{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "eval_steps": 0, "global_step": 1992, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0015060240963855422, "grad_norm": 0.39453125, "learning_rate": 9.994979919678715e-06, "loss": 1.5932, "step": 1 }, { "epoch": 0.0030120481927710845, "grad_norm": 0.3828125, "learning_rate": 9.98995983935743e-06, "loss": 1.606, "step": 2 }, { "epoch": 0.004518072289156626, "grad_norm": 0.388671875, "learning_rate": 9.984939759036144e-06, "loss": 1.6286, "step": 3 }, { "epoch": 0.006024096385542169, "grad_norm": 0.65234375, "learning_rate": 9.97991967871486e-06, "loss": 1.5252, "step": 4 }, { "epoch": 0.007530120481927711, "grad_norm": 0.373046875, "learning_rate": 9.974899598393575e-06, "loss": 1.5902, "step": 5 }, { "epoch": 0.009036144578313253, "grad_norm": 0.345703125, "learning_rate": 9.96987951807229e-06, "loss": 1.5367, "step": 6 }, { "epoch": 0.010542168674698794, "grad_norm": 0.359375, "learning_rate": 9.964859437751004e-06, "loss": 1.654, "step": 7 }, { "epoch": 0.012048192771084338, "grad_norm": 0.357421875, "learning_rate": 9.959839357429719e-06, "loss": 1.6527, "step": 8 }, { "epoch": 0.01355421686746988, "grad_norm": 0.32421875, "learning_rate": 9.954819277108435e-06, "loss": 1.5265, "step": 9 }, { "epoch": 0.015060240963855422, "grad_norm": 0.30859375, "learning_rate": 9.94979919678715e-06, "loss": 1.4826, "step": 10 }, { "epoch": 0.016566265060240965, "grad_norm": 0.3046875, "learning_rate": 9.944779116465864e-06, "loss": 1.5049, "step": 11 }, { "epoch": 0.018072289156626505, "grad_norm": 0.255859375, "learning_rate": 9.93975903614458e-06, "loss": 1.4511, "step": 12 }, { "epoch": 0.01957831325301205, "grad_norm": 0.26953125, "learning_rate": 9.934738955823295e-06, "loss": 1.522, "step": 13 }, { "epoch": 0.02108433734939759, "grad_norm": 0.25, "learning_rate": 9.92971887550201e-06, "loss": 1.4482, "step": 14 }, { "epoch": 0.022590361445783132, "grad_norm": 0.2451171875, "learning_rate": 9.924698795180724e-06, "loss": 1.4168, "step": 15 }, { "epoch": 0.024096385542168676, "grad_norm": 0.25390625, "learning_rate": 9.919678714859438e-06, "loss": 1.4518, "step": 16 }, { "epoch": 0.025602409638554216, "grad_norm": 0.27734375, "learning_rate": 9.914658634538153e-06, "loss": 1.3936, "step": 17 }, { "epoch": 0.02710843373493976, "grad_norm": 0.265625, "learning_rate": 9.909638554216869e-06, "loss": 1.4488, "step": 18 }, { "epoch": 0.0286144578313253, "grad_norm": 0.25390625, "learning_rate": 9.904618473895584e-06, "loss": 1.3833, "step": 19 }, { "epoch": 0.030120481927710843, "grad_norm": 0.25, "learning_rate": 9.899598393574298e-06, "loss": 1.384, "step": 20 }, { "epoch": 0.03162650602409638, "grad_norm": 0.25390625, "learning_rate": 9.894578313253013e-06, "loss": 1.4026, "step": 21 }, { "epoch": 0.03313253012048193, "grad_norm": 0.2353515625, "learning_rate": 9.889558232931727e-06, "loss": 1.3655, "step": 22 }, { "epoch": 0.03463855421686747, "grad_norm": 0.232421875, "learning_rate": 9.884538152610442e-06, "loss": 1.3341, "step": 23 }, { "epoch": 0.03614457831325301, "grad_norm": 0.216796875, "learning_rate": 9.879518072289156e-06, "loss": 1.3796, "step": 24 }, { "epoch": 0.03765060240963856, "grad_norm": 0.2138671875, "learning_rate": 9.874497991967872e-06, "loss": 1.4216, "step": 25 }, { "epoch": 0.0391566265060241, "grad_norm": 0.2060546875, "learning_rate": 9.869477911646587e-06, "loss": 1.3679, "step": 26 }, { "epoch": 0.04066265060240964, "grad_norm": 0.2275390625, "learning_rate": 9.864457831325303e-06, "loss": 1.3756, "step": 27 }, { "epoch": 0.04216867469879518, "grad_norm": 0.208984375, "learning_rate": 9.859437751004018e-06, "loss": 1.3403, "step": 28 }, { "epoch": 0.043674698795180725, "grad_norm": 0.1904296875, "learning_rate": 9.854417670682732e-06, "loss": 1.2785, "step": 29 }, { "epoch": 0.045180722891566265, "grad_norm": 0.1845703125, "learning_rate": 9.849397590361447e-06, "loss": 1.2673, "step": 30 }, { "epoch": 0.046686746987951805, "grad_norm": 0.1943359375, "learning_rate": 9.844377510040161e-06, "loss": 1.2827, "step": 31 }, { "epoch": 0.04819277108433735, "grad_norm": 0.2197265625, "learning_rate": 9.839357429718876e-06, "loss": 1.2705, "step": 32 }, { "epoch": 0.04969879518072289, "grad_norm": 0.181640625, "learning_rate": 9.834337349397592e-06, "loss": 1.3024, "step": 33 }, { "epoch": 0.05120481927710843, "grad_norm": 0.1865234375, "learning_rate": 9.829317269076307e-06, "loss": 1.2268, "step": 34 }, { "epoch": 0.05271084337349398, "grad_norm": 0.1767578125, "learning_rate": 9.824297188755021e-06, "loss": 1.2543, "step": 35 }, { "epoch": 0.05421686746987952, "grad_norm": 0.1806640625, "learning_rate": 9.819277108433736e-06, "loss": 1.2749, "step": 36 }, { "epoch": 0.05572289156626506, "grad_norm": 0.263671875, "learning_rate": 9.81425702811245e-06, "loss": 1.2143, "step": 37 }, { "epoch": 0.0572289156626506, "grad_norm": 0.169921875, "learning_rate": 9.809236947791165e-06, "loss": 1.2441, "step": 38 }, { "epoch": 0.058734939759036146, "grad_norm": 0.1669921875, "learning_rate": 9.804216867469881e-06, "loss": 1.2035, "step": 39 }, { "epoch": 0.060240963855421686, "grad_norm": 0.166015625, "learning_rate": 9.799196787148595e-06, "loss": 1.2236, "step": 40 }, { "epoch": 0.061746987951807226, "grad_norm": 0.1708984375, "learning_rate": 9.79417670682731e-06, "loss": 1.2486, "step": 41 }, { "epoch": 0.06325301204819277, "grad_norm": 0.18359375, "learning_rate": 9.789156626506024e-06, "loss": 1.2194, "step": 42 }, { "epoch": 0.06475903614457831, "grad_norm": 0.1669921875, "learning_rate": 9.784136546184739e-06, "loss": 1.2326, "step": 43 }, { "epoch": 0.06626506024096386, "grad_norm": 0.283203125, "learning_rate": 9.779116465863454e-06, "loss": 1.2236, "step": 44 }, { "epoch": 0.0677710843373494, "grad_norm": 0.1689453125, "learning_rate": 9.77409638554217e-06, "loss": 1.2386, "step": 45 }, { "epoch": 0.06927710843373494, "grad_norm": 0.1650390625, "learning_rate": 9.769076305220884e-06, "loss": 1.2123, "step": 46 }, { "epoch": 0.07078313253012049, "grad_norm": 0.177734375, "learning_rate": 9.764056224899599e-06, "loss": 1.2109, "step": 47 }, { "epoch": 0.07228915662650602, "grad_norm": 0.16015625, "learning_rate": 9.759036144578315e-06, "loss": 1.1781, "step": 48 }, { "epoch": 0.07379518072289157, "grad_norm": 0.1650390625, "learning_rate": 9.75401606425703e-06, "loss": 1.2171, "step": 49 }, { "epoch": 0.07530120481927711, "grad_norm": 0.1689453125, "learning_rate": 9.748995983935744e-06, "loss": 1.1435, "step": 50 }, { "epoch": 0.07680722891566265, "grad_norm": 0.1533203125, "learning_rate": 9.743975903614459e-06, "loss": 1.1986, "step": 51 }, { "epoch": 0.0783132530120482, "grad_norm": 0.15625, "learning_rate": 9.738955823293173e-06, "loss": 1.1341, "step": 52 }, { "epoch": 0.07981927710843373, "grad_norm": 0.1513671875, "learning_rate": 9.733935742971888e-06, "loss": 1.1576, "step": 53 }, { "epoch": 0.08132530120481928, "grad_norm": 0.171875, "learning_rate": 9.728915662650604e-06, "loss": 1.1739, "step": 54 }, { "epoch": 0.08283132530120482, "grad_norm": 0.28515625, "learning_rate": 9.723895582329318e-06, "loss": 1.178, "step": 55 }, { "epoch": 0.08433734939759036, "grad_norm": 0.16015625, "learning_rate": 9.718875502008033e-06, "loss": 1.1974, "step": 56 }, { "epoch": 0.0858433734939759, "grad_norm": 0.17578125, "learning_rate": 9.713855421686747e-06, "loss": 1.155, "step": 57 }, { "epoch": 0.08734939759036145, "grad_norm": 0.154296875, "learning_rate": 9.708835341365462e-06, "loss": 1.1368, "step": 58 }, { "epoch": 0.08885542168674698, "grad_norm": 0.1640625, "learning_rate": 9.703815261044177e-06, "loss": 1.1419, "step": 59 }, { "epoch": 0.09036144578313253, "grad_norm": 0.150390625, "learning_rate": 9.698795180722893e-06, "loss": 1.1508, "step": 60 }, { "epoch": 0.09186746987951808, "grad_norm": 0.1591796875, "learning_rate": 9.693775100401607e-06, "loss": 1.1879, "step": 61 }, { "epoch": 0.09337349397590361, "grad_norm": 0.1572265625, "learning_rate": 9.688755020080322e-06, "loss": 1.1514, "step": 62 }, { "epoch": 0.09487951807228916, "grad_norm": 0.1533203125, "learning_rate": 9.683734939759038e-06, "loss": 1.1118, "step": 63 }, { "epoch": 0.0963855421686747, "grad_norm": 0.1494140625, "learning_rate": 9.678714859437753e-06, "loss": 1.1248, "step": 64 }, { "epoch": 0.09789156626506024, "grad_norm": 0.15625, "learning_rate": 9.673694779116467e-06, "loss": 1.1078, "step": 65 }, { "epoch": 0.09939759036144578, "grad_norm": 0.1533203125, "learning_rate": 9.668674698795182e-06, "loss": 1.1114, "step": 66 }, { "epoch": 0.10090361445783133, "grad_norm": 0.166015625, "learning_rate": 9.663654618473896e-06, "loss": 1.1493, "step": 67 }, { "epoch": 0.10240963855421686, "grad_norm": 0.146484375, "learning_rate": 9.65863453815261e-06, "loss": 1.0754, "step": 68 }, { "epoch": 0.10391566265060241, "grad_norm": 0.1533203125, "learning_rate": 9.653614457831327e-06, "loss": 1.0923, "step": 69 }, { "epoch": 0.10542168674698796, "grad_norm": 0.1982421875, "learning_rate": 9.648594377510041e-06, "loss": 1.1163, "step": 70 }, { "epoch": 0.10692771084337349, "grad_norm": 0.19140625, "learning_rate": 9.643574297188756e-06, "loss": 1.1058, "step": 71 }, { "epoch": 0.10843373493975904, "grad_norm": 0.14453125, "learning_rate": 9.63855421686747e-06, "loss": 1.0821, "step": 72 }, { "epoch": 0.10993975903614457, "grad_norm": 0.146484375, "learning_rate": 9.633534136546185e-06, "loss": 1.083, "step": 73 }, { "epoch": 0.11144578313253012, "grad_norm": 0.2060546875, "learning_rate": 9.6285140562249e-06, "loss": 1.1067, "step": 74 }, { "epoch": 0.11295180722891567, "grad_norm": 0.1552734375, "learning_rate": 9.623493975903616e-06, "loss": 1.1463, "step": 75 }, { "epoch": 0.1144578313253012, "grad_norm": 0.1416015625, "learning_rate": 9.61847389558233e-06, "loss": 1.0727, "step": 76 }, { "epoch": 0.11596385542168675, "grad_norm": 0.1689453125, "learning_rate": 9.613453815261045e-06, "loss": 1.1306, "step": 77 }, { "epoch": 0.11746987951807229, "grad_norm": 0.1591796875, "learning_rate": 9.60843373493976e-06, "loss": 1.0931, "step": 78 }, { "epoch": 0.11897590361445783, "grad_norm": 0.1650390625, "learning_rate": 9.603413654618474e-06, "loss": 1.1027, "step": 79 }, { "epoch": 0.12048192771084337, "grad_norm": 0.1982421875, "learning_rate": 9.598393574297188e-06, "loss": 1.0994, "step": 80 }, { "epoch": 0.12198795180722892, "grad_norm": 0.1630859375, "learning_rate": 9.593373493975905e-06, "loss": 1.0964, "step": 81 }, { "epoch": 0.12349397590361445, "grad_norm": 0.162109375, "learning_rate": 9.588353413654619e-06, "loss": 1.1048, "step": 82 }, { "epoch": 0.125, "grad_norm": 0.1708984375, "learning_rate": 9.583333333333335e-06, "loss": 1.087, "step": 83 }, { "epoch": 0.12650602409638553, "grad_norm": 0.1796875, "learning_rate": 9.57831325301205e-06, "loss": 1.0802, "step": 84 }, { "epoch": 0.1280120481927711, "grad_norm": 0.3046875, "learning_rate": 9.573293172690764e-06, "loss": 1.0701, "step": 85 }, { "epoch": 0.12951807228915663, "grad_norm": 0.1474609375, "learning_rate": 9.568273092369479e-06, "loss": 1.0778, "step": 86 }, { "epoch": 0.13102409638554216, "grad_norm": 0.18359375, "learning_rate": 9.563253012048193e-06, "loss": 1.0608, "step": 87 }, { "epoch": 0.13253012048192772, "grad_norm": 0.1572265625, "learning_rate": 9.558232931726908e-06, "loss": 1.0601, "step": 88 }, { "epoch": 0.13403614457831325, "grad_norm": 0.1640625, "learning_rate": 9.553212851405622e-06, "loss": 1.0517, "step": 89 }, { "epoch": 0.1355421686746988, "grad_norm": 0.1591796875, "learning_rate": 9.548192771084339e-06, "loss": 1.0737, "step": 90 }, { "epoch": 0.13704819277108435, "grad_norm": 0.154296875, "learning_rate": 9.543172690763053e-06, "loss": 1.0623, "step": 91 }, { "epoch": 0.13855421686746988, "grad_norm": 0.15234375, "learning_rate": 9.538152610441768e-06, "loss": 1.0366, "step": 92 }, { "epoch": 0.14006024096385541, "grad_norm": 0.1533203125, "learning_rate": 9.533132530120482e-06, "loss": 1.0819, "step": 93 }, { "epoch": 0.14156626506024098, "grad_norm": 0.169921875, "learning_rate": 9.528112449799197e-06, "loss": 1.1081, "step": 94 }, { "epoch": 0.1430722891566265, "grad_norm": 0.150390625, "learning_rate": 9.523092369477911e-06, "loss": 1.0536, "step": 95 }, { "epoch": 0.14457831325301204, "grad_norm": 0.330078125, "learning_rate": 9.518072289156628e-06, "loss": 1.0098, "step": 96 }, { "epoch": 0.1460843373493976, "grad_norm": 0.177734375, "learning_rate": 9.513052208835342e-06, "loss": 1.033, "step": 97 }, { "epoch": 0.14759036144578314, "grad_norm": 0.15234375, "learning_rate": 9.508032128514058e-06, "loss": 1.0678, "step": 98 }, { "epoch": 0.14909638554216867, "grad_norm": 0.169921875, "learning_rate": 9.503012048192773e-06, "loss": 1.0007, "step": 99 }, { "epoch": 0.15060240963855423, "grad_norm": 0.1591796875, "learning_rate": 9.497991967871487e-06, "loss": 1.0943, "step": 100 }, { "epoch": 0.15210843373493976, "grad_norm": 0.1767578125, "learning_rate": 9.492971887550202e-06, "loss": 1.0497, "step": 101 }, { "epoch": 0.1536144578313253, "grad_norm": 0.1455078125, "learning_rate": 9.487951807228916e-06, "loss": 1.0462, "step": 102 }, { "epoch": 0.15512048192771086, "grad_norm": 0.15234375, "learning_rate": 9.482931726907631e-06, "loss": 1.0612, "step": 103 }, { "epoch": 0.1566265060240964, "grad_norm": 0.1669921875, "learning_rate": 9.477911646586347e-06, "loss": 1.073, "step": 104 }, { "epoch": 0.15813253012048192, "grad_norm": 0.197265625, "learning_rate": 9.472891566265062e-06, "loss": 1.0344, "step": 105 }, { "epoch": 0.15963855421686746, "grad_norm": 0.1630859375, "learning_rate": 9.467871485943776e-06, "loss": 1.0274, "step": 106 }, { "epoch": 0.16114457831325302, "grad_norm": 0.158203125, "learning_rate": 9.46285140562249e-06, "loss": 1.0117, "step": 107 }, { "epoch": 0.16265060240963855, "grad_norm": 0.1640625, "learning_rate": 9.457831325301205e-06, "loss": 1.0695, "step": 108 }, { "epoch": 0.16415662650602408, "grad_norm": 0.1552734375, "learning_rate": 9.45281124497992e-06, "loss": 1.0044, "step": 109 }, { "epoch": 0.16566265060240964, "grad_norm": 0.189453125, "learning_rate": 9.447791164658634e-06, "loss": 1.0439, "step": 110 }, { "epoch": 0.16716867469879518, "grad_norm": 0.181640625, "learning_rate": 9.44277108433735e-06, "loss": 1.0295, "step": 111 }, { "epoch": 0.1686746987951807, "grad_norm": 0.34765625, "learning_rate": 9.437751004016065e-06, "loss": 1.0423, "step": 112 }, { "epoch": 0.17018072289156627, "grad_norm": 0.1630859375, "learning_rate": 9.43273092369478e-06, "loss": 1.0315, "step": 113 }, { "epoch": 0.1716867469879518, "grad_norm": 0.1552734375, "learning_rate": 9.427710843373494e-06, "loss": 1.0492, "step": 114 }, { "epoch": 0.17319277108433734, "grad_norm": 0.1552734375, "learning_rate": 