grad_norm": 0.2559826672077179, | |
| "learning_rate": 3.9259259259259255e-05, | |
| "loss": 0.498, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 2.648888888888889, | |
| "grad_norm": 0.3297974765300751, | |
| "learning_rate": 3.901234567901234e-05, | |
| "loss": 0.3908, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 2.651111111111111, | |
| "grad_norm": 0.3367217481136322, | |
| "learning_rate": 3.8765432098765425e-05, | |
| "loss": 0.4256, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 2.6533333333333333, | |
| "grad_norm": 0.24523089826107025, | |
| "learning_rate": 3.851851851851852e-05, | |
| "loss": 0.454, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 2.6555555555555554, | |
| "grad_norm": 0.2197115421295166, | |
| "learning_rate": 3.82716049382716e-05, | |
| "loss": 0.5067, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.6577777777777776, | |
| "grad_norm": 0.3590328097343445, | |
| "learning_rate": 3.802469135802468e-05, | |
| "loss": 0.3838, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 0.25822263956069946, | |
| "learning_rate": 3.777777777777777e-05, | |
| "loss": 0.4362, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 2.6622222222222223, | |
| "grad_norm": 0.24154570698738098, | |
| "learning_rate": 3.7530864197530867e-05, | |
| "loss": 0.4231, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 2.6644444444444444, | |
| "grad_norm": 0.3014829754829407, | |
| "learning_rate": 3.728395061728395e-05, | |
| "loss": 0.548, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.23438327014446259, | |
| "learning_rate": 3.703703703703703e-05, | |
| "loss": 0.2957, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.6688888888888886, | |
| "grad_norm": 0.28357598185539246, | |
| "learning_rate": 3.679012345679012e-05, | |
| "loss": 0.7384, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 2.671111111111111, | |
| "grad_norm": 0.25190889835357666, | |
| "learning_rate": 3.6543209876543207e-05, | |
| "loss": 0.3769, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 2.6733333333333333, | |
| "grad_norm": 0.25062295794487, | |
| "learning_rate": 3.6296296296296295e-05, | |
| "loss": 0.4993, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 2.6755555555555555, | |
| "grad_norm": 0.3105320632457733, | |
| "learning_rate": 3.604938271604938e-05, | |
| "loss": 0.5487, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 2.677777777777778, | |
| "grad_norm": 0.3144405782222748, | |
| "learning_rate": 3.5802469135802465e-05, | |
| "loss": 0.4473, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 0.3181323707103729, | |
| "learning_rate": 3.555555555555555e-05, | |
| "loss": 0.478, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 2.6822222222222223, | |
| "grad_norm": 0.21852333843708038, | |
| "learning_rate": 3.5308641975308635e-05, | |
| "loss": 0.4886, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 2.6844444444444444, | |
| "grad_norm": 0.24893473088741302, | |
| "learning_rate": 3.506172839506172e-05, | |
| "loss": 0.3401, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 2.6866666666666665, | |
| "grad_norm": 0.6770753264427185, | |
| "learning_rate": 3.481481481481481e-05, | |
| "loss": 0.382, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 2.688888888888889, | |
| "grad_norm": 0.292156457901001, | |
| "learning_rate": 3.45679012345679e-05, | |
| "loss": 0.379, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.6911111111111112, | |
| "grad_norm": 0.31315043568611145, | |
| "learning_rate": 3.432098765432098e-05, | |
| "loss": 0.402, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 2.6933333333333334, | |
| "grad_norm": 0.29110002517700195, | |
| "learning_rate": 3.407407407407407e-05, | |
| "loss": 0.5616, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 2.6955555555555555, | |
| "grad_norm": 0.29066193103790283, | |
| "learning_rate": 3.382716049382716e-05, | |
| "loss": 0.5119, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 2.6977777777777776, | |
| "grad_norm": 0.25643596053123474, | |
| "learning_rate": 3.3580246913580247e-05, | |
| "loss": 0.4585, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "grad_norm": 0.23805151879787445, | |
| "learning_rate": 3.333333333333333e-05, | |
| "loss": 0.47, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.7022222222222223, | |
| "grad_norm": 0.300931453704834, | |
| "learning_rate": 3.3086419753086417e-05, | |
| "loss": 0.4589, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 2.7044444444444444, | |
| "grad_norm": 0.2649901807308197, | |
| "learning_rate": 3.2839506172839505e-05, | |
| "loss": 0.4979, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 2.7066666666666666, | |
| "grad_norm": 0.2810843884944916, | |
| "learning_rate": 3.2592592592592586e-05, | |
| "loss": 0.5062, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 2.7088888888888887, | |
| "grad_norm": 0.27211225032806396, | |
| "learning_rate": 3.2345679012345675e-05, | |
| "loss": 0.4575, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 2.7111111111111112, | |
| "grad_norm": 0.28907597064971924, | |
| "learning_rate": 3.209876543209876e-05, | |
| "loss": 0.376, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.7133333333333334, | |
| "grad_norm": 0.31166574358940125, | |
| "learning_rate": 3.185185185185185e-05, | |
| "loss": 0.5869, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 2.7155555555555555, | |
| "grad_norm": 0.27895674109458923, | |
| "learning_rate": 3.160493827160493e-05, | |
| "loss": 0.5236, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 2.7177777777777776, | |
| "grad_norm": 0.41477781534194946, | |
| "learning_rate": 3.135802469135802e-05, | |
| "loss": 0.5468, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 2.7199999999999998, | |
| "grad_norm": 0.2354549616575241, | |
| "learning_rate": 3.111111111111111e-05, | |
| "loss": 0.3541, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 2.7222222222222223, | |
| "grad_norm": 0.2622091472148895, | |
| "learning_rate": 3.08641975308642e-05, | |
| "loss": 0.5945, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.7244444444444444, | |
| "grad_norm": 0.2661958634853363, | |
| "learning_rate": 3.061728395061728e-05, | |
| "loss": 0.4496, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 2.7266666666666666, | |
| "grad_norm": 0.1839124858379364, | |
| "learning_rate": 3.0370370370370368e-05, | |
| "loss": 0.4761, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 2.728888888888889, | |
| "grad_norm": 0.21911154687404633, | |
| "learning_rate": 3.0123456790123453e-05, | |
| "loss": 0.4264, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 2.7311111111111113, | |
| "grad_norm": 0.3091849386692047, | |
| "learning_rate": 2.987654320987654e-05, | |
| "loss": 0.4087, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 2.7333333333333334, | |
| "grad_norm": 0.2811782956123352, | |
| "learning_rate": 2.9629629629629627e-05, | |
| "loss": 0.4906, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.7355555555555555, | |
| "grad_norm": 0.27253296971321106, | |
| "learning_rate": 2.9382716049382715e-05, | |
| "loss": 0.5261, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 2.7377777777777776, | |
| "grad_norm": 0.29757750034332275, | |
| "learning_rate": 2.9135802469135796e-05, | |
| "loss": 0.6007, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 0.3162846565246582, | |
| "learning_rate": 2.8888888888888885e-05, | |
| "loss": 0.4949, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 2.7422222222222223, | |
| "grad_norm": 0.2618942856788635, | |
| "learning_rate": 2.8641975308641973e-05, | |
| "loss": 0.4121, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 2.7444444444444445, | |
| "grad_norm": 0.24067893624305725, | |
| "learning_rate": 2.8395061728395058e-05, | |
| "loss": 0.4042, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.7466666666666666, | |
| "grad_norm": 0.3019217252731323, | |
| "learning_rate": 2.8148148148148147e-05, | |
| "loss": 0.4992, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 2.7488888888888887, | |
| "grad_norm": 0.21723076701164246, | |
| "learning_rate": 2.790123456790123e-05, | |
| "loss": 0.4366, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 2.7511111111111113, | |
| "grad_norm": 0.34433475136756897, | |
| "learning_rate": 2.765432098765432e-05, | |
| "loss": 0.3907, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 2.7533333333333334, | |
| "grad_norm": 0.23262938857078552, | |
| "learning_rate": 2.7407407407407405e-05, | |
| "loss": 0.4287, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 2.7555555555555555, | |
| "grad_norm": 0.2457321286201477, | |
| "learning_rate": 2.7160493827160493e-05, | |
| "loss": 0.2756, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.7577777777777777, | |
| "grad_norm": 0.27166110277175903, | |
| "learning_rate": 2.6913580246913578e-05, | |
| "loss": 0.4958, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 0.41035521030426025, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.3064, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 2.7622222222222224, | |
| "grad_norm": 0.23718339204788208, | |
| "learning_rate": 2.6419753086419748e-05, | |
| "loss": 0.4731, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 2.7644444444444445, | |
| "grad_norm": 0.4064386188983917, | |
| "learning_rate": 2.6172839506172837e-05, | |
| "loss": 0.3994, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 2.7666666666666666, | |