9.422690763052209e-06, "loss": 0.9827, "step": 115 }, { "epoch": 0.1746987951807229, "grad_norm": 0.1689453125, "learning_rate": 9.417670682730925e-06, "loss": 1.0382, "step": 116 }, { "epoch": 0.17620481927710843, "grad_norm": 0.1494140625, "learning_rate": 9.41265060240964e-06, "loss": 1.0273, "step": 117 }, { "epoch": 0.17771084337349397, "grad_norm": 0.2314453125, "learning_rate": 9.407630522088354e-06, "loss": 1.0731, "step": 118 }, { "epoch": 0.17921686746987953, "grad_norm": 0.189453125, "learning_rate": 9.40261044176707e-06, "loss": 1.0193, "step": 119 }, { "epoch": 0.18072289156626506, "grad_norm": 0.1640625, "learning_rate": 9.397590361445785e-06, "loss": 1.0304, "step": 120 }, { "epoch": 0.1822289156626506, "grad_norm": 0.16796875, "learning_rate": 9.3925702811245e-06, "loss": 1.0268, "step": 121 }, { "epoch": 0.18373493975903615, "grad_norm": 0.154296875, "learning_rate": 9.387550200803214e-06, "loss": 1.0029, "step": 122 }, { "epoch": 0.1852409638554217, "grad_norm": 0.1953125, "learning_rate": 9.382530120481928e-06, "loss": 0.9716, "step": 123 }, { "epoch": 0.18674698795180722, "grad_norm": 0.158203125, "learning_rate": 9.377510040160643e-06, "loss": 0.9913, "step": 124 }, { "epoch": 0.18825301204819278, "grad_norm": 0.234375, "learning_rate": 9.372489959839359e-06, "loss": 1.0044, "step": 125 }, { "epoch": 0.1897590361445783, "grad_norm": 0.173828125, "learning_rate": 9.367469879518074e-06, "loss": 1.08, "step": 126 }, { "epoch": 0.19126506024096385, "grad_norm": 0.1865234375, "learning_rate": 9.362449799196788e-06, "loss": 1.0029, "step": 127 }, { "epoch": 0.1927710843373494, "grad_norm": 0.1591796875, "learning_rate": 9.357429718875503e-06, "loss": 1.0228, "step": 128 }, { "epoch": 0.19427710843373494, "grad_norm": 0.1796875, "learning_rate": 9.352409638554217e-06, "loss": 1.0362, "step": 129 }, { "epoch": 0.19578313253012047, "grad_norm": 0.19921875, "learning_rate": 9.347389558232932e-06, "loss": 1.0274, "step": 130 }, { "epoch": 0.19728915662650603, "grad_norm": 0.16015625, "learning_rate": 9.342369477911646e-06, "loss": 1.0631, "step": 131 }, { "epoch": 0.19879518072289157, "grad_norm": 0.162109375, "learning_rate": 9.337349397590362e-06, "loss": 0.9849, "step": 132 }, { "epoch": 0.2003012048192771, "grad_norm": 0.1943359375, "learning_rate": 9.332329317269077e-06, "loss": 1.0331, "step": 133 }, { "epoch": 0.20180722891566266, "grad_norm": 0.1591796875, "learning_rate": 9.327309236947793e-06, "loss": 1.0052, "step": 134 }, { "epoch": 0.2033132530120482, "grad_norm": 0.271484375, "learning_rate": 9.322289156626508e-06, "loss": 1.0121, "step": 135 }, { "epoch": 0.20481927710843373, "grad_norm": 0.1630859375, "learning_rate": 9.317269076305222e-06, "loss": 1.0143, "step": 136 }, { "epoch": 0.2063253012048193, "grad_norm": 0.322265625, "learning_rate": 9.312248995983937e-06, "loss": 0.9929, "step": 137 }, { "epoch": 0.20783132530120482, "grad_norm": 0.2275390625, "learning_rate": 9.307228915662651e-06, "loss": 0.9521, "step": 138 }, { "epoch": 0.20933734939759036, "grad_norm": 0.1796875, "learning_rate": 9.302208835341366e-06, "loss": 1.0541, "step": 139 }, { "epoch": 0.21084337349397592, "grad_norm": 0.1650390625, "learning_rate": 9.297188755020082e-06, "loss": 0.9156, "step": 140 }, { "epoch": 0.21234939759036145, "grad_norm": 0.1669921875, "learning_rate": 9.292168674698796e-06, "loss": 1.0575, "step": 141 }, { "epoch": 0.21385542168674698, "grad_norm": 0.2216796875, "learning_rate": 9.287148594377511e-06, "loss": 1.0212, "step": 142 }, { "epoch": 0.21536144578313254, "grad_norm": 0.21484375, "learning_rate": 9.282128514056226e-06, "loss": 1.0113, "step": 143 }, { "epoch": 0.21686746987951808, "grad_norm": 0.26171875, "learning_rate": 9.27710843373494e-06, "loss": 0.9747, "step": 144 }, { "epoch": 0.2183734939759036, "grad_norm": 0.1650390625, "learning_rate": 9.272088353413655e-06, "loss": 0.9684, "step": 145 }, { "epoch": 0.21987951807228914, "grad_norm": 0.17578125, "learning_rate": 9.26706827309237e-06, "loss": 0.9896, "step": 146 }, { "epoch": 0.2213855421686747, "grad_norm": 0.1728515625, "learning_rate": 9.262048192771085e-06, "loss": 1.0117, "step": 147 }, { "epoch": 0.22289156626506024, "grad_norm": 0.1796875, "learning_rate": 9.2570281124498e-06, "loss": 0.9748, "step": 148 }, { "epoch": 0.22439759036144577, "grad_norm": 0.16796875, "learning_rate": 9.252008032128514e-06, "loss": 1.0027, "step": 149 }, { "epoch": 0.22590361445783133, "grad_norm": 0.185546875, "learning_rate": 9.246987951807229e-06, "loss": 1.0147, "step": 150 }, { "epoch": 0.22740963855421686, "grad_norm": 0.345703125, "learning_rate": 9.241967871485943e-06, "loss": 0.983, "step": 151 }, { "epoch": 0.2289156626506024, "grad_norm": 0.1748046875, "learning_rate": 9.23694779116466e-06, "loss": 0.9984, "step": 152 }, { "epoch": 0.23042168674698796, "grad_norm": 0.2412109375, "learning_rate": 9.231927710843374e-06, "loss": 1.0294, "step": 153 }, { "epoch": 0.2319277108433735, "grad_norm": 0.171875, "learning_rate": 9.226907630522089e-06, "loss": 0.9763, "step": 154 }, { "epoch": 0.23343373493975902, "grad_norm": 0.1728515625, "learning_rate": 9.221887550200805e-06, "loss": 1.0381, "step": 155 }, { "epoch": 0.23493975903614459, "grad_norm": 0.1796875, "learning_rate": 9.21686746987952e-06, "loss": 0.9951, "step": 156 }, { "epoch": 0.23644578313253012, "grad_norm": 0.1728515625, "learning_rate": 9.211847389558234e-06, "loss": 1.019, "step": 157 }, { "epoch": 0.23795180722891565, "grad_norm": 0.1748046875, "learning_rate": 9.206827309236949e-06, "loss": 0.9672, "step": 158 }, { "epoch": 0.2394578313253012, "grad_norm": 0.2109375, "learning_rate": 9.201807228915663e-06, "loss": 0.9932, "step": 159 }, { "epoch": 0.24096385542168675, "grad_norm": 0.1953125, "learning_rate": 9.196787148594378e-06, "loss": 0.9786, "step": 160 }, { "epoch": 0.24246987951807228, "grad_norm": 0.1748046875, "learning_rate": 9.191767068273094e-06, "loss": 0.9539, "step": 161 }, { "epoch": 0.24397590361445784, "grad_norm": 0.1708984375, "learning_rate": 9.186746987951808e-06, "loss": 1.0208, "step": 162 }, { "epoch": 0.24548192771084337, "grad_norm": 0.1728515625, "learning_rate": 9.181726907630523e-06, "loss": 1.0275, "step": 163 }, { "epoch": 0.2469879518072289, "grad_norm": 0.373046875, "learning_rate": 9.176706827309237e-06, "loss": 0.9987, "step": 164 }, { "epoch": 0.24849397590361447, "grad_norm": 0.1787109375, "learning_rate": 9.171686746987952e-06, "loss": 1.0042, "step": 165 }, { "epoch": 0.25, "grad_norm": 0.1748046875, "learning_rate": 9.166666666666666e-06, "loss": 0.967, "step": 166 }, { "epoch": 0.25150602409638556, "grad_norm": 0.1650390625, "learning_rate": 9.161646586345383e-06, "loss": 0.9648, "step": 167 }, { "epoch": 0.25301204819277107, "grad_norm": 0.1806640625, "learning_rate": 9.156626506024097e-06, "loss": 0.9491, "step": 168 }, { "epoch": 0.2545180722891566, "grad_norm": 0.291015625, "learning_rate": 9.151606425702812e-06, "loss": 0.9949, "step": 169 }, { "epoch": 0.2560240963855422, "grad_norm": 0.1845703125, "learning_rate": 9.146586345381528e-06, "loss": 0.9959, "step": 170 }, { "epoch": 0.2575301204819277, "grad_norm": 0.1728515625, "learning_rate": 9.141566265060242e-06, "loss": 1.0221, "step": 171 }, { "epoch": 0.25903614457831325, "grad_norm": 0.1787109375, "learning_rate": 9.136546184738957e-06, "loss": 0.99, "step": 172 }, { "epoch": 0.2605421686746988, "grad_norm": 0.1708984375, "learning_rate": 9.131526104417672e-06, "loss": 1.02, "step": 173 }, { "epoch": 0.2620481927710843, "grad_norm": 0.1669921875, "learning_rate": 9.126506024096386e-06, "loss": 0.9701, "step": 174 }, { "epoch": 0.2635542168674699, "grad_norm": 0.17578125, "learning_rate": 9.1214859437751e-06, "loss": 1.0061, "step": 175 }, { "epoch": 0.26506024096385544, "grad_norm": 0.166015625, "learning_rate": 9.116465863453817e-06, "loss": 0.9882, "step": 176 }, { "epoch": 0.26656626506024095, "grad_norm": 0.1982421875, "learning_rate": 9.111445783132531e-06, "loss": 1.0182, "step": 177 }, { "epoch": 0.2680722891566265, "grad_norm": 0.212890625, "learning_rate": 9.106425702811246e-06, "loss": 0.9606, "step": 178 }, { "epoch": 0.26957831325301207, "grad_norm": 0.20703125, "learning_rate": 9.10140562248996e-06, "loss": 0.9901, "step": 179 }, { "epoch": 0.2710843373493976, "grad_norm": 0.173828125, "learning_rate": 9.096385542168675e-06, "loss": 0.9395, "step": 180 }, { "epoch": 0.27259036144578314, "grad_norm": 0.21875, "learning_rate": 9.09136546184739e-06, "loss": 0.9876, "step": 181 }, { "epoch": 0.2740963855421687, "grad_norm": 0.205078125, "learning_rate": 9.086345381526106e-06, "loss": 1.0252, "step": 182 }, { "epoch": 0.2756024096385542, "grad_norm": 0.228515625, "learning_rate": 9.08132530120482e-06, "loss": 0.9905, "step": 183 }, { "epoch": 0.27710843373493976, "grad_norm": 0.1865234375, "learning_rate": 9.076305220883535e-06, "loss": 0.9775, "step": 184 }, { "epoch": 0.2786144578313253, "grad_norm": 0.169921875, "learning_rate": 9.07128514056225e-06, "loss": 1.0046, "step": 185 }, { "epoch": 0.28012048192771083, "grad_norm": 0.173828125, "learning_rate": 9.066265060240964e-06, "loss": 0.9419, "step": 186 }, { "epoch": 0.2816265060240964, "grad_norm": 0.2001953125, "learning_rate": 9.06124497991968e-06, "loss": 0.9767, "step": 187 }, { "epoch": 0.28313253012048195, "grad_norm": 0.228515625, "learning_rate": 9.056224899598394e-06, "loss": 0.9601, "step": 188 }, { "epoch": 0.28463855421686746, "grad_norm": 0.1865234375, "learning_rate": 9.051204819277109e-06, "loss": 0.9777, "step": 189 }, { "epoch": 0.286144578313253, "grad_norm": 0.2490234375, "learning_rate": 9.046184738955825e-06, "loss": 0.9501, "step": 190 }, { "epoch": 0.2876506024096386, "grad_norm": 0.177734375, "learning_rate": 9.04116465863454e-06, "loss": 0.9844, "step": 191 }, { "epoch": 0.2891566265060241, "grad_norm": 0.1708984375, "learning_rate": 9.036144578313254e-06, "loss": 0.9687, "step": 192 }, { "epoch": 0.29066265060240964, "grad_norm": 0.189453125, "learning_rate": 9.031124497991969e-06, "loss": 0.9974, "step": 193 }, { "epoch": 0.2921686746987952, "grad_norm": 0.197265625, "learning_rate": 9.026104417670683e-06, "loss": 0.9668, "step": 194 }, { "epoch": 0.2936746987951807, "grad_norm": 0.1923828125, "learning_rate": 9.021084337349398e-06, "loss": 0.9499, "step": 195 }, { "epoch": 0.29518072289156627, "grad_norm": 0.2158203125, "learning_rate": 9.016064257028112e-06, "loss": 0.9477, "step": 196 }, { "epoch": 0.29668674698795183, "grad_norm": 0.17578125, "learning_rate": 9.011044176706829e-06, "loss": 0.9662, "step": 197 }, { "epoch": 0.29819277108433734, "grad_norm": 0.26171875, "learning_rate": 9.006024096385543e-06, "loss": 0.9314, "step": 198 }, { "epoch": 0.2996987951807229, "grad_norm": 0.1796875, "learning_rate": 9.001004016064258e-06, "loss": 0.9513, "step": 199 }, { "epoch": 0.30120481927710846, "grad_norm": 0.18359375, "learning_rate": 8.995983935742972e-06, "loss": 0.9366, "step": 200 }, { "epoch": 0.30271084337349397, "grad_norm": 0.1689453125, "learning_rate": 8.990963855421687e-06, "loss": 0.9668, "step": 201 }, { "epoch": 0.3042168674698795, "grad_norm": 0.2041015625, "learning_rate": 8.985943775100401e-06, "loss": 0.9829, "step": 202 }, { "epoch": 0.3057228915662651, "grad_norm": 0.171875, "learning_rate": 8.980923694779117e-06, "loss": 0.9489, "step": 203 }, { "epoch": 0.3072289156626506, "grad_norm": 0.205078125, "learning_rate": 8.975903614457832e-06, "loss": 0.9974, "step": 204 }, { "epoch": 0.30873493975903615, "grad_norm": 0.177734375, "learning_rate": 8.970883534136548e-06, "loss": 0.9668, "step": 205 }, { "epoch": 0.3102409638554217, "grad_norm": 0.1875, "learning_rate": 8.965863453815263e-06, "loss": 0.9722, "step": 206 }, { "epoch": 0.3117469879518072, "grad_norm": 0.197265625, "learning_rate": 8.960843373493977e-06, "loss": 0.9763, "step": 207 }, { "epoch": 0.3132530120481928, "grad_norm": 0.2001953125, "learning_rate": 8.955823293172692e-06, "loss": 0.961, "step": 208 }, { "epoch": 0.3147590361445783, "grad_norm": 0.62890625, "learning_rate": 8.950803212851406e-06, "loss": 1.0608, "step": 209 }, { "epoch": 0.31626506024096385, "grad_norm": 0.1806640625, "learning_rate": 8.94578313253012e-06, "loss": 0.9932, "step": 210 }, { "epoch": 0.3177710843373494, "grad_norm": 0.23046875, "learning_rate": 8.940763052208837e-06, "loss": 0.9298, "step": 211 }, { "epoch": 0.3192771084337349, "grad_norm": 0.1748046875, "learning_rate": 8.935742971887552e-06, "loss": 0.9381, "step": 212 }, { "epoch": 0.3207831325301205, "grad_norm": 0.26953125, "learning_rate": 8.930722891566266e-06, "loss": 0.9291, "step": 213 }, { "epoch": 0.32228915662650603, "grad_norm": 0.1767578125, "learning_rate": 8.92570281124498e-06, "loss": 0.9046, "step": 214 }, { "epoch": 0.32379518072289154, "grad_norm": 0.1826171875, "learning_rate": 8.920682730923695e-06, "loss": 0.9615, "step": 215 }, { "epoch": 0.3253012048192771, "grad_norm": 0.1806640625, "learning_rate": 8.91566265060241e-06, "loss": 0.9442, "step": 216 }, { "epoch": 0.32680722891566266, "grad_norm": 0.1884765625, "learning_rate": 8.910642570281124e-06, "loss": 0.9378, "step": 217 }, { "epoch": 0.32831325301204817, "grad_norm": 0.17578125, "learning_rate": 8.90562248995984e-06, "loss": 0.9782, "step": 218 }, { "epoch": 0.32981927710843373, "grad_norm": 0.1884765625, "learning_rate": 8.900602409638555e-06, "loss": 0.9192, "step": 219 }, { "epoch": 0.3313253012048193, "grad_norm": 0.232421875, "learning_rate": 8.89558232931727e-06, "loss": 0.9268, "step": 220 }, { "epoch": 0.3328313253012048, "grad_norm": 0.1923828125, "learning_rate": 8.890562248995984e-06, "loss": 0.9588, "step": 221 }, { "epoch": 0.33433734939759036, "grad_norm": 0.21484375, "learning_rate": 8.885542168674699e-06, "loss": 0.9849, "step": 222 }, { "epoch": 0.3358433734939759, "grad_norm": 0.1767578125, "learning_rate": 8.880522088353415e-06, "loss": 0.9692, "step": 223 }, { "epoch": 0.3373493975903614, "grad_norm": 0.2734375, "learning_rate": 8.87550200803213e-06, "loss": 0.9682, "step": 224 }, { "epoch": 0.338855421686747, "grad_norm": 0.173828125, "learning_rate": 8.870481927710844e-06, "loss": 0.9498, "step": 225 }, { "epoch": 0.34036144578313254, "grad_norm": 0.1845703125, "learning_rate": 8.86546184738956e-06, "loss": 0.9406, "step": 226 }, { "epoch": 0.34186746987951805, "grad_norm": 0.1904296875, "learning_rate": 8.860441767068275e-06, "loss": 0.9397, "step": 227 }, { "epoch": 0.3433734939759036, "grad_norm": 0.2080078125, "learning_rate": 8.855421686746989e-06, "loss": 0.9375, "step": 228 }, { "epoch": 0.34487951807228917, "grad_norm": 0.298828125, "learning_rate": 8.850401606425704e-06, "loss": 0.9437, "step": 229 }, { "epoch": 0.3463855421686747, "grad_norm": 0.2216796875, "learning_rate": 8.845381526104418e-06, "loss": 