| "grad_norm": 0.1951911300420761, | |
| "learning_rate": 2.592592592592592e-05, | |
| "loss": 0.4474, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.7688888888888887, | |
| "grad_norm": 0.21489310264587402, | |
| "learning_rate": 2.567901234567901e-05, | |
| "loss": 0.4107, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 2.771111111111111, | |
| "grad_norm": 0.20124374330043793, | |
| "learning_rate": 2.5432098765432095e-05, | |
| "loss": 0.4534, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 2.7733333333333334, | |
| "grad_norm": 0.33558300137519836, | |
| "learning_rate": 2.5185185185185183e-05, | |
| "loss": 0.451, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 2.7755555555555556, | |
| "grad_norm": 0.24915766716003418, | |
| "learning_rate": 2.4938271604938268e-05, | |
| "loss": 0.3313, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 0.18676912784576416, | |
| "learning_rate": 2.4691358024691357e-05, | |
| "loss": 0.5069, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.7800000000000002, | |
| "grad_norm": 0.3069996237754822, | |
| "learning_rate": 2.4444444444444445e-05, | |
| "loss": 0.4469, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 2.7822222222222224, | |
| "grad_norm": 0.35303255915641785, | |
| "learning_rate": 2.419753086419753e-05, | |
| "loss": 0.6305, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 2.7844444444444445, | |
| "grad_norm": 0.24550585448741913, | |
| "learning_rate": 2.3950617283950618e-05, | |
| "loss": 0.3975, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 2.7866666666666666, | |
| "grad_norm": 0.2411435842514038, | |
| "learning_rate": 2.37037037037037e-05, | |
| "loss": 0.4139, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 2.7888888888888888, | |
| "grad_norm": 0.34191983938217163, | |
| "learning_rate": 2.3456790123456788e-05, | |
| "loss": 0.5396, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.7911111111111113, | |
| "grad_norm": 0.23888371884822845, | |
| "learning_rate": 2.3209876543209873e-05, | |
| "loss": 0.4926, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 2.7933333333333334, | |
| "grad_norm": 0.19129949808120728, | |
| "learning_rate": 2.296296296296296e-05, | |
| "loss": 0.4754, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 2.7955555555555556, | |
| "grad_norm": 0.2999621033668518, | |
| "learning_rate": 2.2716049382716047e-05, | |
| "loss": 0.6211, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 2.7977777777777777, | |
| "grad_norm": 0.31490370631217957, | |
| "learning_rate": 2.2469135802469135e-05, | |
| "loss": 0.4823, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 0.2625960409641266, | |
| "learning_rate": 2.222222222222222e-05, | |
| "loss": 0.5804, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.8022222222222224, | |
| "grad_norm": 0.2711751163005829, | |
| "learning_rate": 2.1975308641975308e-05, | |
| "loss": 0.4685, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 2.8044444444444445, | |
| "grad_norm": 0.2509281039237976, | |
| "learning_rate": 2.1728395061728393e-05, | |
| "loss": 0.54, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 2.8066666666666666, | |
| "grad_norm": 0.24419257044792175, | |
| "learning_rate": 2.148148148148148e-05, | |
| "loss": 0.6527, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 2.8088888888888888, | |
| "grad_norm": 0.26806503534317017, | |
| "learning_rate": 2.1234567901234563e-05, | |
| "loss": 0.4227, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 2.811111111111111, | |
| "grad_norm": 0.24897365272045135, | |
| "learning_rate": 2.098765432098765e-05, | |
| "loss": 0.3335, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.8133333333333335, | |
| "grad_norm": 0.21197307109832764, | |
| "learning_rate": 2.0740740740740737e-05, | |
| "loss": 0.3434, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 2.8155555555555556, | |
| "grad_norm": 0.25956013798713684, | |
| "learning_rate": 2.0493827160493825e-05, | |
| "loss": 0.4276, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 2.8177777777777777, | |
| "grad_norm": 0.25997549295425415, | |
| "learning_rate": 2.0246913580246913e-05, | |
| "loss": 0.3793, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 0.3059636056423187, | |
| "learning_rate": 1.9999999999999998e-05, | |
| "loss": 0.381, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 2.822222222222222, | |
| "grad_norm": 0.22168779373168945, | |
| "learning_rate": 1.9753086419753087e-05, | |
| "loss": 0.4418, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.8244444444444445, | |
| "grad_norm": 0.26615574955940247, | |
| "learning_rate": 1.950617283950617e-05, | |
| "loss": 0.4083, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 2.8266666666666667, | |
| "grad_norm": 0.2455672025680542, | |
| "learning_rate": 1.925925925925926e-05, | |
| "loss": 0.4181, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 2.828888888888889, | |
| "grad_norm": 0.2567649483680725, | |
| "learning_rate": 1.901234567901234e-05, | |
| "loss": 0.5391, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 2.8311111111111114, | |
| "grad_norm": 0.2008361965417862, | |
| "learning_rate": 1.8765432098765433e-05, | |
| "loss": 0.3542, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 2.8333333333333335, | |
| "grad_norm": 0.29503872990608215, | |
| "learning_rate": 1.8518518518518515e-05, | |
| "loss": 0.3966, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.8355555555555556, | |
| "grad_norm": 0.29749149084091187, | |
| "learning_rate": 1.8271604938271603e-05, | |
| "loss": 0.5504, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 2.8377777777777777, | |
| "grad_norm": 0.78473299741745, | |
| "learning_rate": 1.802469135802469e-05, | |
| "loss": 0.3548, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 0.3943696618080139, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 0.4358, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 2.8422222222222224, | |
| "grad_norm": 0.3001830279827118, | |
| "learning_rate": 1.753086419753086e-05, | |
| "loss": 0.4179, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 2.8444444444444446, | |
| "grad_norm": 0.33909693360328674, | |
| "learning_rate": 1.728395061728395e-05, | |
| "loss": 0.4276, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.8466666666666667, | |
| "grad_norm": 0.33997949957847595, | |
| "learning_rate": 1.7037037037037035e-05, | |
| "loss": 0.4834, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 2.848888888888889, | |
| "grad_norm": 0.24325981736183167, | |
| "learning_rate": 1.6790123456790123e-05, | |
| "loss": 0.4327, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 2.851111111111111, | |
| "grad_norm": 0.26005110144615173, | |
| "learning_rate": 1.6543209876543208e-05, | |
| "loss": 0.4714, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 2.8533333333333335, | |
| "grad_norm": 0.34214770793914795, | |
| "learning_rate": 1.6296296296296293e-05, | |
| "loss": 0.4656, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 2.8555555555555556, | |
| "grad_norm": 0.32353708148002625, | |
| "learning_rate": 1.604938271604938e-05, | |
| "loss": 0.4362, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.8577777777777778, | |
| "grad_norm": 0.27080851793289185, | |
| "learning_rate": 1.5802469135802467e-05, | |
| "loss": 0.3997, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "grad_norm": 0.2615989148616791, | |
| "learning_rate": 1.5555555555555555e-05, | |
| "loss": 0.3359, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 2.862222222222222, | |
| "grad_norm": 0.22018349170684814, | |
| "learning_rate": 1.530864197530864e-05, | |
| "loss": 0.3581, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 2.8644444444444446, | |
| "grad_norm": 0.280860960483551, | |
| "learning_rate": 1.5061728395061727e-05, | |
| "loss": 0.4006, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 2.8666666666666667, | |
| "grad_norm": 0.2675721049308777, | |
| "learning_rate": 1.4814814814814813e-05, | |
| "loss": 0.4196, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.868888888888889, | |
| "grad_norm": 0.22121883928775787, | |
| "learning_rate": 1.4567901234567898e-05, | |
| "loss": 0.4088, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 2.871111111111111, | |
| "grad_norm": 0.2731737196445465, | |
| "learning_rate": 1.4320987654320987e-05, | |
| "loss": 0.3216, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 2.873333333333333, | |
| "grad_norm": 0.24813485145568848, | |
| "learning_rate": 1.4074074074074073e-05, | |
| "loss": 0.3485, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 2.8755555555555556, | |
| "grad_norm": 0.2645670473575592, | |
| "learning_rate": 1.382716049382716e-05, | |
| "loss": 0.4655, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 2.8777777777777778, | |
| "grad_norm": 0.2540053725242615, | |
| "learning_rate": 1.3580246913580247e-05, | |
| "loss": 0.368, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 0.3078966736793518, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.498, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 2.8822222222222225, | |
| "grad_norm": 0.2583654224872589, | |
| "learning_rate": 1.3086419753086418e-05, | |
| "loss": 0.5326, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 2.8844444444444446, | |
| "grad_norm": 0.2601797878742218, | |
| "learning_rate": 1.2839506172839505e-05, | |
| "loss": 0.4029, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 2.8866666666666667, | |