0.9593, "step": 230 }, { "epoch": 0.34789156626506024, "grad_norm": 0.302734375, "learning_rate": 8.840361445783133e-06, "loss": 0.928, "step": 231 }, { "epoch": 0.3493975903614458, "grad_norm": 0.19140625, "learning_rate": 8.835341365461849e-06, "loss": 0.9218, "step": 232 }, { "epoch": 0.3509036144578313, "grad_norm": 0.1875, "learning_rate": 8.830321285140563e-06, "loss": 0.947, "step": 233 }, { "epoch": 0.35240963855421686, "grad_norm": 0.185546875, "learning_rate": 8.825301204819278e-06, "loss": 0.9349, "step": 234 }, { "epoch": 0.3539156626506024, "grad_norm": 0.2255859375, "learning_rate": 8.820281124497992e-06, "loss": 0.9166, "step": 235 }, { "epoch": 0.35542168674698793, "grad_norm": 0.578125, "learning_rate": 8.815261044176707e-06, "loss": 0.9781, "step": 236 }, { "epoch": 0.3569277108433735, "grad_norm": 0.1826171875, "learning_rate": 8.810240963855422e-06, "loss": 0.9195, "step": 237 }, { "epoch": 0.35843373493975905, "grad_norm": 0.212890625, "learning_rate": 8.805220883534136e-06, "loss": 0.9726, "step": 238 }, { "epoch": 0.35993975903614456, "grad_norm": 0.1875, "learning_rate": 8.800200803212852e-06, "loss": 0.9128, "step": 239 }, { "epoch": 0.3614457831325301, "grad_norm": 0.185546875, "learning_rate": 8.795180722891567e-06, "loss": 0.9074, "step": 240 }, { "epoch": 0.3629518072289157, "grad_norm": 0.2021484375, "learning_rate": 8.790160642570283e-06, "loss": 0.9618, "step": 241 }, { "epoch": 0.3644578313253012, "grad_norm": 0.19921875, "learning_rate": 8.785140562248998e-06, "loss": 0.9005, "step": 242 }, { "epoch": 0.36596385542168675, "grad_norm": 0.2001953125, "learning_rate": 8.780120481927712e-06, "loss": 0.9734, "step": 243 }, { "epoch": 0.3674698795180723, "grad_norm": 0.2578125, "learning_rate": 8.775100401606427e-06, "loss": 0.9527, "step": 244 }, { "epoch": 0.3689759036144578, "grad_norm": 0.30859375, "learning_rate": 8.770080321285141e-06, "loss": 0.9738, "step": 245 }, { "epoch": 0.3704819277108434, "grad_norm": 0.189453125, "learning_rate": 8.765060240963856e-06, "loss": 0.9303, "step": 246 }, { "epoch": 0.37198795180722893, "grad_norm": 0.201171875, "learning_rate": 8.760040160642572e-06, "loss": 0.9516, "step": 247 }, { "epoch": 0.37349397590361444, "grad_norm": 0.19140625, "learning_rate": 8.755020080321286e-06, "loss": 0.9519, "step": 248 }, { "epoch": 0.375, "grad_norm": 0.1953125, "learning_rate": 8.750000000000001e-06, "loss": 0.9654, "step": 249 }, { "epoch": 0.37650602409638556, "grad_norm": 0.1845703125, "learning_rate": 8.744979919678715e-06, "loss": 0.9106, "step": 250 }, { "epoch": 0.37801204819277107, "grad_norm": 0.19140625, "learning_rate": 8.73995983935743e-06, "loss": 0.9259, "step": 251 }, { "epoch": 0.3795180722891566, "grad_norm": 0.185546875, "learning_rate": 8.734939759036145e-06, "loss": 1.0022, "step": 252 }, { "epoch": 0.3810240963855422, "grad_norm": 0.1767578125, "learning_rate": 8.72991967871486e-06, "loss": 0.9379, "step": 253 }, { "epoch": 0.3825301204819277, "grad_norm": 0.1796875, "learning_rate": 8.724899598393575e-06, "loss": 0.9185, "step": 254 }, { "epoch": 0.38403614457831325, "grad_norm": 0.18359375, "learning_rate": 8.71987951807229e-06, "loss": 0.9193, "step": 255 }, { "epoch": 0.3855421686746988, "grad_norm": 0.185546875, "learning_rate": 8.714859437751004e-06, "loss": 0.9617, "step": 256 }, { "epoch": 0.3870481927710843, "grad_norm": 0.1865234375, "learning_rate": 8.709839357429719e-06, "loss": 0.9203, "step": 257 }, { "epoch": 0.3885542168674699, "grad_norm": 0.181640625, "learning_rate": 8.704819277108435e-06, "loss": 0.9802, "step": 258 }, { "epoch": 0.39006024096385544, "grad_norm": 0.197265625, "learning_rate": 8.69979919678715e-06, "loss": 1.006, "step": 259 }, { "epoch": 0.39156626506024095, "grad_norm": 0.189453125, "learning_rate": 8.694779116465864e-06, "loss": 0.9566, "step": 260 }, { "epoch": 0.3930722891566265, "grad_norm": 0.1962890625, "learning_rate": 8.689759036144579e-06, "loss": 0.9437, "step": 261 }, { "epoch": 0.39457831325301207, "grad_norm": 0.39453125, "learning_rate": 8.684738955823295e-06, "loss": 0.914, "step": 262 }, { "epoch": 0.3960843373493976, "grad_norm": 0.21484375, "learning_rate": 8.67971887550201e-06, "loss": 0.895, "step": 263 }, { "epoch": 0.39759036144578314, "grad_norm": 0.1728515625, "learning_rate": 8.674698795180724e-06, "loss": 0.9272, "step": 264 }, { "epoch": 0.3990963855421687, "grad_norm": 0.224609375, "learning_rate": 8.669678714859438e-06, "loss": 0.9916, "step": 265 }, { "epoch": 0.4006024096385542, "grad_norm": 0.1806640625, "learning_rate": 8.664658634538153e-06, "loss": 0.9136, "step": 266 }, { "epoch": 0.40210843373493976, "grad_norm": 0.2001953125, "learning_rate": 8.659638554216867e-06, "loss": 0.9654, "step": 267 }, { "epoch": 0.4036144578313253, "grad_norm": 0.21484375, "learning_rate": 8.654618473895584e-06, "loss": 0.9571, "step": 268 }, { "epoch": 0.40512048192771083, "grad_norm": 0.1728515625, "learning_rate": 8.649598393574298e-06, "loss": 0.9241, "step": 269 }, { "epoch": 0.4066265060240964, "grad_norm": 0.181640625, "learning_rate": 8.644578313253013e-06, "loss": 0.938, "step": 270 }, { "epoch": 0.40813253012048195, "grad_norm": 0.201171875, "learning_rate": 8.639558232931727e-06, "loss": 0.9553, "step": 271 }, { "epoch": 0.40963855421686746, "grad_norm": 0.2236328125, "learning_rate": 8.634538152610442e-06, "loss": 0.9195, "step": 272 }, { "epoch": 0.411144578313253, "grad_norm": 0.2412109375, "learning_rate": 8.629518072289156e-06, "loss": 0.9441, "step": 273 }, { "epoch": 0.4126506024096386, "grad_norm": 0.1923828125, "learning_rate": 8.624497991967873e-06, "loss": 0.915, "step": 274 }, { "epoch": 0.4141566265060241, "grad_norm": 0.1923828125, "learning_rate": 8.619477911646587e-06, "loss": 0.9853, "step": 275 }, { "epoch": 0.41566265060240964, "grad_norm": 0.19921875, "learning_rate": 8.614457831325302e-06, "loss": 0.9247, "step": 276 }, { "epoch": 0.4171686746987952, "grad_norm": 0.185546875, "learning_rate": 8.609437751004018e-06, "loss": 0.9398, "step": 277 }, { "epoch": 0.4186746987951807, "grad_norm": 0.1982421875, "learning_rate": 8.604417670682732e-06, "loss": 0.9158, "step": 278 }, { "epoch": 0.42018072289156627, "grad_norm": 0.2041015625, "learning_rate": 8.599397590361447e-06, "loss": 0.9032, "step": 279 }, { "epoch": 0.42168674698795183, "grad_norm": 0.1953125, "learning_rate": 8.594377510040161e-06, "loss": 0.9139, "step": 280 }, { "epoch": 0.42319277108433734, "grad_norm": 0.19921875, "learning_rate": 8.589357429718876e-06, "loss": 0.9486, "step": 281 }, { "epoch": 0.4246987951807229, "grad_norm": 0.181640625, "learning_rate": 8.58433734939759e-06, "loss": 0.894, "step": 282 }, { "epoch": 0.42620481927710846, "grad_norm": 0.1923828125, "learning_rate": 8.579317269076307e-06, "loss": 0.9804, "step": 283 }, { "epoch": 0.42771084337349397, "grad_norm": 0.1962890625, "learning_rate": 8.574297188755021e-06, "loss": 0.9538, "step": 284 }, { "epoch": 0.4292168674698795, "grad_norm": 0.1875, "learning_rate": 8.569277108433736e-06, "loss": 0.9171, "step": 285 }, { "epoch": 0.4307228915662651, "grad_norm": 0.19921875, "learning_rate": 8.56425702811245e-06, "loss": 0.9324, "step": 286 }, { "epoch": 0.4322289156626506, "grad_norm": 0.1962890625, "learning_rate": 8.559236947791165e-06, "loss": 0.9166, "step": 287 }, { "epoch": 0.43373493975903615, "grad_norm": 0.1943359375, "learning_rate": 8.55421686746988e-06, "loss": 0.9813, "step": 288 }, { "epoch": 0.4352409638554217, "grad_norm": 0.234375, "learning_rate": 8.549196787148596e-06, "loss": 0.953, "step": 289 }, { "epoch": 0.4367469879518072, "grad_norm": 0.419921875, "learning_rate": 8.54417670682731e-06, "loss": 0.9414, "step": 290 }, { "epoch": 0.4382530120481928, "grad_norm": 0.2001953125, "learning_rate": 8.539156626506025e-06, "loss": 0.9416, "step": 291 }, { "epoch": 0.4397590361445783, "grad_norm": 0.244140625, "learning_rate": 8.534136546184739e-06, "loss": 0.9186, "step": 292 }, { "epoch": 0.44126506024096385, "grad_norm": 0.1845703125, "learning_rate": 8.529116465863454e-06, "loss": 0.906, "step": 293 }, { "epoch": 0.4427710843373494, "grad_norm": 0.1923828125, "learning_rate": 8.52409638554217e-06, "loss": 0.9469, "step": 294 }, { "epoch": 0.4442771084337349, "grad_norm": 0.1943359375, "learning_rate": 8.519076305220884e-06, "loss": 0.8992, "step": 295 }, { "epoch": 0.4457831325301205, "grad_norm": 0.25390625, "learning_rate": 8.514056224899599e-06, "loss": 0.9386, "step": 296 }, { "epoch": 0.44728915662650603, "grad_norm": 0.208984375, "learning_rate": 8.509036144578313e-06, "loss": 0.9254, "step": 297 }, { "epoch": 0.44879518072289154, "grad_norm": 0.1943359375, "learning_rate": 8.50401606425703e-06, "loss": 0.9328, "step": 298 }, { "epoch": 0.4503012048192771, "grad_norm": 0.1982421875, "learning_rate": 8.498995983935744e-06, "loss": 0.9169, "step": 299 }, { "epoch": 0.45180722891566266, "grad_norm": 0.201171875, "learning_rate": 8.493975903614459e-06, "loss": 0.8949, "step": 300 }, { "epoch": 0.45331325301204817, "grad_norm": 0.189453125, "learning_rate": 8.488955823293173e-06, "loss": 0.8969, "step": 301 }, { "epoch": 0.45481927710843373, "grad_norm": 0.2021484375, "learning_rate": 8.483935742971888e-06, "loss": 0.8944, "step": 302 }, { "epoch": 0.4563253012048193, "grad_norm": 0.23046875, "learning_rate": 8.478915662650602e-06, "loss": 0.9411, "step": 303 }, { "epoch": 0.4578313253012048, "grad_norm": 0.2158203125, "learning_rate": 8.473895582329319e-06, "loss": 0.9283, "step": 304 }, { "epoch": 0.45933734939759036, "grad_norm": 0.177734375, "learning_rate": 8.468875502008033e-06, "loss": 0.9662, "step": 305 }, { "epoch": 0.4608433734939759, "grad_norm": 0.19140625, "learning_rate": 8.463855421686748e-06, "loss": 0.9404, "step": 306 }, { "epoch": 0.4623493975903614, "grad_norm": 0.2021484375, "learning_rate": 8.458835341365462e-06, "loss": 0.9456, "step": 307 }, { "epoch": 0.463855421686747, "grad_norm": 0.2041015625, "learning_rate": 8.453815261044177e-06, "loss": 0.8976, "step": 308 }, { "epoch": 0.46536144578313254, "grad_norm": 0.2177734375, "learning_rate": 8.448795180722891e-06, "loss": 0.8823, "step": 309 }, { "epoch": 0.46686746987951805, "grad_norm": 0.1904296875, "learning_rate": 8.443775100401607e-06, "loss": 0.8968, "step": 310 }, { "epoch": 0.4683734939759036, "grad_norm": 0.2578125, "learning_rate": 8.438755020080322e-06, "loss": 0.8937, "step": 311 }, { "epoch": 0.46987951807228917, "grad_norm": 0.1865234375, "learning_rate": 8.433734939759038e-06, "loss": 0.9068, "step": 312 }, { "epoch": 0.4713855421686747, "grad_norm": 0.205078125, "learning_rate": 8.428714859437753e-06, "loss": 0.9654, "step": 313 }, { "epoch": 0.47289156626506024, "grad_norm": 0.1982421875, "learning_rate": 8.423694779116467e-06, "loss": 0.9205, "step": 314 }, { "epoch": 0.4743975903614458, "grad_norm": 0.197265625, "learning_rate": 8.418674698795182e-06, "loss": 0.9187, "step": 315 }, { "epoch": 0.4759036144578313, "grad_norm": 0.1884765625, "learning_rate": 8.413654618473896e-06, "loss": 0.9263, "step": 316 }, { "epoch": 0.47740963855421686, "grad_norm": 0.19921875, "learning_rate": 8.40863453815261e-06, "loss": 0.9237, "step": 317 }, { "epoch": 0.4789156626506024, "grad_norm": 0.28125, "learning_rate": 8.403614457831325e-06, "loss": 0.9333, "step": 318 }, { "epoch": 0.48042168674698793, "grad_norm": 0.197265625, "learning_rate": 8.398594377510041e-06, "loss": 0.8847, "step": 319 }, { "epoch": 0.4819277108433735, "grad_norm": 0.271484375, "learning_rate": 8.393574297188756e-06, "loss": 0.8904, "step": 320 }, { "epoch": 0.48343373493975905, "grad_norm": 0.212890625, "learning_rate": 8.38855421686747e-06, "loss": 0.9415, "step": 321 }, { "epoch": 0.48493975903614456, "grad_norm": 0.2578125, "learning_rate": 8.383534136546185e-06, "loss": 0.894, "step": 322 }, { "epoch": 0.4864457831325301, "grad_norm": 0.23828125, "learning_rate": 8.3785140562249e-06, "loss": 0.9397, "step": 323 }, { "epoch": 0.4879518072289157, "grad_norm": 0.201171875, "learning_rate": 8.373493975903614e-06, "loss": 0.9131, "step": 324 }, { "epoch": 0.4894578313253012, "grad_norm": 0.1923828125, "learning_rate": 8.36847389558233e-06, "loss": 0.9397, "step": 325 }, { "epoch": 0.49096385542168675, "grad_norm": 0.1943359375, "learning_rate": 8.363453815261045e-06, "loss": 0.8796, "step": 326 }, { "epoch": 0.4924698795180723, "grad_norm": 0.2060546875, "learning_rate": 8.35843373493976e-06, "loss": 0.9035, "step": 327 }, { "epoch": 0.4939759036144578, "grad_norm": 0.259765625, "learning_rate": 8.353413654618474e-06, "loss": 0.907, "step": 328 }, { "epoch": 0.4954819277108434, "grad_norm": 0.2001953125, "learning_rate": 8.348393574297188e-06, "loss": 0.9461, "step": 329 }, { "epoch": 0.49698795180722893, "grad_norm": 0.1962890625, "learning_rate": 8.343373493975905e-06, "loss": 0.9428, "step": 330 }, { "epoch": 0.49849397590361444, "grad_norm": 0.1904296875, "learning_rate": 8.33835341365462e-06, "loss": 0.8631, "step": 331 }, { "epoch": 0.5, "grad_norm": 0.201171875, "learning_rate": 8.333333333333334e-06, "loss": 0.8787, "step": 332 }, { "epoch": 0.5015060240963856, "grad_norm": 0.2099609375, "learning_rate": 8.32831325301205e-06, "loss": 0.8986, "step": 333 }, { "epoch": 0.5030120481927711, "grad_norm": 0.2109375, "learning_rate": 8.323293172690764e-06, "loss": 0.8913, "step": 334 }, { "epoch": 0.5045180722891566, "grad_norm": 0.1982421875, "learning_rate": 8.318273092369479e-06, "loss": 0.9031, "step": 335 }, { "epoch": 0.5060240963855421, "grad_norm": 0.1796875, "learning_rate": 8.313253012048194e-06, "loss": 0.8727, "step": 336 }, { "epoch": 0.5075301204819277, "grad_norm": 0.232421875, "learning_rate": 8.308232931726908e-06, "loss": 0.9072, "step": 337 }, { "epoch": 0.5090361445783133, "grad_norm": 0.205078125, "learning_rate": 8.303212851405623e-06, "loss": 0.9562, "step": 338 }, { "epoch": 0.5105421686746988, "grad_norm": 0.25390625, "learning_rate": 8.298192771084339e-06, "loss": 0.9241, "step": 339 }, { "epoch": 0.5120481927710844, "grad_norm": 0.1962890625, "learning_rate": 8.293172690763053e-06, "loss": 0.9156, "step": 340 }, { "epoch": 0.5135542168674698, "grad_norm": 0.2119140625, "learning_rate": 8.288152610441768e-06, "loss": 0.8993, "step": 341 }, { "epoch": 0.5150602409638554, "grad_norm": 0.263671875, "learning_rate": 8.283132530120482e-06, "loss": 0.9104, "step": 342 }, { "epoch": 0.516566265060241, "grad_norm": 0.2021484375, "learning_rate": 8.278112449799197e-06, "loss": 0.8753, "step": 343 }, { "epoch": 0.5180722891566265, "grad_norm": 0.189453125, "learning_rate": 8.273092369477911e-06, "loss": 0.8722, "step": 344 }, { "epoch": 0.5195783132530121, "grad_norm": 0.1982421875, "learning_rate": 8.268072289156626e-06, "loss": 0.9064, "step": 345 }, { "epoch": 0.5210843373493976, "grad_norm": 0.1962890625, "learning_rate": 8.263052208835342e-06, "loss": 0.94, "step": 346 }, { "epoch": 0.5225903614457831, "grad_norm": 0.2333984375, "learning_rate": 8.258032128514057e-06, "loss": 0.948, "step": 347 }, { "epoch": 0.5240963855421686, "grad_norm": 0.2080078125, "learning_rate": 8.253012048192773e-06, "loss": 0.8985, "step": 348 }, { "epoch": 0.5256024096385542, "grad_norm": 0.294921875, "learning_rate": 8.247991967871487e-06, "loss": 0.912, "step": 349 }, { "epoch": 0.5271084337349398, "grad_norm": 0.19140625, "learning_rate": 8.242971887550202e-06, "loss": 0.9315, "step": 350 }, { "epoch": 0.5286144578313253, "grad_norm": 0.1904296875, "learning_rate": 8.237951807228917e-06, "loss": 0.8591, "step": 351 }, { "epoch": 0.5301204819277109, "grad_norm": 0.265625, "learning_rate": 8.232931726907631e-06, "loss": 0.9568, "step": 352 }, { "epoch": 0.5316265060240963, "grad_norm": 0.2412109375, "learning_rate": 8.227911646586346e-06, "loss": 0.9194, "step": 353 }, { "epoch": 0.5331325301204819, "grad_norm": 0.1962890625, "learning_rate": 8.222891566265062e-06, "loss": 0.9177, "step": 354 }, { "epoch": 0.5346385542168675, "grad_norm": 0.21875, "learning_rate": 8.217871485943776e-06, "loss": 0.9144, "step": 355 }, { "epoch": 0.536144578313253, "grad_norm": 0.1953125, "learning_rate": 8.21285140562249e-06, "loss": 0.9084, "step": 356 }, { "epoch": 0.5376506024096386, "grad_norm": 0.20703125, "learning_rate": 8.207831325301205e-06, "loss": 0.9014, "step": 357 }, { "epoch": 0.5391566265060241, "grad_norm": 0.2275390625, "learning_rate": 8.20281124497992e-06, "loss": 0.9062, "step": 358 }, { "epoch": 0.5406626506024096, "grad_norm": 0.1953125, "learning_rate": 8.197791164658634e-06, "loss": 0.9346, "step": 359 }, { "epoch": 0.5421686746987951, "grad_norm": 0.205078125, "learning_rate": 8.19277108433735e-06, "loss": 0.906, "step": 360 }, { "epoch": 0.5436746987951807, "grad_norm": 0.22265625, "learning_rate": 8.187751004016065e-06, "loss": 0.9104, "step": 361 }, { "epoch": 0.5451807228915663, "grad_norm": 0.20703125, "learning_rate": 8.18273092369478e-06, "loss": 0.899, "step": 362 }, { "epoch": 0.5466867469879518, "grad_norm": 0.19140625, "learning_rate": 8.177710843373494e-06, "loss": 0.8641, "step": 363 }, { "epoch": 0.5481927710843374, "grad_norm": 0.1953125, "learning_rate": 8.172690763052209e-06, "loss": 0.893, "step": 364 }, { "epoch": 0.5496987951807228, "grad_norm": 0.1962890625, "learning_rate": 8.167670682730925e-06, "loss": 0.9062, "step": 365 }, { "epoch": 0.5512048192771084, "grad_norm": 0.19140625, "learning_rate": 8.16265060240964e-06, "loss": 0.8972, "step": 366 }, { "epoch": 0.552710843373494, "grad_norm": 0.2109375, "learning_rate": 8.157630522088354e-06, "loss": 0.9061, "step": 367 }, { "epoch": 0.5542168674698795, "grad_norm": 0.205078125, "learning_rate": 8.152610441767069e-06, "loss": 0.9516, "step": 368 }, { "epoch": 0.5557228915662651, "grad_norm": 0.220703125, "learning_rate": 8.147590361445785e-06, "loss": 0.9578, "step": 369 }, { "epoch": 0.5572289156626506, "grad_norm": 0.203125, "learning_rate": 8.1425702811245e-06, "loss": 0.9347, "step": 370 }, { "epoch": 0.5587349397590361, "grad_norm": 0.2080078125, "learning_rate": 8.137550200803214e-06, "loss": 0.9764, "step": 371 }, { "epoch": 0.5602409638554217, "grad_norm": 0.1962890625, "learning_rate": 8.132530120481928e-06, "loss": 0.9088, "step": 372 }, { "epoch": 0.5617469879518072, "grad_norm": 0.30859375, "learning_rate": 8.127510040160643e-06, "loss": 0.8624, "step": 373 }, { "epoch": 0.5632530120481928, "grad_norm": 0.19921875, "learning_rate": 8.122489959839357e-06, "loss": 0.9244, "step": 374 }, { "epoch": 0.5647590361445783, "grad_norm": 0.2021484375, "learning_rate": 8.117469879518074e-06, "loss": 0.8896, "step": 375 }, { "epoch": 0.5662650602409639, "grad_norm": 0.2255859375, "learning_rate": 8.112449799196788e-06, "loss": 0.8952, "step": 376 }, { "epoch": 0.5677710843373494, "grad_norm": 0.1875, "learning_rate": 8.107429718875503e-06, "loss": 0.8307, "step": 377 }, { "epoch": 0.5692771084337349, "grad_norm": 0.201171875, "learning_rate": 8.102409638554217e-06, "loss": 0.9034, "step": 378 }, { "epoch": 0.5707831325301205, "grad_norm": 0.21484375, "learning_rate": 8.097389558232932e-06, "loss": 0.9421, "step": 379 }, { "epoch": 0.572289156626506, "grad_norm": 0.2041015625, "learning_rate": 8.092369477911646e-06, "loss": 0.9167, "step": 380 }, { "epoch": 0.5737951807228916, "grad_norm": 0.2734375, "learning_rate": 8.087349397590362e-06, "loss": 0.8939, "step": 381 }, { "epoch": 0.5753012048192772, "grad_norm": 0.2158203125, "learning_rate": 8.082329317269077e-06, "loss": 0.9113, "step": 382 }, { "epoch": 0.5768072289156626, "grad_norm": 0.2177734375, "learning_rate": 8.077309236947792e-06, "loss": 0.9454, "step": 383 }, { "epoch": 0.5783132530120482, "grad_norm": 0.2255859375, "learning_rate": 8.072289156626508e-06, "loss": 0.8946, "step": 384 }, { "epoch": 0.5798192771084337, "grad_norm": 0.2080078125, "learning_rate": 8.067269076305222e-06, "loss": 0.9053, "step": 385 }, { "epoch": 0.5813253012048193, "grad_norm": 0.255859375, "learning_rate": 8.062248995983937e-06, "loss": 0.8906, "step": 386 }, { "epoch": 0.5828313253012049, "grad_norm": 0.2265625, "learning_rate": 8.057228915662651e-06, "loss": 0.8813, "step": 387 }, { "epoch": 0.5843373493975904, "grad_norm": 0.201171875, "learning_rate": 8.052208835341366e-06, "loss": 0.8939, "step": 388 }, { "epoch": 0.5858433734939759, "grad_norm": 0.2470703125, "learning_rate": 8.04718875502008e-06, "loss": 0.8703, "step": 389 }, { "epoch": 0.5873493975903614, "grad_norm": 0.2412109375, "learning_rate": 8.042168674698797e-06, "loss": 0.926, "step": 390 }, { "epoch": 0.588855421686747, "grad_norm": 0.2255859375, "learning_rate": 8.037148594377511e-06, "loss": 0.9002, "step": 391 }, { "epoch": 0.5903614457831325, "grad_norm": 0.2060546875, "learning_rate": 8.032128514056226e-06, "loss": 0.858, "step": 392 }, { "epoch": 0.5918674698795181, "grad_norm": 0.2109375, "learning_rate": 8.02710843373494e-06, "loss": 0.9492, "step": 393 }, { "epoch": 0.5933734939759037, "grad_norm": 0.205078125, "learning_rate": 8.022088353413655e-06, "loss": 0.8925, "step": 394 }, { "epoch": 0.5948795180722891, "grad_norm": 0.208984375, "learning_rate": 8.01706827309237e-06, "loss": 0.8987, "step": 395 }, { "epoch": 0.5963855421686747, "grad_norm": 0.2109375, "learning_rate": 8.012048192771085e-06, "loss": 0.9186, "step": 396 }, { "epoch": 0.5978915662650602, "grad_norm": 0.2314453125, "learning_rate": 8.0070281124498e-06, "loss": 0.9233, "step": 397 }, { "epoch": 0.5993975903614458, "grad_norm": 0.22265625, "learning_rate": 8.002008032128515e-06, "loss": 0.913, "step": 398 }, { "epoch": 0.6009036144578314, "grad_norm": 0.240234375, "learning_rate": 7.996987951807229e-06, "loss": 0.897, "step": 399 }, { "epoch": 0.6024096385542169, "grad_norm": 0.21484375, "learning_rate": 7.991967871485944e-06, "loss": 0.914, "step": 400 }, { "epoch": 0.6039156626506024, "grad_norm": 0.2080078125, "learning_rate": 7.98694779116466e-06, "loss": 0.919, "step": 401 }, { "epoch": 0.6054216867469879, "grad_norm": 0.3828125, "learning_rate": 7.981927710843374e-06, "loss": 0.9063, "step": 402 }, { "epoch": 0.6069277108433735, "grad_norm": 0.205078125, "learning_rate": 7.976907630522089e-06, "loss": 0.9033, "step": 403 }, { "epoch": 0.608433734939759, "grad_norm": 0.2490234375, "learning_rate": 7.971887550200803e-06, "loss": 0.9061, "step": 404 }, { "epoch": 0.6099397590361446, "grad_norm": 0.201171875, "learning_rate": 7.96686746987952e-06, "loss": 0.9349, "step": 405 }, { "epoch": 0.6114457831325302, "grad_norm": 0.220703125, "learning_rate": 7.961847389558234e-06, "loss": 0.9076, "step": 406 }, { "epoch": 0.6129518072289156, "grad_norm": 0.2138671875, "learning_rate": 7.956827309236949e-06, "loss": 0.8564, "step": 407 }, { "epoch": 0.6144578313253012, "grad_norm": 0.220703125, "learning_rate": 7.951807228915663e-06, "loss": 0.9484, "step": 408 }, { "epoch": 0.6159638554216867, "grad_norm": 0.22265625, "learning_rate": 7.946787148594378e-06, "loss": 0.9349, "step": 409 }, { "epoch": 0.6174698795180723, "grad_norm": 0.208984375, "learning_rate": 7.941767068273092e-06, "loss": 0.9321, "step": 410 }, { "epoch": 0.6189759036144579, "grad_norm": 0.20703125, "learning_rate": 7.936746987951808e-06, "loss": 0.8603, "step": 411 }, { "epoch": 0.6204819277108434, "grad_norm": 0.2001953125, "learning_rate": 7.931726907630523e-06, "loss": 0.8549, "step": 412 }, { "epoch": 0.6219879518072289, "grad_norm": 0.234375, "learning_rate": 7.926706827309237e-06, "loss": 0.8942, "step": 413 }, { "epoch": 0.6234939759036144, "grad_norm": 0.2158203125, "learning_rate": 7.921686746987952e-06, "loss": 0.8991, "step": 414 }, { "epoch": 0.625, "grad_norm": 0.2021484375, "learning_rate": 7.916666666666667e-06, "loss": 0.8873, "step": 415 }, { "epoch": 0.6265060240963856, "grad_norm": 0.2275390625, "learning_rate": 7.911646586345381e-06, "loss": 0.9012, "step": 416 }, { "epoch": 0.6280120481927711, "grad_norm": 0.22265625, "learning_rate": 7.906626506024097e-06, "loss": 0.8718, "step": 417 }, { "epoch": 0.6295180722891566, "grad_norm": 0.232421875, "learning_rate": 7.901606425702812e-06, "loss": 0.8999, "step": 418 }, { "epoch": 0.6310240963855421, "grad_norm": 0.20703125, "learning_rate": 7.896586345381528e-06, "loss": 0.8841, "step": 419 }, { "epoch": 0.6325301204819277, "grad_norm": 0.21875, "learning_rate": 7.891566265060243e-06, "loss": 0.8888, "step": 420 }, { "epoch": 0.6340361445783133, "grad_norm": 0.2119140625, "learning_rate": 7.886546184738957e-06, "loss": 0.8662, "step": 421 }, { "epoch": 0.6355421686746988, "grad_norm": 0.2578125, "learning_rate": 7.881526104417672e-06, "loss": 0.8796, "step": 422 }, { "epoch": 0.6370481927710844, "grad_norm": 1.5234375, "learning_rate": 7.876506024096386e-06, "loss": 0.8935, "step": 423 }, { "epoch": 0.6385542168674698, "grad_norm": 0.22265625, "learning_rate": 7.8714859437751e-06, "loss": 0.9052, "step": 424 }, { "epoch": 0.6400602409638554, "grad_norm": 0.236328125, "learning_rate": 7.866465863453815e-06, "loss": 0.9357, "step": 425 }, { "epoch": 0.641566265060241, "grad_norm": 0.216796875, "learning_rate": 7.861445783132531e-06, "loss": 0.9021, "step": 426 }, { "epoch": 0.6430722891566265, "grad_norm": 0.2255859375, "learning_rate": 7.856425702811246e-06, "loss": 0.8775, "step": 427 }, { "epoch": 0.6445783132530121, "grad_norm": 0.255859375, "learning_rate": 7.85140562248996e-06, "loss": 0.907, "step": 428 }, { "epoch": 0.6460843373493976, "grad_norm": 0.318359375, "learning_rate": 7.846385542168675e-06, "loss": 0.9229, "step": 429 }, { "epoch": 0.6475903614457831, "grad_norm": 0.328125, "learning_rate": 7.84136546184739e-06, "loss": 0.9079, "step": 430 }, { "epoch": 0.6490963855421686, "grad_norm": 0.23046875, "learning_rate": 7.836345381526104e-06, "loss": 0.8903, "step": 431 }, { "epoch": 0.6506024096385542, "grad_norm": 0.2314453125, "learning_rate": 7.83132530120482e-06, "loss": 0.8986, "step": 432 }, { "epoch": 0.6521084337349398, "grad_norm": 0.2236328125, "learning_rate": 7.826305220883535e-06, "loss": 0.8843, "step": 433 }, { "epoch": 0.6536144578313253, "grad_norm": 0.2138671875, "learning_rate": 7.82128514056225e-06, "loss": 0.8853, "step": 434 }, { "epoch": 0.6551204819277109, "grad_norm": 0.2158203125, "learning_rate": 7.816265060240964e-06, "loss": 0.8738, "step": 435 }, { "epoch": 0.6566265060240963, "grad_norm": 0.2421875, "learning_rate": 7.81124497991968e-06, "loss": 0.9206, "step": 436 }, { "epoch": 0.6581325301204819, "grad_norm": 0.216796875, "learning_rate": 7.806224899598395e-06, "loss": 0.8855, "step": 437 }, { "epoch": 0.6596385542168675, "grad_norm": 0.205078125, "learning_rate": 7.801204819277109e-06, "loss": 0.8776, "step": 438 }, { "epoch": 0.661144578313253, "grad_norm": 0.2099609375, "learning_rate": 7.796184738955824e-06, "loss": 0.9167, "step": 439 }, { "epoch": 0.6626506024096386, "grad_norm": 0.228515625, "learning_rate": 7.79116465863454e-06, "loss": 0.8995, "step": 440 }, { "epoch": 0.6641566265060241, "grad_norm": 0.2041015625, "learning_rate": 7.786144578313254e-06, "loss": 0.8937, "step": 441 }, { "epoch": 0.6656626506024096, "grad_norm": 0.2265625, "learning_rate": 7.781124497991969e-06, "loss": 0.8849, "step": 442 }, { "epoch": 0.6671686746987951, "grad_norm": 0.25, "learning_rate": 7.776104417670683e-06, "loss": 0.8921, "step": 443 }, { "epoch": 0.6686746987951807, "grad_norm": 0.208984375, "learning_rate": 7.771084337349398e-06, "loss": 0.8794, "step": 444 }, { "epoch": 0.6701807228915663, "grad_norm": 0.205078125, "learning_rate": 7.766064257028112e-06, "loss": 0.8706, "step": 445 }, { "epoch": 0.6716867469879518, "grad_norm": 0.515625, "learning_rate": 7.761044176706827e-06, "loss": 0.882, "step": 446 }, { "epoch": 0.6731927710843374, "grad_norm": 0.236328125, "learning_rate": 7.756024096385543e-06, "loss": 0.9158, "step": 447 }, { "epoch": 0.6746987951807228, "grad_norm": 0.2255859375, "learning_rate": 7.751004016064258e-06, "loss": 0.9052, "step": 448 }, { "epoch": 0.6762048192771084, "grad_norm": 0.2314453125, "learning_rate": 7.745983935742972e-06, "loss": 0.8401, "step": 449 }, { "epoch": 0.677710843373494, "grad_norm": 0.279296875, "learning_rate": 7.740963855421687e-06, "loss": 0.8841, "step": 450 }, { "epoch": 0.6792168674698795, "grad_norm": 0.2099609375, "learning_rate": 7.735943775100401e-06, "loss": 0.912, "step": 451 }, { "epoch": 0.6807228915662651, "grad_norm": 0.2109375, "learning_rate": 7.730923694779116e-06, "loss": 0.8904, "step": 452 }, { "epoch": 0.6822289156626506, "grad_norm": 0.203125, "learning_rate": 7.725903614457832e-06, "loss": 0.8802, "step": 453 }, { "epoch": 0.6837349397590361, "grad_norm": 0.3359375, "learning_rate": 7.720883534136547e-06, "loss": 0.917, "step": 454 }, { "epoch": 0.6852409638554217, "grad_norm": 0.251953125, "learning_rate": 7.715863453815263e-06, "loss": 0.8987, "step": 455 }, { "epoch": 0.6867469879518072, "grad_norm": 0.2353515625, "learning_rate": 7.710843373493977e-06, "loss": 0.8751, "step": 456 }, { "epoch": 0.6882530120481928, "grad_norm": 0.2041015625, "learning_rate": 7.705823293172692e-06, "loss": 0.9132, "step": 457 }, { "epoch": 0.6897590361445783, "grad_norm": 0.224609375, "learning_rate": 7.700803212851406e-06, "loss": 0.8691, "step": 458 }, { "epoch": 0.6912650602409639, "grad_norm": 0.205078125, "learning_rate": 7.695783132530121e-06, "loss": 0.8613, "step": 459 }, { "epoch": 0.6927710843373494, "grad_norm": 0.2177734375, "learning_rate": 7.690763052208835e-06, "loss": 0.9013, "step": 460 }, { "epoch": 0.6942771084337349, "grad_norm": 0.2177734375, "learning_rate": 7.685742971887552e-06, "loss": 0.9268, "step": 461 }, { "epoch": 0.6957831325301205, "grad_norm": 