| "grad_norm": 0.24041469395160675, | |
| "learning_rate": 1.2592592592592592e-05, | |
| "loss": 0.4195, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 2.888888888888889, | |
| "grad_norm": 0.3383660316467285, | |
| "learning_rate": 1.2345679012345678e-05, | |
| "loss": 0.4559, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.891111111111111, | |
| "grad_norm": 0.2917707860469818, | |
| "learning_rate": 1.2098765432098765e-05, | |
| "loss": 0.4998, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 2.8933333333333335, | |
| "grad_norm": 0.29064223170280457, | |
| "learning_rate": 1.185185185185185e-05, | |
| "loss": 0.4735, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 2.8955555555555557, | |
| "grad_norm": 0.28389549255371094, | |
| "learning_rate": 1.1604938271604937e-05, | |
| "loss": 0.5133, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 2.897777777777778, | |
| "grad_norm": 0.2980717718601227, | |
| "learning_rate": 1.1358024691358023e-05, | |
| "loss": 0.5401, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.24981656670570374, | |
| "learning_rate": 1.111111111111111e-05, | |
| "loss": 0.5033, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.902222222222222, | |
| "grad_norm": 0.2664763331413269, | |
| "learning_rate": 1.0864197530864197e-05, | |
| "loss": 0.4684, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 2.9044444444444446, | |
| "grad_norm": 0.277413547039032, | |
| "learning_rate": 1.0617283950617282e-05, | |
| "loss": 0.4596, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 2.9066666666666667, | |
| "grad_norm": 0.2499413788318634, | |
| "learning_rate": 1.0370370370370368e-05, | |
| "loss": 0.3985, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 2.908888888888889, | |
| "grad_norm": 0.2742132842540741, | |
| "learning_rate": 1.0123456790123457e-05, | |
| "loss": 0.5249, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 2.911111111111111, | |
| "grad_norm": 0.24763798713684082, | |
| "learning_rate": 9.876543209876543e-06, | |
| "loss": 0.5399, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.913333333333333, | |
| "grad_norm": 0.3568689823150635, | |
| "learning_rate": 9.62962962962963e-06, | |
| "loss": 0.4471, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 2.9155555555555557, | |
| "grad_norm": 0.26705747842788696, | |
| "learning_rate": 9.382716049382717e-06, | |
| "loss": 0.3619, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 2.917777777777778, | |
| "grad_norm": 0.2750731110572815, | |
| "learning_rate": 9.135802469135802e-06, | |
| "loss": 0.3774, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 0.29260754585266113, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.4241, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 2.9222222222222225, | |
| "grad_norm": 0.28297215700149536, | |
| "learning_rate": 8.641975308641975e-06, | |
| "loss": 0.4242, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.924444444444444, | |
| "grad_norm": 0.30213314294815063, | |
| "learning_rate": 8.395061728395062e-06, | |
| "loss": 0.511, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 2.9266666666666667, | |
| "grad_norm": 0.29815974831581116, | |
| "learning_rate": 8.148148148148147e-06, | |
| "loss": 0.5409, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 2.928888888888889, | |
| "grad_norm": 0.28545916080474854, | |
| "learning_rate": 7.901234567901233e-06, | |
| "loss": 0.4803, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 2.931111111111111, | |
| "grad_norm": 0.2818659543991089, | |
| "learning_rate": 7.65432098765432e-06, | |
| "loss": 0.4151, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 2.9333333333333336, | |
| "grad_norm": 0.26690900325775146, | |
| "learning_rate": 7.407407407407407e-06, | |
| "loss": 0.4973, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.9355555555555557, | |
| "grad_norm": 0.2489357590675354, | |
| "learning_rate": 7.160493827160493e-06, | |
| "loss": 0.4027, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 2.937777777777778, | |
| "grad_norm": 0.2781828045845032, | |
| "learning_rate": 6.91358024691358e-06, | |
| "loss": 0.6148, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "grad_norm": 0.32088056206703186, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.6265, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 2.942222222222222, | |
| "grad_norm": 0.2306426763534546, | |
| "learning_rate": 6.4197530864197525e-06, | |
| "loss": 0.4634, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 2.9444444444444446, | |
| "grad_norm": 0.2687436044216156, | |
| "learning_rate": 6.172839506172839e-06, | |
| "loss": 0.4067, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.9466666666666668, | |