0.2080078125, "learning_rate": 7.680722891566266e-06, "loss": 0.9236, "step": 462 }, { "epoch": 0.697289156626506, "grad_norm": 0.310546875, "learning_rate": 7.67570281124498e-06, "loss": 0.9153, "step": 463 }, { "epoch": 0.6987951807228916, "grad_norm": 0.275390625, "learning_rate": 7.670682730923695e-06, "loss": 0.9038, "step": 464 }, { "epoch": 0.7003012048192772, "grad_norm": 0.265625, "learning_rate": 7.66566265060241e-06, "loss": 0.8768, "step": 465 }, { "epoch": 0.7018072289156626, "grad_norm": 0.271484375, "learning_rate": 7.660642570281124e-06, "loss": 0.9009, "step": 466 }, { "epoch": 0.7033132530120482, "grad_norm": 0.2333984375, "learning_rate": 7.655622489959839e-06, "loss": 0.8774, "step": 467 }, { "epoch": 0.7048192771084337, "grad_norm": 0.2109375, "learning_rate": 7.650602409638555e-06, "loss": 0.86, "step": 468 }, { "epoch": 0.7063253012048193, "grad_norm": 0.20703125, "learning_rate": 7.64558232931727e-06, "loss": 0.8934, "step": 469 }, { "epoch": 0.7078313253012049, "grad_norm": 0.20703125, "learning_rate": 7.640562248995984e-06, "loss": 0.8552, "step": 470 }, { "epoch": 0.7093373493975904, "grad_norm": 0.2412109375, "learning_rate": 7.635542168674699e-06, "loss": 0.8795, "step": 471 }, { "epoch": 0.7108433734939759, "grad_norm": 0.205078125, "learning_rate": 7.630522088353415e-06, "loss": 0.8905, "step": 472 }, { "epoch": 0.7123493975903614, "grad_norm": 0.22265625, "learning_rate": 7.625502008032129e-06, "loss": 0.8762, "step": 473 }, { "epoch": 0.713855421686747, "grad_norm": 0.22265625, "learning_rate": 7.620481927710845e-06, "loss": 0.9204, "step": 474 }, { "epoch": 0.7153614457831325, "grad_norm": 0.2138671875, "learning_rate": 7.615461847389559e-06, "loss": 0.9241, "step": 475 }, { "epoch": 0.7168674698795181, "grad_norm": 0.203125, "learning_rate": 7.610441767068274e-06, "loss": 0.8496, "step": 476 }, { "epoch": 0.7183734939759037, "grad_norm": 0.2421875, "learning_rate": 7.605421686746988e-06, "loss": 0.9456, "step": 477 }, { "epoch": 0.7198795180722891, "grad_norm": 0.271484375, "learning_rate": 7.600401606425704e-06, "loss": 0.9034, "step": 478 }, { "epoch": 0.7213855421686747, "grad_norm": 0.2099609375, "learning_rate": 7.595381526104418e-06, "loss": 0.9107, "step": 479 }, { "epoch": 0.7228915662650602, "grad_norm": 0.216796875, "learning_rate": 7.590361445783133e-06, "loss": 0.8449, "step": 480 }, { "epoch": 0.7243975903614458, "grad_norm": 0.244140625, "learning_rate": 7.585341365461848e-06, "loss": 0.8846, "step": 481 }, { "epoch": 0.7259036144578314, "grad_norm": 0.2099609375, "learning_rate": 7.580321285140563e-06, "loss": 0.8918, "step": 482 }, { "epoch": 0.7274096385542169, "grad_norm": 0.2109375, "learning_rate": 7.575301204819277e-06, "loss": 0.9189, "step": 483 }, { "epoch": 0.7289156626506024, "grad_norm": 0.19921875, "learning_rate": 7.570281124497993e-06, "loss": 0.8532, "step": 484 }, { "epoch": 0.7304216867469879, "grad_norm": 0.462890625, "learning_rate": 7.565261044176707e-06, "loss": 0.8696, "step": 485 }, { "epoch": 0.7319277108433735, "grad_norm": 0.2119140625, "learning_rate": 7.560240963855422e-06, "loss": 0.9119, "step": 486 }, { "epoch": 0.733433734939759, "grad_norm": 0.21875, "learning_rate": 7.555220883534137e-06, "loss": 0.8745, "step": 487 }, { "epoch": 0.7349397590361446, "grad_norm": 0.2109375, "learning_rate": 7.5502008032128515e-06, "loss": 0.8942, "step": 488 }, { "epoch": 0.7364457831325302, "grad_norm": 0.22265625, "learning_rate": 7.545180722891566e-06, "loss": 0.9233, "step": 489 }, { "epoch": 0.7379518072289156, "grad_norm": 0.2158203125, "learning_rate": 7.540160642570282e-06, "loss": 0.8716, "step": 490 }, { "epoch": 0.7394578313253012, "grad_norm": 0.2470703125, "learning_rate": 7.535140562248997e-06, "loss": 0.9035, "step": 491 }, { "epoch": 0.7409638554216867, "grad_norm": 0.26171875, "learning_rate": 7.530120481927712e-06, "loss": 0.8819, "step": 492 }, { "epoch": 0.7424698795180723, "grad_norm": 0.20703125, "learning_rate": 7.525100401606427e-06, "loss": 0.8935, "step": 493 }, { "epoch": 0.7439759036144579, "grad_norm": 0.216796875, "learning_rate": 7.520080321285141e-06, "loss": 0.9025, "step": 494 }, { "epoch": 0.7454819277108434, "grad_norm": 0.212890625, "learning_rate": 7.515060240963857e-06, "loss": 0.8897, "step": 495 }, { "epoch": 0.7469879518072289, "grad_norm": 0.26171875, "learning_rate": 7.510040160642571e-06, "loss": 0.9257, "step": 496 }, { "epoch": 0.7484939759036144, "grad_norm": 0.2080078125, "learning_rate": 7.505020080321286e-06, "loss": 0.8823, "step": 497 }, { "epoch": 0.75, "grad_norm": 0.2080078125, "learning_rate": 7.500000000000001e-06, "loss": 0.8816, "step": 498 }, { "epoch": 0.7515060240963856, "grad_norm": 0.2392578125, "learning_rate": 7.4949799196787156e-06, "loss": 0.874, "step": 499 }, { "epoch": 0.7530120481927711, "grad_norm": 0.6328125, "learning_rate": 7.48995983935743e-06, "loss": 0.8545, "step": 500 }, { "epoch": 0.7545180722891566, "grad_norm": 0.2099609375, "learning_rate": 7.484939759036145e-06, "loss": 0.8474, "step": 501 }, { "epoch": 0.7560240963855421, "grad_norm": 0.2021484375, "learning_rate": 7.47991967871486e-06, "loss": 0.8436, "step": 502 }, { "epoch": 0.7575301204819277, "grad_norm": 0.2734375, "learning_rate": 7.4748995983935745e-06, "loss": 0.9038, "step": 503 }, { "epoch": 0.7590361445783133, "grad_norm": 0.208984375, "learning_rate": 7.469879518072289e-06, "loss": 0.8349, "step": 504 }, { "epoch": 0.7605421686746988, "grad_norm": 0.267578125, "learning_rate": 7.464859437751004e-06, "loss": 0.8854, "step": 505 }, { "epoch": 0.7620481927710844, "grad_norm": 0.263671875, "learning_rate": 7.459839357429719e-06, "loss": 0.9337, "step": 506 }, { "epoch": 0.7635542168674698, "grad_norm": 0.21484375, "learning_rate": 7.454819277108435e-06, "loss": 0.8829, "step": 507 }, { "epoch": 0.7650602409638554, "grad_norm": 0.21875, "learning_rate": 7.44979919678715e-06, "loss": 0.8254, "step": 508 }, { "epoch": 0.766566265060241, "grad_norm": 0.259765625, "learning_rate": 7.444779116465864e-06, "loss": 0.855, "step": 509 }, { "epoch": 0.7680722891566265, "grad_norm": 0.23828125, "learning_rate": 7.43975903614458e-06, "loss": 0.8876, "step": 510 }, { "epoch": 0.7695783132530121, "grad_norm": 0.2158203125, "learning_rate": 7.434738955823294e-06, "loss": 0.8853, "step": 511 }, { "epoch": 0.7710843373493976, "grad_norm": 0.2236328125, "learning_rate": 7.429718875502009e-06, "loss": 0.8635, "step": 512 }, { "epoch": 0.7725903614457831, "grad_norm": 0.228515625, "learning_rate": 7.424698795180724e-06, "loss": 0.8292, "step": 513 }, { "epoch": 0.7740963855421686, "grad_norm": 0.2138671875, "learning_rate": 7.4196787148594385e-06, "loss": 0.8691, "step": 514 }, { "epoch": 0.7756024096385542, "grad_norm": 0.236328125, "learning_rate": 7.414658634538153e-06, "loss": 0.92, "step": 515 }, { "epoch": 0.7771084337349398, "grad_norm": 0.224609375, "learning_rate": 7.4096385542168684e-06, "loss": 0.8907, "step": 516 }, { "epoch": 0.7786144578313253, "grad_norm": 0.3671875, "learning_rate": 7.404618473895583e-06, "loss": 0.8776, "step": 517 }, { "epoch": 0.7801204819277109, "grad_norm": 0.2109375, "learning_rate": 7.3995983935742975e-06, "loss": 0.8483, "step": 518 }, { "epoch": 0.7816265060240963, "grad_norm": 0.2216796875, "learning_rate": 7.394578313253013e-06, "loss": 0.8943, "step": 519 }, { "epoch": 0.7831325301204819, "grad_norm": 0.267578125, "learning_rate": 7.389558232931727e-06, "loss": 0.884, "step": 520 }, { "epoch": 0.7846385542168675, "grad_norm": 0.23046875, "learning_rate": 7.384538152610442e-06, "loss": 0.8675, "step": 521 }, { "epoch": 0.786144578313253, "grad_norm": 0.23828125, "learning_rate": 7.3795180722891564e-06, "loss": 0.8562, "step": 522 }, { "epoch": 0.7876506024096386, "grad_norm": 0.2109375, "learning_rate": 7.374497991967872e-06, "loss": 0.8594, "step": 523 }, { "epoch": 0.7891566265060241, "grad_norm": 0.23046875, "learning_rate": 7.369477911646586e-06, "loss": 0.8655, "step": 524 }, { "epoch": 0.7906626506024096, "grad_norm": 0.2197265625, "learning_rate": 7.3644578313253026e-06, "loss": 0.8967, "step": 525 }, { "epoch": 0.7921686746987951, "grad_norm": 0.2119140625, "learning_rate": 7.359437751004017e-06, "loss": 0.8559, "step": 526 }, { "epoch": 0.7936746987951807, "grad_norm": 0.3515625, "learning_rate": 7.354417670682732e-06, "loss": 0.8436, "step": 527 }, { "epoch": 0.7951807228915663, "grad_norm": 0.26171875, "learning_rate": 7.349397590361447e-06, "loss": 0.8564, "step": 528 }, { "epoch": 0.7966867469879518, "grad_norm": 0.2197265625, "learning_rate": 7.3443775100401615e-06, "loss": 0.8686, "step": 529 }, { "epoch": 0.7981927710843374, "grad_norm": 0.2197265625, "learning_rate": 7.339357429718876e-06, "loss": 0.8876, "step": 530 }, { "epoch": 0.7996987951807228, "grad_norm": 0.2060546875, "learning_rate": 7.334337349397591e-06, "loss": 0.854, "step": 531 }, { "epoch": 0.8012048192771084, "grad_norm": 0.2177734375, "learning_rate": 7.329317269076306e-06, "loss": 0.9133, "step": 532 }, { "epoch": 0.802710843373494, "grad_norm": 0.2265625, "learning_rate": 7.3242971887550205e-06, "loss": 0.8877, "step": 533 }, { "epoch": 0.8042168674698795, "grad_norm": 0.369140625, "learning_rate": 7.319277108433736e-06, "loss": 0.8468, "step": 534 }, { "epoch": 0.8057228915662651, "grad_norm": 0.2080078125, "learning_rate": 7.31425702811245e-06, "loss": 0.8619, "step": 535 }, { "epoch": 0.8072289156626506, "grad_norm": 0.2177734375, "learning_rate": 7.309236947791165e-06, "loss": 0.8366, "step": 536 }, { "epoch": 0.8087349397590361, "grad_norm": 0.2080078125, "learning_rate": 7.30421686746988e-06, "loss": 0.8777, "step": 537 }, { "epoch": 0.8102409638554217, "grad_norm": 0.2177734375, "learning_rate": 7.299196787148595e-06, "loss": 0.8597, "step": 538 }, { "epoch": 0.8117469879518072, "grad_norm": 0.26953125, "learning_rate": 7.294176706827309e-06, "loss": 0.8738, "step": 539 }, { "epoch": 0.8132530120481928, "grad_norm": 0.2197265625, "learning_rate": 7.289156626506025e-06, "loss": 0.8579, "step": 540 }, { "epoch": 0.8147590361445783, "grad_norm": 0.2236328125, "learning_rate": 7.284136546184739e-06, "loss": 0.8624, "step": 541 }, { "epoch": 0.8162650602409639, "grad_norm": 0.21875, "learning_rate": 7.279116465863454e-06, "loss": 0.9425, "step": 542 }, { "epoch": 0.8177710843373494, "grad_norm": 0.251953125, "learning_rate": 7.27409638554217e-06, "loss": 0.8858, "step": 543 }, { "epoch": 0.8192771084337349, "grad_norm": 0.2490234375, "learning_rate": 7.2690763052208845e-06, "loss": 0.8669, "step": 544 }, { "epoch": 0.8207831325301205, "grad_norm": 0.3046875, "learning_rate": 7.264056224899599e-06, "loss": 0.8612, "step": 545 }, { "epoch": 0.822289156626506, "grad_norm": 0.232421875, "learning_rate": 7.259036144578314e-06, "loss": 0.8727, "step": 546 }, { "epoch": 0.8237951807228916, "grad_norm": 0.232421875, "learning_rate": 7.254016064257029e-06, "loss": 0.8879, "step": 547 }, { "epoch": 0.8253012048192772, "grad_norm": 0.255859375, "learning_rate": 7.2489959839357435e-06, "loss": 0.9011, "step": 548 }, { "epoch": 0.8268072289156626, "grad_norm": 0.232421875, "learning_rate": 7.243975903614459e-06, "loss": 0.8287, "step": 549 }, { "epoch": 0.8283132530120482, "grad_norm": 0.21875, "learning_rate": 7.238955823293173e-06, "loss": 0.8136, "step": 550 }, { "epoch": 0.8298192771084337, "grad_norm": 0.2392578125, "learning_rate": 7.233935742971888e-06, "loss": 0.8775, "step": 551 }, { "epoch": 0.8313253012048193, "grad_norm": 0.283203125, "learning_rate": 7.228915662650603e-06, "loss": 0.8767, "step": 552 }, { "epoch": 0.8328313253012049, "grad_norm": 0.2314453125, "learning_rate": 7.223895582329318e-06, "loss": 0.8702, "step": 553 }, { "epoch": 0.8343373493975904, "grad_norm": 0.2255859375, "learning_rate": 7.218875502008032e-06, "loss": 0.9172, "step": 554 }, { "epoch": 0.8358433734939759, "grad_norm": 0.212890625, "learning_rate": 7.213855421686748e-06, "loss": 0.8462, "step": 555 }, { "epoch": 0.8373493975903614, "grad_norm": 0.21484375, "learning_rate": 7.208835341365462e-06, "loss": 0.8658, "step": 556 }, { "epoch": 0.838855421686747, "grad_norm": 0.2265625, "learning_rate": 7.203815261044177e-06, "loss": 0.8364, "step": 557 }, { "epoch": 0.8403614457831325, "grad_norm": 0.2353515625, "learning_rate": 7.198795180722892e-06, "loss": 0.8591, "step": 558 }, { "epoch": 0.8418674698795181, "grad_norm": 0.23828125, "learning_rate": 7.193775100401607e-06, "loss": 0.8932, "step": 559 }, { "epoch": 0.8433734939759037, "grad_norm": 0.27734375, "learning_rate": 7.188755020080321e-06, "loss": 0.878, "step": 560 }, { "epoch": 0.8448795180722891, "grad_norm": 0.240234375, "learning_rate": 7.183734939759037e-06, "loss": 0.8808, "step": 561 }, { "epoch": 0.8463855421686747, "grad_norm": 0.2197265625, "learning_rate": 7.178714859437752e-06, "loss": 0.8871, "step": 562 }, { "epoch": 0.8478915662650602, "grad_norm": 0.216796875, "learning_rate": 7.1736947791164664e-06, "loss": 0.8504, "step": 563 }, { "epoch": 0.8493975903614458, "grad_norm": 0.228515625, "learning_rate": 7.168674698795182e-06, "loss": 0.8602, "step": 564 }, { "epoch": 0.8509036144578314, "grad_norm": 0.224609375, "learning_rate": 7.163654618473896e-06, "loss": 0.896, "step": 565 }, { "epoch": 0.8524096385542169, "grad_norm": 0.2119140625, "learning_rate": 7.158634538152611e-06, "loss": 0.8912, "step": 566 }, { "epoch": 0.8539156626506024, "grad_norm": 0.22265625, "learning_rate": 7.153614457831326e-06, "loss": 0.8678, "step": 567 }, { "epoch": 0.8554216867469879, "grad_norm": 0.23828125, "learning_rate": 7.148594377510041e-06, "loss": 0.8916, "step": 568 }, { "epoch": 0.8569277108433735, "grad_norm": 0.25, "learning_rate": 7.143574297188755e-06, "loss": 0.8961, "step": 569 }, { "epoch": 0.858433734939759, "grad_norm": 0.2255859375, "learning_rate": 7.138554216867471e-06, "loss": 0.8888, "step": 570 }, { "epoch": 0.8599397590361446, "grad_norm": 0.2265625, "learning_rate": 7.133534136546185e-06, "loss": 0.8369, "step": 571 }, { "epoch": 0.8614457831325302, "grad_norm": 0.23046875, "learning_rate": 7.1285140562249e-06, "loss": 0.8444, "step": 572 }, { "epoch": 0.8629518072289156, "grad_norm": 0.32421875, "learning_rate": 7.123493975903615e-06, "loss": 0.8591, "step": 573 }, { "epoch": 0.8644578313253012, "grad_norm": 0.2216796875, "learning_rate": 7.11847389558233e-06, "loss": 0.8621, "step": 574 }, { "epoch": 0.8659638554216867, "grad_norm": 0.578125, "learning_rate": 7.113453815261044e-06, "loss": 0.8212, "step": 575 }, { "epoch": 0.8674698795180723, "grad_norm": 0.435546875, "learning_rate": 7.1084337349397595e-06, "loss": 0.8754, "step": 576 }, { "epoch": 0.8689759036144579, "grad_norm": 0.2578125, "learning_rate": 7.103413654618474e-06, "loss": 0.842, "step": 577 }, { "epoch": 0.8704819277108434, "grad_norm": 0.2353515625, "learning_rate": 7.0983935742971886e-06, "loss": 0.8936, "step": 578 }, { "epoch": 0.8719879518072289, "grad_norm": 0.2265625, "learning_rate": 7.093373493975905e-06, "loss": 0.8746, "step": 579 }, { "epoch": 0.8734939759036144, "grad_norm": 0.224609375, "learning_rate": 7.088353413654619e-06, "loss": 0.9165, "step": 580 }, { "epoch": 0.875, "grad_norm": 0.2197265625, "learning_rate": 7.083333333333335e-06, "loss": 0.8868, "step": 581 }, { "epoch": 0.8765060240963856, "grad_norm": 0.2314453125, "learning_rate": 7.078313253012049e-06, "loss": 0.8867, "step": 582 }, { "epoch": 0.8780120481927711, "grad_norm": 0.216796875, "learning_rate": 7.073293172690764e-06, "loss": 0.882, "step": 583 }, { "epoch": 0.8795180722891566, "grad_norm": 0.232421875, "learning_rate": 7.068273092369478e-06, "loss": 0.8629, "step": 584 }, { "epoch": 0.8810240963855421, "grad_norm": 0.2265625, "learning_rate": 7.063253012048194e-06, "loss": 0.8536, "step": 585 }, { "epoch": 0.8825301204819277, "grad_norm": 0.232421875, "learning_rate": 7.058232931726908e-06, "loss": 0.8583, "step": 586 }, { "epoch": 0.8840361445783133, "grad_norm": 0.255859375, "learning_rate": 7.053212851405623e-06, "loss": 0.9088, "step": 587 }, { "epoch": 0.8855421686746988, "grad_norm": 0.2255859375, "learning_rate": 7.048192771084338e-06, "loss": 0.8807, "step": 588 }, { "epoch": 0.8870481927710844, "grad_norm": 0.298828125, "learning_rate": 7.043172690763053e-06, "loss": 0.8074, "step": 589 }, { "epoch": 0.8885542168674698, "grad_norm": 0.2216796875, "learning_rate": 7.038152610441767e-06, "loss": 0.8962, "step": 590 }, { "epoch": 0.8900602409638554, "grad_norm": 0.2392578125, "learning_rate": 7.0331325301204825e-06, "loss": 0.9045, "step": 591 }, { "epoch": 0.891566265060241, "grad_norm": 0.212890625, "learning_rate": 7.028112449799197e-06, "loss": 0.7944, "step": 592 }, { "epoch": 0.8930722891566265, "grad_norm": 0.271484375, "learning_rate": 7.0230923694779115e-06, "loss": 0.9126, "step": 593 }, { "epoch": 0.8945783132530121, "grad_norm": 0.2294921875, "learning_rate": 7.018072289156627e-06, "loss": 0.899, "step": 594 }, { "epoch": 0.8960843373493976, "grad_norm": 0.22265625, "learning_rate": 7.0130522088353414e-06, "loss": 0.8695, "step": 595 }, { "epoch": 0.8975903614457831, "grad_norm": 0.265625, "learning_rate": 7.008032128514058e-06, "loss": 0.8619, "step": 596 }, { "epoch": 0.8990963855421686, "grad_norm": 0.2197265625, "learning_rate": 7.003012048192772e-06, "loss": 0.8642, "step": 597 }, { "epoch": 0.9006024096385542, "grad_norm": 0.2451171875, "learning_rate": 6.997991967871487e-06, "loss": 0.9016, "step": 598 }, { "epoch": 0.9021084337349398, "grad_norm": 0.208984375, "learning_rate": 6.992971887550202e-06, "loss": 0.8555, "step": 599 }, { "epoch": 0.9036144578313253, "grad_norm": 0.2734375, "learning_rate": 6.987951807228917e-06, "loss": 0.9057, "step": 600 }, { "epoch": 0.9051204819277109, "grad_norm": 0.2392578125, "learning_rate": 6.982931726907631e-06, "loss": 0.8653, "step": 601 }, { "epoch": 0.9066265060240963, "grad_norm": 0.2138671875, "learning_rate": 6.9779116465863465e-06, "loss": 0.8904, "step": 602 }, { "epoch": 0.9081325301204819, "grad_norm": 0.2333984375, "learning_rate": 6.972891566265061e-06, "loss": 0.8405, "step": 603 }, { "epoch": 0.9096385542168675, "grad_norm": 0.41796875, "learning_rate": 6.9678714859437756e-06, "loss": 0.8692, "step": 604 }, { "epoch": 0.911144578313253, "grad_norm": 0.365234375, "learning_rate": 6.96285140562249e-06, "loss": 0.8214, "step": 605 }, { "epoch": 0.9126506024096386, "grad_norm": 0.2197265625, "learning_rate": 6.9578313253012055e-06, "loss": 0.8801, "step": 606 }, { "epoch": 0.9141566265060241, "grad_norm": 0.2294921875, "learning_rate": 6.95281124497992e-06, "loss": 0.8742, "step": 607 }, { "epoch": 0.9156626506024096, "grad_norm": 0.2216796875, "learning_rate": 6.9477911646586345e-06, "loss": 0.8552, "step": 608 }, { "epoch": 0.9171686746987951, "grad_norm": 0.2216796875, "learning_rate": 6.94277108433735e-06, "loss": 0.8405, "step": 609 }, { "epoch": 0.9186746987951807, "grad_norm": 0.216796875, "learning_rate": 6.9377510040160644e-06, "loss": 0.8302, "step": 610 }, { "epoch": 0.9201807228915663, "grad_norm": 0.21875, "learning_rate": 6.932730923694779e-06, "loss": 0.9197, "step": 611 }, { "epoch": 0.9216867469879518, "grad_norm": 0.2265625, "learning_rate": 6.927710843373494e-06, "loss": 0.8648, "step": 612 }, { "epoch": 0.9231927710843374, "grad_norm": 0.2275390625, "learning_rate": 6.922690763052209e-06, "loss": 0.8639, "step": 613 }, { "epoch": 0.9246987951807228, "grad_norm": 0.2197265625, "learning_rate": 6.917670682730925e-06, "loss": 0.8643, "step": 614 }, { "epoch": 0.9262048192771084, "grad_norm": 0.21875, "learning_rate": 6.91265060240964e-06, "loss": 0.8824, "step": 615 }, { "epoch": 0.927710843373494, "grad_norm": 0.228515625, "learning_rate": 6.907630522088354e-06, "loss": 0.8694, "step": 616 }, { "epoch": 0.9292168674698795, "grad_norm": 0.287109375, "learning_rate": 6.9026104417670695e-06, "loss": 0.8186, "step": 617 }, { "epoch": 0.9307228915662651, "grad_norm": 0.251953125, "learning_rate": 6.897590361445784e-06, "loss": 0.8261, "step": 618 }, { "epoch": 0.9322289156626506, "grad_norm": 0.228515625, "learning_rate": 6.8925702811244986e-06, "loss": 0.883, "step": 619 }, { "epoch": 0.9337349397590361, "grad_norm": 0.236328125, "learning_rate": 6.887550200803214e-06, "loss": 0.8986, "step": 620 }, { "epoch": 0.9352409638554217, "grad_norm": 0.26171875, "learning_rate": 6.8825301204819285e-06, "loss": 0.8818, "step": 621 }, { "epoch": 0.9367469879518072, "grad_norm": 0.2109375, "learning_rate": 6.877510040160643e-06, "loss": 0.8388, "step": 622 }, { "epoch": 0.9382530120481928, "grad_norm": 0.2333984375, "learning_rate": 6.872489959839358e-06, "loss": 0.8526, "step": 623 }, { "epoch": 0.9397590361445783, "grad_norm": 0.2578125, "learning_rate": 6.867469879518073e-06, "loss": 0.9028, "step": 624 }, { "epoch": 0.9412650602409639, "grad_norm": 0.2294921875, "learning_rate": 6.862449799196787e-06, "loss": 0.8903, "step": 625 }, { "epoch": 0.9427710843373494, "grad_norm": 0.236328125, "learning_rate": 6.857429718875502e-06, "loss": 0.86, "step": 626 }, { "epoch": 0.9442771084337349, "grad_norm": 0.22265625, "learning_rate": 6.852409638554217e-06, "loss": 0.8675, "step": 627 }, { "epoch": 0.9457831325301205, "grad_norm": 0.208984375, "learning_rate": 6.847389558232932e-06, "loss": 0.854, "step": 628 }, { "epoch": 0.947289156626506, "grad_norm": 0.2177734375, "learning_rate": 6.842369477911646e-06, "loss": 0.8423, "step": 629 }, { "epoch": 0.9487951807228916, "grad_norm": 0.2353515625, "learning_rate": 6.837349397590362e-06, "loss": 0.8789, "step": 630 }, { "epoch": 0.9503012048192772, "grad_norm": 0.21875, "learning_rate": 6.832329317269076e-06, "loss": 0.8397, "step": 631 }, { "epoch": 0.9518072289156626, "grad_norm": 0.2294921875, "learning_rate": 6.8273092369477925e-06, "loss": 0.8597, "step": 632 }, { "epoch": 0.9533132530120482, "grad_norm": 0.212890625, "learning_rate": 6.822289156626507e-06, "loss": 0.877, "step": 633 }, { "epoch": 0.9548192771084337, "grad_norm": 0.20703125, "learning_rate": 6.8172690763052215e-06, "loss": 0.8442, "step": 634 }, { "epoch": 0.9563253012048193, "grad_norm": 0.294921875, "learning_rate": 6.812248995983937e-06, "loss": 0.8679, "step": 635 }, { "epoch": 0.9578313253012049, "grad_norm": 0.220703125, "learning_rate": 6.8072289156626514e-06, "loss": 0.8825, "step": 636 }, { "epoch": 0.9593373493975904, "grad_norm": 0.232421875, "learning_rate": 6.802208835341366e-06, "loss": 0.9002, "step": 637 }, { "epoch": 0.9608433734939759, "grad_norm": 0.232421875, "learning_rate": 6.797188755020081e-06, "loss": 0.8688, "step": 638 }, { "epoch": 0.9623493975903614, "grad_norm": 0.271484375, "learning_rate": 6.792168674698796e-06, "loss": 0.8349, "step": 639 }, { "epoch": 0.963855421686747, "grad_norm": 0.220703125, "learning_rate": 6.78714859437751e-06, "loss": 0.8671, "step": 640 }, { "epoch": 0.9653614457831325, "grad_norm": 0.365234375, "learning_rate": 6.782128514056226e-06, "loss": 0.8771, "step": 641 }, { "epoch": 0.9668674698795181, "grad_norm": 0.251953125, "learning_rate": 6.77710843373494e-06, "loss": 0.859, "step": 642 }, { "epoch": 0.9683734939759037, "grad_norm": 0.25390625, "learning_rate": 6.772088353413655e-06, "loss": 0.8884, "step": 643 }, { "epoch": 0.9698795180722891, "grad_norm": 0.2265625, "learning_rate": 6.76706827309237e-06, "loss": 0.8773, "step": 644 }, { "epoch": 0.9713855421686747, "grad_norm": 0.224609375, "learning_rate": 6.762048192771085e-06, "loss": 0.8675, "step": 645 }, { "epoch": 0.9728915662650602, "grad_norm": 0.220703125, "learning_rate": 6.757028112449799e-06, "loss": 0.8645, "step": 646 }, { "epoch": 0.9743975903614458, "grad_norm": 0.265625, "learning_rate": 6.752008032128514e-06, "loss": 0.873, "step": 647 }, { "epoch": 0.9759036144578314, "grad_norm": 0.2373046875, "learning_rate": 6.746987951807229e-06, "loss": 0.856, "step": 648 }, { "epoch": 0.9774096385542169, "grad_norm": 0.2578125, "learning_rate": 6.741967871485944e-06, "loss": 0.8588, "step": 649 }, { "epoch": 0.9789156626506024, "grad_norm": 0.228515625, "learning_rate": 6.73694779116466e-06, "loss": 0.8432, "step": 650 }, { "epoch": 0.9804216867469879, "grad_norm": 0.21484375, "learning_rate": 6.731927710843374e-06, "loss": 0.8612, "step": 651 }, { "epoch": 0.9819277108433735, "grad_norm": 0.216796875, "learning_rate": 6.726907630522089e-06, "loss": 0.8115, "step": 652 }, { "epoch": 0.983433734939759, "grad_norm": 0.25390625, "learning_rate": 6.721887550200804e-06, "loss": 0.8988, "step": 653 }, { "epoch": 0.9849397590361446, "grad_norm": 1.5078125, "learning_rate": 6.716867469879519e-06, "loss": 0.892, "step": 654 }, { "epoch": 0.9864457831325302, "grad_norm": 0.248046875, "learning_rate": 6.711847389558233e-06, "loss": 0.8641, "step": 655 }, { "epoch": 0.9879518072289156, "grad_norm": 0.216796875, "learning_rate": 6.706827309236949e-06, "loss": 0.8659, "step": 656 }, { "epoch": 0.9894578313253012, "grad_norm": 0.37109375, "learning_rate": 6.701807228915663e-06, "loss": 0.8672, "step": 657 }, { "epoch": 0.9909638554216867, "grad_norm": 0.2392578125, "learning_rate": 6.696787148594378e-06, "loss": 0.8598, "step": 658 }, { "epoch": 0.9924698795180723, "grad_norm": 0.2265625, "learning_rate": 6.691767068273093e-06, "loss": 0.86, "step": 659 }, { "epoch": 0.9939759036144579, "grad_norm": 0.228515625, "learning_rate": 6.686746987951808e-06, "loss": 0.8378, "step": 660 }, { "epoch": 0.9954819277108434, "grad_norm": 0.263671875, "learning_rate": 6.681726907630522e-06, "loss": 0.8586, "step": 661 }, { "epoch": 0.9969879518072289, "grad_norm": 0.380859375, "learning_rate": 6.676706827309238e-06, "loss": 0.8684, "step": 662 }, { "epoch": 0.9984939759036144, "grad_norm": 0.220703125, "learning_rate": 6.671686746987952e-06, "loss": 0.8111, "step": 663 }, { "epoch": 1.0, "grad_norm": 0.2255859375, "learning_rate": 6.666666666666667e-06, "loss": 0.8634, "step": 664 }, { "epoch": 1.0015060240963856, "grad_norm": 0.2470703125, "learning_rate": 6.661646586345382e-06, "loss": 0.8614, "step": 665 }, { "epoch": 1.0030120481927711, "grad_norm": 0.228515625, "learning_rate": 6.6566265060240965e-06, "loss": 0.8532, "step": 666 }, { "epoch": 1.0045180722891567, "grad_norm": 0.283203125, "learning_rate": 6.651606425702812e-06, "loss": 0.8567, "step": 667 }, { "epoch": 1.0060240963855422, "grad_norm": 0.2265625, "learning_rate": 6.646586345381527e-06, "loss": 0.8582, "step": 668 }, { "epoch": 1.0075301204819278, "grad_norm": 0.2333984375, "learning_rate": 6.641566265060242e-06, "loss": 0.8707, "step": 669 }, { "epoch": 1.0090361445783131, "grad_norm": 0.2216796875, "learning_rate": 6.636546184738956e-06, "loss": 0.8329, "step": 670 }, { "epoch": 1.0105421686746987, "grad_norm": 0.255859375, "learning_rate": 6.631526104417672e-06, "loss": 0.8413, "step": 671 }, { "epoch": 1.0120481927710843, "grad_norm": 0.2734375, "learning_rate": 6.626506024096386e-06, "loss": 0.8569, "step": 672 }, { "epoch": 1.0135542168674698, "grad_norm": 0.2421875, "learning_rate": 6.621485943775101e-06, "loss": 0.8338, "step": 673 }, { "epoch": 1.0150602409638554, "grad_norm": 0.2314453125, "learning_rate": 6.616465863453816e-06, "loss": 0.891, "step": 674 }, { "epoch": 1.016566265060241, "grad_norm": 0.2314453125, "learning_rate": 6.611445783132531e-06, "loss": 0.8461, "step": 675 }, { "epoch": 1.0180722891566265, "grad_norm": 0.234375, "learning_rate": 6.606425702811245e-06, "loss": 0.8551, "step": 676 }, { "epoch": 1.019578313253012, "grad_norm": 0.2236328125, "learning_rate": 6.6014056224899606e-06, "loss": 0.9115, "step": 677 }, { "epoch": 1.0210843373493976, "grad_norm": 0.2470703125, "learning_rate": 6.596385542168675e-06, "loss": 0.8637, "step": 678 }, { "epoch": 1.0225903614457832, "grad_norm": 0.232421875, "learning_rate": 6.59136546184739e-06, "loss": 0.901, "step": 679 }, { "epoch": 1.0240963855421688, "grad_norm": 0.220703125, "learning_rate": 6.586345381526105e-06, "loss": 0.8602, "step": 680 }, { "epoch": 1.0256024096385543, "grad_norm": 0.2578125, "learning_rate": 6.5813253012048195e-06, "loss": 0.884, "step": 681 }, { "epoch": 1.0271084337349397, "grad_norm": 0.279296875, "learning_rate": 6.576305220883534e-06, "loss": 0.8538, "step": 682 }, { "epoch": 1.0286144578313252, "grad_norm": 0.224609375, "learning_rate": 6.5712851405622494e-06, "loss": 0.8677, "step": 683 }, { "epoch": 1.0301204819277108, "grad_norm": 0.267578125, "learning_rate": 6.566265060240964e-06, "loss": 0.8748, "step": 684 }, { "epoch": 1.0316265060240963, "grad_norm": 0.2216796875, "learning_rate": 6.56124497991968e-06, "loss": 0.891, "step": 685 }, { "epoch": 1.033132530120482, "grad_norm": 0.302734375, "learning_rate": 6.556224899598395e-06, "loss": 0.8293, "step": 686 }, { "epoch": 1.0346385542168675, "grad_norm": 0.2412109375, "learning_rate": 6.551204819277109e-06, "loss": 0.8904, "step": 687 }, { "epoch": 1.036144578313253, "grad_norm": 0.2490234375, "learning_rate": 6.546184738955825e-06, "loss": 0.8274, "step": 688 }, { "epoch": 1.0376506024096386, "grad_norm": 0.255859375, "learning_rate": 6.541164658634539e-06, "loss": 0.8446, "step": 689 }, { "epoch": 1.0391566265060241, "grad_norm": 0.2373046875, "learning_rate": 6.536144578313254e-06, "loss": 0.874, "step": 690 }, { "epoch": 1.0406626506024097, "grad_norm": 0.234375, "learning_rate": 6.531124497991968e-06, "loss": 0.8823, "step": 691 }, { "epoch": 1.0421686746987953, "grad_norm": 0.240234375, "learning_rate": 6.5261044176706836e-06, "loss": 0.8162, "step": 692 }, { "epoch": 1.0436746987951808, "grad_norm": 0.255859375, "learning_rate": 6.521084337349398e-06, "loss": 0.8989, "step": 693 }, { "epoch": 1.0451807228915662, "grad_norm": 0.234375, "learning_rate": 6.516064257028113e-06, "loss": 0.8451, "step": 694 }, { "epoch": 1.0466867469879517, "grad_norm": 0.224609375, "learning_rate": 6.511044176706828e-06, "loss": 0.8399, "step": 695 }, { "epoch": 1.0481927710843373, "grad_norm": 0.2578125, "learning_rate": 6.5060240963855425e-06, "loss": 0.8223, "step": 696 }, { "epoch": 1.0496987951807228, "grad_norm": 0.2333984375, "learning_rate": 6.501004016064257e-06, "loss": 0.8888, "step": 697 }, { "epoch": 1.0512048192771084, "grad_norm": 0.259765625, "learning_rate": 6.495983935742972e-06, "loss": 0.8256, "step": 698 }, { "epoch": 1.052710843373494, "grad_norm": 0.25, "learning_rate": 6.490963855421687e-06, "loss": 0.8598, "step": 699 }, { "epoch": 1.0542168674698795, "grad_norm": 0.2275390625, "learning_rate": 6.4859437751004015e-06, "loss": 0.8554, "step": 700 }, { "epoch": 1.055722891566265, "grad_norm": 0.2392578125, "learning_rate": 6.480923694779117e-06, "loss": 0.8402, "step": 701 }, { "epoch": 1.0572289156626506, "grad_norm": 0.232421875, "learning_rate": 6.475903614457831e-06, "loss": 0.8329, "step": 702 }, { "epoch": 1.0587349397590362, "grad_norm": 0.2451171875, "learning_rate": 6.470883534136548e-06, "loss": 0.8521, "step": 703 }, { "epoch": 1.0602409638554218, "grad_norm": 0.333984375, "learning_rate": 6.465863453815262e-06, "loss": 0.8338, "step": 704 }, { "epoch": 1.0617469879518073, "grad_norm": 0.2255859375, "learning_rate": 6.460843373493977e-06, "loss": 0.8156, "step": 705 }, { "epoch": 1.0632530120481927, "grad_norm": 0.2392578125, "learning_rate": 6.455823293172692e-06, "loss": 0.8592, "step": 706 }, { "epoch": 1.0647590361445782, "grad_norm": 0.2412109375, "learning_rate": 6.4508032128514065e-06, "loss": 0.8615, "step": 707 }, { "epoch": 1.0662650602409638, "grad_norm": 0.232421875, "learning_rate": 6.445783132530121e-06, "loss": 0.8685, "step": 708 }, { "epoch": 1.0677710843373494, "grad_norm": 0.236328125, "learning_rate": 6.4407630522088364e-06, "loss": 0.8961, "step": 709 }, { "epoch": 1.069277108433735, "grad_norm": 0.2275390625, "learning_rate": 6.435742971887551e-06, "loss": 0.8537, "step": 710 }, { "epoch": 1.0707831325301205, "grad_norm": 0.263671875, "learning_rate": 6.4307228915662655e-06, "loss": 0.8122, "step": 711 }, { "epoch": 1.072289156626506, "grad_norm": 0.2890625, "learning_rate": 6.42570281124498e-06, "loss": 0.8461, "step": 712 }, { "epoch": 1.0737951807228916, "grad_norm": 0.275390625, "learning_rate": 6.420682730923695e-06, "loss": 0.8243, "step": 713 }, { "epoch": 1.0753012048192772, "grad_norm": 0.234375, "learning_rate": 6.41566265060241e-06, "loss": 0.8428, "step": 714 }, { "epoch": 1.0768072289156627, "grad_norm": 0.232421875, "learning_rate": 6.4106425702811244e-06, "loss": 0.8742, "step": 715 }, { "epoch": 1.0783132530120483, "grad_norm": 0.2314453125, "learning_rate": 6.40562248995984e-06, "loss": 0.8374, "step": 716 }, { "epoch": 1.0798192771084336, "grad_norm": 0.546875, "learning_rate": 6.400602409638554e-06, "loss": 0.9472, "step": 717 }, { "epoch": 1.0813253012048192, "grad_norm": 0.28125, "learning_rate": 6.395582329317269e-06, "loss": 0.8606, "step": 718 }, { "epoch": 1.0828313253012047, "grad_norm": 0.2373046875, "learning_rate": 6.390562248995984e-06, "loss": 0.8874, "step": 719 }, { "epoch": 1.0843373493975903, "grad_norm": 0.26171875, "learning_rate": 6.385542168674699e-06, "loss": 0.843, "step": 720 }, { "epoch": 1.0858433734939759, "grad_norm": 0.271484375, "learning_rate": 6.380522088353415e-06, "loss": 0.8349, "step": 721 }, { "epoch": 1.0873493975903614, "grad_norm": 0.2412109375, "learning_rate": 6.3755020080321295e-06, "loss": 0.8869, "step": 722 }, { "epoch": 1.088855421686747, "grad_norm": 0.2431640625, "learning_rate": 6.370481927710844e-06, "loss": 0.8185, "step": 723 }, { "epoch": 1.0903614457831325, "grad_norm": 0.2431640625, "learning_rate": 6.365461847389559e-06, "loss": 0.8568, "step": 724 }, { "epoch": 1.091867469879518, "grad_norm": 0.228515625, "learning_rate": 6.360441767068274e-06, "loss": 0.8611, "step": 725 }, { "epoch": 1.0933734939759037, "grad_norm": 0.2392578125, "learning_rate": 6.3554216867469885e-06, "loss": 0.8418, "step": 726 }, { "epoch": 1.0948795180722892, "grad_norm": 0.2412109375, "learning_rate": 6.350401606425704e-06, "loss": 0.8681, "step": 727 }, { "epoch": 1.0963855421686748, "grad_norm": 0.2197265625, "learning_rate": 6.345381526104418e-06, "loss": 0.8298, "step": 728 }, { "epoch": 1.0978915662650603, "grad_norm": 0.248046875, "learning_rate": 6.340361445783133e-06, "loss": 0.8452, "step": 729 }, { "epoch": 1.0993975903614457, "grad_norm": 0.2470703125, "learning_rate": 6.335341365461848e-06, "loss": 0.8472, "step": 730 }, { "epoch": 1.1009036144578312, "grad_norm": 0.291015625, "learning_rate": 6.330321285140563e-06, "loss": 0.8561, "step": 731 }, { "epoch": 1.1024096385542168, "grad_norm": 0.30078125, "learning_rate": 6.325301204819277e-06, "loss": 0.7886, "step": 732 }, { "epoch": 1.1039156626506024, "grad_norm": 0.2412109375, "learning_rate": 6.320281124497992e-06, "loss": 0.8663, "step": 733 }, { "epoch": 1.105421686746988, "grad_norm": 0.2470703125, "learning_rate": 6.315261044176707e-06, "loss": 0.8559, "step": 734 }, { "epoch": 1.1069277108433735, "grad_norm": 0.296875, "learning_rate": 6.310240963855422e-06, "loss": 0.8621, "step": 735 }, { "epoch": 1.108433734939759, "grad_norm": 0.23828125, "learning_rate": 6.305220883534136e-06, "loss": 0.8276, "step": 736 }, { "epoch": 1.1099397590361446, "grad_norm": 0.2392578125, "learning_rate": 6.300200803212852e-06, "loss": 0.8514, "step": 737 }, { "epoch": 1.1114457831325302, "grad_norm": 0.486328125, "learning_rate": 6.295180722891566e-06, "loss": 0.8408, "step": 738 }, { "epoch": 1.1129518072289157, "grad_norm": 0.23046875, "learning_rate": 6.290160642570282e-06, "loss": 0.8918, "step": 739 }, { "epoch": 1.1144578313253013, "grad_norm": 0.361328125, "learning_rate": 6.285140562248997e-06, "loss": 0.8645, "step": 740 }, { "epoch": 1.1159638554216866, "grad_norm": 0.2392578125, "learning_rate": 6.2801204819277114e-06, "loss": 0.8452, "step": 741 }, { "epoch": 1.1174698795180722, "grad_norm": 0.255859375, "learning_rate": 6.275100401606427e-06, "loss": 0.8022, "step": 742 }, { "epoch": 1.1189759036144578, "grad_norm": 0.2255859375, "learning_rate": 6.270080321285141e-06, "loss": 0.8569, "step": 743 }, { "epoch": 1.1204819277108433, "grad_norm": 0.2373046875, "learning_rate": 6.265060240963856e-06, "loss": 0.8705, "step": 744 }, { "epoch": 1.1219879518072289, "grad_norm": 0.2216796875, "learning_rate": 6.260040160642571e-06, "loss": 0.8326, "step": 745 }, { "epoch": 1.1234939759036144, "grad_norm": 0.2275390625, "learning_rate": 6.255020080321286e-06, "loss": 0.8973, "step": 746 }, { "epoch": 1.125, "grad_norm": 0.23046875, "learning_rate": 6.25e-06, "loss": 0.8385, "step": 747 }, { "epoch": 1.1265060240963856, "grad_norm": 0.236328125, "learning_rate": 6.244979919678716e-06, "loss": 0.8531, "step": 748 }, { "epoch": 1.1280120481927711, "grad_norm": 0.30078125, "learning_rate": 6.23995983935743e-06, "loss": 0.8473, "step": 749 }, { "epoch": 1.1295180722891567, "grad_norm": 0.2353515625, "learning_rate": 6.234939759036145e-06, "loss": 0.8807, "step": 750 }, { "epoch": 1.1310240963855422, "grad_norm": 0.251953125, "learning_rate": 6.22991967871486e-06, "loss": 0.8233, "step": 751 }, { "epoch": 1.1325301204819278, "grad_norm": 0.251953125, "learning_rate": 6.224899598393575e-06, "loss": 0.8455, "step": 752 }, { "epoch": 1.1340361445783134, "grad_norm": 0.2470703125, "learning_rate": 6.219879518072289e-06, "loss": 0.8401, "step": 753 }, { "epoch": 1.1355421686746987, "grad_norm": 0.2490234375, "learning_rate": 6.214859437751004e-06, "loss": 0.8512, "step": 754 }, { "epoch": 1.1370481927710843, "grad_norm": 0.2392578125, "learning_rate": 6.209839357429719e-06, "loss": 0.8815, "step": 755 }, { "epoch": 1.1385542168674698, "grad_norm": 0.2265625, "learning_rate": 6.2048192771084344e-06, "loss": 0.8632, "step": 756 }, { "epoch": 1.1400602409638554, "grad_norm": 0.2255859375, "learning_rate": 6.19979919678715e-06, "loss": 0.8738, "step": 757 }, { "epoch": 1.141566265060241, "grad_norm": 0.25390625, "learning_rate": 6.194779116465864e-06, "loss": 0.8533, "step": 758 }, { "epoch": 1.1430722891566265, "grad_norm": 0.248046875, "learning_rate": 6.189759036144579e-06, "loss": 0.8521, "step": 759 }, { "epoch": 1.144578313253012, "grad_norm": 0.271484375, "learning_rate": 6.184738955823294e-06, "loss": 0.8658, "step": 760 }, { "epoch": 1.1460843373493976, "grad_norm": 0.2373046875, "learning_rate": 6.179718875502009e-06, "loss": 0.8702, "step": 761 }, { "epoch": 1.1475903614457832, "grad_norm": 0.314453125, "learning_rate": 6.174698795180723e-06, "loss": 0.8475, "step": 762 }, { "epoch": 1.1490963855421688, "grad_norm": 1.0234375, "learning_rate": 6.169678714859439e-06, "loss": 0.8262, "step": 763 }, { "epoch": 1.1506024096385543, "grad_norm": 0.2421875, "learning_rate": 6.164658634538153e-06, "loss": 0.8305, "step": 764 }, { "epoch": 1.1521084337349397, "grad_norm": 0.2255859375, "learning_rate": 6.159638554216868e-06, "loss": 0.8206, "step": 765 }, { "epoch": 1.1536144578313252, "grad_norm": 0.2392578125, "learning_rate": 6.154618473895583e-06, "loss": 0.8573, "step": 766 }, { "epoch": 1.1551204819277108, "grad_norm": 0.337890625, "learning_rate": 6.149598393574298e-06, "loss": 0.862, "step": 767 }, { "epoch": 1.1566265060240963, "grad_norm": 0.322265625, "learning_rate": 6.144578313253012e-06, "loss": 0.8347, "step": 768 }, { "epoch": 1.158132530120482, "grad_norm": 0.31640625, "learning_rate": 6.1395582329317275e-06, "loss": 0.8581, "step": 769 }, { "epoch": 1.1596385542168675, "grad_norm": 0.2333984375, "learning_rate": 6.134538152610442e-06, "loss": 0.8535, "step": 770 }, { "epoch": 1.161144578313253, "grad_norm": 0.255859375, "learning_rate": 6.1295180722891566e-06, "loss": 0.8569, "step": 771 }, { "epoch": 1.1626506024096386, "grad_norm": 0.2373046875, "learning_rate": 6.124497991967872e-06, "loss": 0.8691, "step": 772 }, { "epoch": 1.1641566265060241, "grad_norm": 0.263671875, "learning_rate": 6.1194779116465865e-06, "loss": 0.8139, "step": 773 }, { "epoch": 1.1656626506024097, "grad_norm": 0.2353515625, "learning_rate": 6.114457831325302e-06, "loss": 0.8553, "step": 774 }, { "epoch": 1.1671686746987953, "grad_norm": 0.2333984375, "learning_rate": 6.109437751004017e-06, "loss": 0.869, "step": 775 }, { "epoch": 1.1686746987951806, "grad_norm": 0.2470703125, "learning_rate": 6.104417670682732e-06, "loss": 0.8857, "step": 776 }, { "epoch": 1.1701807228915664, "grad_norm": 0.33984375, "learning_rate": 6.099397590361446e-06, "loss": 0.8528, "step": 777 }, { "epoch": 1.1716867469879517, "grad_norm": 0.330078125, "learning_rate": 6.094377510040162e-06, "loss": 0.8102, "step": 778 }, { "epoch": 1.1731927710843373, "grad_norm": 0.240234375, "learning_rate": 6.089357429718876e-06, "loss": 0.8443, "step": 779 }, { "epoch": 1.1746987951807228, "grad_norm": 0.2421875, "learning_rate": 6.084337349397591e-06, "loss": 0.8494, "step": 780 }, { "epoch": 1.1762048192771084, "grad_norm": 0.236328125, "learning_rate": 6.079317269076306e-06, "loss": 0.8694, "step": 781 }, { "epoch": 1.177710843373494, "grad_norm": 0.240234375, "learning_rate": 6.074297188755021e-06, "loss": 0.8658, "step": 782 }, { "epoch": 1.1792168674698795, "grad_norm": 0.2373046875, "learning_rate": 6.069277108433735e-06, "loss": 0.875, "step": 783 }, { "epoch": 1.180722891566265, "grad_norm": 0.23828125, "learning_rate": 6.0642570281124505e-06, "loss": 0.8495, "step": 784 }, { "epoch": 1.1822289156626506, "grad_norm": 0.2294921875, "learning_rate": 6.059236947791165e-06, "loss": 0.8373, "step": 785 }, { "epoch": 1.1837349397590362, "grad_norm": 0.2451171875, "learning_rate": 6.0542168674698795e-06, "loss": 0.8455, "step": 786 }, { "epoch": 1.1852409638554218, "grad_norm": 0.2373046875, "learning_rate": 6.049196787148595e-06, "loss": 0.8692, "step": 787 }, { "epoch": 1.1867469879518073, "grad_norm": 0.2314453125, "learning_rate": 6.0441767068273094e-06, "loss": 0.8211, "step": 788 }, { "epoch": 1.1882530120481927, "grad_norm": 0.2734375, "learning_rate": 6.039156626506024e-06, "loss": 0.8832, "step": 789 }, { "epoch": 1.1897590361445782, "grad_norm": 0.2734375, "learning_rate": 6.034136546184739e-06, "loss": 0.8103, "step": 790 }, { "epoch": 1.1912650602409638, "grad_norm": 0.2470703125, "learning_rate": 6.029116465863454e-06, "loss": 0.8587, "step": 791 }, { "epoch": 1.1927710843373494, "grad_norm": 0.2470703125, "learning_rate": 6.02409638554217e-06, "loss": 0.896, "step": 792 }, { "epoch": 1.194277108433735, "grad_norm": 0.275390625, "learning_rate": 6.019076305220885e-06, "loss": 0.8723, "step": 793 }, { "epoch": 1.1957831325301205, "grad_norm": 0.240234375, "learning_rate": 6.014056224899599e-06, "loss": 0.8264, "step": 794 }, { "epoch": 1.197289156626506, "grad_norm": 0.2236328125, "learning_rate": 6.009036144578314e-06, "loss": 0.8053, "step": 795 }, { "epoch": 1.1987951807228916, "grad_norm": 0.2490234375, "learning_rate": 6.004016064257029e-06, "loss": 0.8384, "step": 796 }, { "epoch": 1.2003012048192772, "grad_norm": 0.2255859375, "learning_rate": 5.9989959839357436e-06, "loss": 0.7939, "step": 797 }, { "epoch": 1.2018072289156627, "grad_norm": 0.2373046875, "learning_rate": 5.993975903614458e-06, "loss": 0.8772, "step": 798 }, { "epoch": 1.2033132530120483, "grad_norm": 0.2373046875, "learning_rate": 5.9889558232931735e-06, "loss": 0.8642, "step": 799 }, { "epoch": 1.2048192771084336, "grad_norm": 0.23828125, "learning_rate": 5.983935742971888e-06, "loss": 0.8671, "step": 800 }, { "epoch": 1.2063253012048194, "grad_norm": 0.240234375, "learning_rate": 5.9789156626506025e-06, "loss": 0.8361, "step": 801 }, { "epoch": 1.2078313253012047, "grad_norm": 0.255859375, "learning_rate": 5.973895582329318e-06, "loss": 0.8561, "step": 802 }, { "epoch": 1.2093373493975903, "grad_norm": 0.47265625, "learning_rate": 5.968875502008032e-06, "loss": 0.8467, "step": 803 }, { "epoch": 1.2108433734939759, "grad_norm": 0.2294921875, "learning_rate": 5.963855421686747e-06, "loss": 0.8757, "step": 804 }, { "epoch": 1.2123493975903614, "grad_norm": 0.26953125, "learning_rate": 5.958835341365462e-06, "loss": 0.8217, "step": 805 }, { "epoch": 1.213855421686747, "grad_norm": 0.75390625, "learning_rate": 5.953815261044177e-06, "loss": 0.8377, "step": 806 }, { "epoch": 1.2153614457831325, "grad_norm": 0.2294921875, "learning_rate": 5.948795180722891e-06, "loss": 0.863, "step": 807 }, { "epoch": 1.216867469879518, "grad_norm": 0.23046875, "learning_rate": 5.943775100401607e-06, "loss": 0.8463, "step": 808 }, { "epoch": 1.2183734939759037, "grad_norm": 0.23828125, "learning_rate": 5.938755020080321e-06, "loss": 0.8121, "step": 809 }, { "epoch": 1.2198795180722892, "grad_norm": 0.2412109375, "learning_rate": 5.9337349397590375e-06, "loss": 0.8659, "step": 810 }, { "epoch": 1.2213855421686748, "grad_norm": 0.283203125, "learning_rate": 5.928714859437752e-06, "loss": 0.8658, "step": 811 }, { "epoch": 1.2228915662650603, "grad_norm": 0.23046875, "learning_rate": 5.9236947791164665e-06, "loss": 0.8608, "step": 812 }, { "epoch": 1.2243975903614457, "grad_norm": 0.224609375, "learning_rate": 5.918674698795182e-06, "loss": 0.8358, "step": 813 }, { "epoch": 1.2259036144578312, "grad_norm": 0.2333984375, "learning_rate": 5.9136546184738964e-06, "loss": 0.8393, "step": 814 }, { "epoch": 1.2274096385542168, "grad_norm": 0.2412109375, "learning_rate": 5.908634538152611e-06, "loss": 0.8877, "step": 815 }, { "epoch": 1.2289156626506024, "grad_norm": 0.265625, "learning_rate": 5.9036144578313255e-06, "loss": 0.8296, "step": 816 }, { "epoch": 1.230421686746988, "grad_norm": 0.2314453125, "learning_rate": 5.898594377510041e-06, "loss": 0.8034, "step": 817 }, { "epoch": 1.2319277108433735, "grad_norm": 0.2431640625, "learning_rate": 5.893574297188755e-06, "loss": 0.8549, "step": 818 }, { "epoch": 1.233433734939759, "grad_norm": 0.404296875, "learning_rate": 5.88855421686747e-06, "loss": 0.8685, "step": 819 }, { "epoch": 1.2349397590361446, "grad_norm": 0.2373046875, "learning_rate": 5.883534136546185e-06, "loss": 0.8858, "step": 820 }, { "epoch": 1.2364457831325302, "grad_norm": 0.232421875, "learning_rate": 5.8785140562249e-06, "loss": 0.8374, "step": 821 }, { "epoch": 1.2379518072289157, "grad_norm": 0.2431640625, "learning_rate": 5.873493975903614e-06, "loss": 0.8509, "step": 822 }, { "epoch": 1.2394578313253013, "grad_norm": 0.2431640625, "learning_rate": 5.86847389558233e-06, "loss": 0.8218, "step": 823 }, { "epoch": 1.2409638554216866, "grad_norm": 0.25390625, "learning_rate": 5.863453815261044e-06, "loss": 0.8491, "step": 824 }, { "epoch": 1.2424698795180722, "grad_norm": 0.23046875, "learning_rate": 5.858433734939759e-06, "loss": 0.8769, "step": 825 }, { "epoch": 1.2439759036144578, "grad_norm": 0.25, "learning_rate": 5.853413654618474e-06, "loss": 0.8749, "step": 826 }, { "epoch": 1.2454819277108433, "grad_norm": 0.23046875, "learning_rate": 5.848393574297189e-06, "loss": 0.8226, "step": 827 }, { "epoch": 1.2469879518072289, "grad_norm": 0.25390625, "learning_rate": 5.843373493975905e-06, "loss": 0.8611, "step": 828 }, { "epoch": 1.2484939759036144, "grad_norm": 0.2412109375, "learning_rate": 5.8383534136546194e-06, "loss": 0.8506, "step": 829 }, { "epoch": 1.25, "grad_norm": 0.251953125, "learning_rate": 5.833333333333334e-06, "loss": 0.8743, "step": 830 }, { "epoch": 1.2515060240963856, "grad_norm": 0.2294921875, "learning_rate": 5.828313253012049e-06, "loss": 0.8218, "step": 831 }, { "epoch": 1.2530120481927711, "grad_norm": 0.248046875, "learning_rate": 5.823293172690764e-06, "loss": 0.8428, "step": 832 }, { "epoch": 1.2545180722891567, "grad_norm": 0.2392578125, "learning_rate": 5.818273092369478e-06, "loss": 0.8641, "step": 833 }, { "epoch": 1.2560240963855422, "grad_norm": 0.234375, "learning_rate": 5.813253012048194e-06, "loss": 0.8411, "step": 834 }, { "epoch": 1.2575301204819276, "grad_norm": 0.380859375, "learning_rate": 5.808232931726908e-06, "loss": 0.835, "step": 835 }, { "epoch": 1.2590361445783134, "grad_norm": 0.232421875, "learning_rate": 5.803212851405623e-06, "loss": 0.8481, "step": 836 }, { "epoch": 1.2605421686746987, "grad_norm": 0.2470703125, "learning_rate": 5.798192771084338e-06, "loss": 0.8285, "step": 837 }, { "epoch": 1.2620481927710843, "grad_norm": 0.263671875, "learning_rate": 5.793172690763053e-06, "loss": 0.8458, "step": 838 }, { "epoch": 1.2635542168674698, "grad_norm": 0.25, "learning_rate": 5.788152610441767e-06, "loss": 0.8681, "step": 839 }, { "epoch": 1.2650602409638554, "grad_norm": 0.236328125, "learning_rate": 5.783132530120482e-06, "loss": 0.841, "step": 840 }, { "epoch": 1.266566265060241, "grad_norm": 0.2255859375, "learning_rate": 5.778112449799197e-06, "loss": 0.8141, "step": 841 }, { "epoch": 1.2680722891566265, "grad_norm": 0.294921875, "learning_rate": 5.773092369477912e-06, "loss": 0.8568, "step": 842 }, { "epoch": 1.269578313253012, "grad_norm": 0.251953125, "learning_rate": 5.768072289156626e-06, "loss": 0.8483, "step": 843 }, { "epoch": 1.2710843373493976, "grad_norm": 0.69140625, "learning_rate": 5.7630522088353416e-06, "loss": 0.8103, "step": 844 }, { "epoch": 1.2725903614457832, "grad_norm": 0.25, "learning_rate": 5.758032128514057e-06, "loss": 0.8145, "step": 845 }, { "epoch": 1.2740963855421688, "grad_norm": 0.251953125, "learning_rate": 5.753012048192772e-06, "loss": 0.8484, "step": 846 }, { "epoch": 1.2756024096385543, "grad_norm": 0.2392578125, "learning_rate": 5.747991967871487e-06, "loss": 0.8267, "step": 847 }, { "epoch": 1.2771084337349397, "grad_norm": 0.2314453125, "learning_rate": 5.742971887550201e-06, "loss": 0.8393, "step": 848 }, { "epoch": 1.2786144578313254, "grad_norm": 0.255859375, "learning_rate": 5.737951807228917e-06, "loss": 0.874, "step": 849 }, { "epoch": 1.2801204819277108, "grad_norm": 0.2451171875, "learning_rate": 5.732931726907631e-06, "loss": 0.8035, "step": 850 }, { "epoch": 1.2816265060240963, "grad_norm": 0.23828125, "learning_rate": 5.727911646586346e-06, "loss": 0.858, "step": 851 }, { "epoch": 1.283132530120482, "grad_norm": 0.2412109375, "learning_rate": 5.722891566265061e-06, "loss": 0.8165, "step": 852 }, { "epoch": 1.2846385542168675, "grad_norm": 0.32421875, "learning_rate": 5.717871485943776e-06, "loss": 0.8394, "step": 853 }, { "epoch": 1.286144578313253, "grad_norm": 0.27734375, "learning_rate": 5.71285140562249e-06, "loss": 0.8721, "step": 854 }, { "epoch": 1.2876506024096386, "grad_norm": 0.234375, "learning_rate": 5.707831325301206e-06, "loss": 0.8317, "step": 855 }, { "epoch": 1.2891566265060241, "grad_norm": 0.2490234375, "learning_rate": 5.70281124497992e-06, "loss": 0.8221, "step": 856 }, { "epoch": 1.2906626506024097, "grad_norm": 0.2373046875, "learning_rate": 5.697791164658635e-06, "loss": 0.8662, "step": 857 }, { "epoch": 1.2921686746987953, "grad_norm": 0.244140625, "learning_rate": 5.69277108433735e-06, "loss": 0.842, "step": 858 }, { "epoch": 1.2936746987951806, "grad_norm": 0.2470703125, "learning_rate": 5.6877510040160645e-06, "loss": 0.8869, "step": 859 }, { "epoch": 1.2951807228915664, "grad_norm": 0.23828125, "learning_rate": 5.682730923694779e-06, "loss": 0.8842, "step": 860 }, { "epoch": 1.2966867469879517, "grad_norm": 0.2392578125, "learning_rate": 5.677710843373494e-06, "loss": 0.8037, "step": 861 }, { "epoch": 1.2981927710843373, "grad_norm": 0.287109375, "learning_rate": 5.672690763052209e-06, "loss": 0.7978, "step": 862 }, { "epoch": 1.2996987951807228, "grad_norm": 0.236328125, "learning_rate": 5.667670682730924e-06, "loss": 0.8426, "step": 863 }, { "epoch": 1.3012048192771084, "grad_norm": 0.2578125, "learning_rate": 5.66265060240964e-06, "loss": 0.8504, "step": 864 }, { "epoch": 1.302710843373494, "grad_norm": 0.2578125, "learning_rate": 5.657630522088354e-06, "loss": 0.8392, "step": 865 }, { "epoch": 1.3042168674698795, "grad_norm": 0.2314453125, "learning_rate": 5.652610441767069e-06, "loss": 0.8135, "step": 866 }, { "epoch": 1.305722891566265, "grad_norm": 0.2314453125, "learning_rate": 5.647590361445784e-06, "loss": 0.829, "step": 867 }, { "epoch": 1.3072289156626506, "grad_norm": 0.2431640625, "learning_rate": 5.642570281124499e-06, "loss": 0.8268, "step": 868 }, { "epoch": 1.3087349397590362, "grad_norm": 0.2294921875, "learning_rate": 5.637550200803213e-06, "loss": 0.8066, "step": 869 }, { "epoch": 1.3102409638554218, "grad_norm": 0.25390625, "learning_rate": 5.6325301204819286e-06, "loss": 0.8302, "step": 870 }, { "epoch": 1.3117469879518073, "grad_norm": 0.291015625, "learning_rate": 5.627510040160643e-06, "loss": 0.8117, "step": 871 }, { "epoch": 1.3132530120481927, "grad_norm": 0.2265625, "learning_rate": 5.622489959839358e-06, "loss": 0.8242, "step": 872 }, { "epoch": 1.3147590361445782, "grad_norm": 0.28515625, "learning_rate": 5.617469879518073e-06, "loss": 0.8484, "step": 873 }, { "epoch": 1.3162650602409638, "grad_norm": 0.234375, "learning_rate": 5.6124497991967875e-06, "loss": 0.8384, "step": 874 }, { "epoch": 1.3177710843373494, "grad_norm": 0.55078125, "learning_rate": 5.607429718875502e-06, "loss": 0.8614, "step": 875 }, { "epoch": 1.319277108433735, "grad_norm": 0.2470703125, "learning_rate": 5.602409638554217e-06, "loss": 0.8209, "step": 876 }, { "epoch": 1.3207831325301205, "grad_norm": 0.333984375, "learning_rate": 5.597389558232932e-06, "loss": 0.8609, "step": 877 }, { "epoch": 1.322289156626506, "grad_norm": 0.283203125, "learning_rate": 5.5923694779116465e-06, "loss": 0.8559, "step": 878 }, { "epoch": 1.3237951807228916, "grad_norm": 0.283203125, "learning_rate": 5.587349397590362e-06, "loss": 0.8824, "step": 879 }, { "epoch": 1.3253012048192772, "grad_norm": 0.2451171875, "learning_rate": 5.582329317269076e-06, "loss": 0.8656, "step": 880 }, { "epoch": 1.3268072289156627, "grad_norm": 0.31640625, "learning_rate": 5.577309236947792e-06, "loss": 0.8086, "step": 881 }, { "epoch": 1.3283132530120483, "grad_norm": 0.236328125, "learning_rate": 5.572289156626507e-06, "loss": 0.8768, "step": 882 }, { "epoch": 1.3298192771084336, "grad_norm": 0.2431640625, "learning_rate": 5.567269076305222e-06, "loss": 0.8692, "step": 883 }, { "epoch": 1.3313253012048194, "grad_norm": 0.263671875, "learning_rate": 5.562248995983936e-06, "loss": 0.8424, "step": 884 }, { "epoch": 1.3328313253012047, "grad_norm": 0.2353515625, "learning_rate": 5.5572289156626515e-06, "loss": 0.8419, "step": 885 }, { "epoch": 1.3343373493975903, "grad_norm": 0.26953125, "learning_rate": 5.552208835341366e-06, "loss": 0.8398, "step": 886 }, { "epoch": 1.3358433734939759, "grad_norm": 0.2275390625, "learning_rate": 5.547188755020081e-06, "loss": 0.8373, "step": 887 }, { "epoch": 1.3373493975903614, "grad_norm": 0.302734375, "learning_rate": 5.542168674698796e-06, "loss": 0.8455, "step": 888 }, { "epoch": 1.338855421686747, "grad_norm": 0.232421875, "learning_rate": 5.5371485943775105e-06, "loss": 0.8202, "step": 889 }, { "epoch": 1.3403614457831325, "grad_norm": 0.2451171875, "learning_rate": 5.532128514056225e-06, "loss": 0.8622, "step": 890 }, { "epoch": 1.341867469879518, "grad_norm": 0.26953125, "learning_rate": 5.52710843373494e-06, "loss": 0.8298, "step": 891 }, { "epoch": 1.3433734939759037, "grad_norm": 0.2431640625, "learning_rate": 5.522088353413655e-06, "loss": 0.8223, "step": 892 }, { "epoch": 1.3448795180722892, "grad_norm": 0.2333984375, "learning_rate": 5.5170682730923695e-06, "loss": 0.8217, "step": 893 }, { "epoch": 1.3463855421686746, "grad_norm": 0.314453125, "learning_rate": 5.512048192771085e-06, "loss": 0.8559, "step": 894 }, { "epoch": 1.3478915662650603, "grad_norm": 0.27734375, "learning_rate": 5.507028112449799e-06, "loss": 0.8415, "step": 895 }, { "epoch": 1.3493975903614457, "grad_norm": 0.2451171875, "learning_rate": 5.502008032128514e-06, "loss": 0.8716, "step": 896 }, { "epoch": 1.3509036144578312, "grad_norm": 0.263671875, "learning_rate": 5.496987951807229e-06, "loss": 0.8324, "step": 897 }, { "epoch": 1.3524096385542168, "grad_norm": 0.59375, "learning_rate": 5.491967871485944e-06, "loss": 0.8764, "step": 898 }, { "epoch": 1.3539156626506024, "grad_norm": 0.2333984375, "learning_rate": 5.48694779116466e-06, "loss": 0.8827, "step": 899 }, { "epoch": 1.355421686746988, "grad_norm": 0.23828125, "learning_rate": 5.4819277108433745e-06, "loss": 0.8716, "step": 900 }, { "epoch": 1.3569277108433735, "grad_norm": 0.58984375, "learning_rate": 5.476907630522089e-06, "loss": 0.8721, "step": 901 }, { "epoch": 1.358433734939759, "grad_norm": 0.271484375, "learning_rate": 5.471887550200804e-06, "loss": 0.8707, "step": 902 }, { "epoch": 1.3599397590361446, "grad_norm": 0.251953125, "learning_rate": 5.466867469879519e-06, "loss": 0.8071, "step": 903 }, { "epoch": 1.3614457831325302, "grad_norm": 0.267578125, "learning_rate": 5.4618473895582335e-06, "loss": 0.8399, "step": 904 }, { "epoch": 1.3629518072289157, "grad_norm": 0.6328125, "learning_rate": 5.456827309236948e-06, "loss": 0.8178, "step": 905 }, { "epoch": 1.3644578313253013, "grad_norm": 0.240234375, "learning_rate": 5.451807228915663e-06, "loss": 0.8015, "step": 906 }, { "epoch": 1.3659638554216866, "grad_norm": 0.27734375, "learning_rate": 5.446787148594378e-06, "loss": 0.8218, "step": 907 }, { "epoch": 1.3674698795180724, "grad_norm": 0.244140625, "learning_rate": 5.4417670682730924e-06, "loss": 0.8154, "step": 908 }, { "epoch": 1.3689759036144578, "grad_norm": 0.251953125, "learning_rate": 5.436746987951808e-06, "loss": 0.8483, "step": 909 }, { "epoch": 1.3704819277108433, "grad_norm": 0.259765625, "learning_rate": 5.431726907630522e-06, "loss": 0.8413, "step": 910 }, { "epoch": 1.3719879518072289, "grad_norm": 0.251953125, "learning_rate": 5.426706827309237e-06, "loss": 0.8243, "step": 911 }, { "epoch": 1.3734939759036144, "grad_norm": 0.2314453125, "learning_rate": 5.421686746987952e-06, "loss": 0.7856, "step": 912 }, { "epoch": 1.375, "grad_norm": 0.265625, "learning_rate": 5.416666666666667e-06, "loss": 0.8588, "step": 913 }, { "epoch": 1.3765060240963856, "grad_norm": 0.271484375, "learning_rate": 5.411646586345381e-06, "loss": 0.8427, "step": 914 }, { "epoch": 1.3780120481927711, "grad_norm": 0.275390625, "learning_rate": 5.406626506024097e-06, "loss": 0.8648, "step": 915 }, { "epoch": 1.3795180722891567, "grad_norm": 0.337890625, "learning_rate": 5.401606425702812e-06, "loss": 0.7738, "step": 916 }, { "epoch": 1.3810240963855422, "grad_norm": 0.259765625, "learning_rate": 5.396586345381527e-06, "loss": 0.8763, "step": 917 }, { "epoch": 1.3825301204819276, "grad_norm": 0.298828125, "learning_rate": 5.391566265060242e-06, "loss": 0.8244, "step": 918 }, { "epoch": 1.3840361445783134, "grad_norm": 0.29296875, "learning_rate": 5.3865461847389565e-06, "loss": 0.8623, "step": 919 }, { "epoch": 1.3855421686746987, "grad_norm": 0.255859375, "learning_rate": 5.381526104417672e-06, "loss": 0.8437, "step": 920 }, { "epoch": 1.3870481927710843, "grad_norm": 0.248046875, "learning_rate": 5.376506024096386e-06, "loss": 0.8831, "step": 921 }, { "epoch": 1.3885542168674698, "grad_norm": 0.2431640625, "learning_rate": 5.371485943775101e-06, "loss": 0.8186, "step": 922 }, { "epoch": 1.3900602409638554, "grad_norm": 0.234375, "learning_rate": 5.366465863453815e-06, "loss": 0.7906, "step": 923 }, { "epoch": 1.391566265060241, "grad_norm": 0.2333984375, "learning_rate": 5.361445783132531e-06, "loss": 0.8376, "step": 924 }, { "epoch": 1.3930722891566265, "grad