| "grad_norm": 0.3024968206882477, | |
| "learning_rate": 5.925925925925925e-06, | |
| "loss": 0.4939, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 2.948888888888889, | |
| "grad_norm": 0.25185179710388184, | |
| "learning_rate": 5.679012345679012e-06, | |
| "loss": 0.5607, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 2.951111111111111, | |
| "grad_norm": 0.31656819581985474, | |
| "learning_rate": 5.432098765432098e-06, | |
| "loss": 0.56, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 2.953333333333333, | |
| "grad_norm": 0.33628830313682556, | |
| "learning_rate": 5.185185185185184e-06, | |
| "loss": 0.4529, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 2.9555555555555557, | |
| "grad_norm": 0.2628391683101654, | |
| "learning_rate": 4.938271604938272e-06, | |
| "loss": 0.4202, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.957777777777778, | |
| "grad_norm": 0.3400716483592987, | |
| "learning_rate": 4.691358024691358e-06, | |
| "loss": 0.5591, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 0.33360472321510315, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.4352, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 2.962222222222222, | |
| "grad_norm": 0.270183801651001, | |
| "learning_rate": 4.197530864197531e-06, | |
| "loss": 0.4192, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 2.964444444444444, | |
| "grad_norm": 0.3092866539955139, | |
| "learning_rate": 3.950617283950617e-06, | |
| "loss": 0.4532, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 2.966666666666667, | |
| "grad_norm": 0.4208994209766388, | |
| "learning_rate": 3.7037037037037033e-06, | |
| "loss": 0.5659, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.968888888888889, | |
| "grad_norm": 0.2298194020986557, | |
| "learning_rate": 3.45679012345679e-06, | |
| "loss": 0.3772, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 2.971111111111111, | |
| "grad_norm": 0.22611771523952484, | |
| "learning_rate": 3.2098765432098762e-06, | |
| "loss": 0.3389, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 2.9733333333333336, | |
| "grad_norm": 0.328031450510025, | |
| "learning_rate": 2.9629629629629625e-06, | |
| "loss": 0.3958, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 2.9755555555555553, | |
| "grad_norm": 0.2716084122657776, | |
| "learning_rate": 2.716049382716049e-06, | |
| "loss": 0.5908, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 2.977777777777778, | |
| "grad_norm": 0.2897849678993225, | |
| "learning_rate": 2.469135802469136e-06, | |
| "loss": 0.5129, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 0.33624279499053955, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.4486, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 2.982222222222222, | |
| "grad_norm": 0.2882382571697235, | |
| "learning_rate": 1.9753086419753083e-06, | |
| "loss": 0.3876, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 2.9844444444444447, | |
| "grad_norm": 0.22958910465240479, | |
| "learning_rate": 1.728395061728395e-06, | |
| "loss": 0.3783, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 2.986666666666667, | |
| "grad_norm": 0.2548253536224365, | |
| "learning_rate": 1.4814814814814812e-06, | |
| "loss": 0.3439, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 2.988888888888889, | |
| "grad_norm": 0.31680935621261597, | |
| "learning_rate": 1.234567901234568e-06, | |
| "loss": 0.4421, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 2.991111111111111, | |
| "grad_norm": 0.3343510627746582, | |
| "learning_rate": 9.876543209876542e-07, | |
| "loss": 0.6147, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 2.993333333333333, | |
| "grad_norm": 0.2711341381072998, | |
| "learning_rate": 7.407407407407406e-07, | |
| "loss": 0.456, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 2.9955555555555557, | |
| "grad_norm": 0.25026512145996094, | |
| "learning_rate": 4.938271604938271e-07, | |
| "loss": 0.5292, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 2.997777777777778, | |
| "grad_norm": 0.2671927511692047, | |
| "learning_rate": 2.4691358024691354e-07, | |
| "loss": 0.2949, | |
| "step": 1349 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.2514517605304718, | |
| "learning_rate": 0.0, | |
| "loss": 0.3747, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 1350, | |
| "total_flos": 5.0942943348496794e+17, | |
| "train_loss": 0.5073228121245349, | |
| "train_runtime": 3670.8069, | |
| "train_samples_per_second": 5.883, | |
| "train_steps_per_second": 0.368 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 1350, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.0942943348496794e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |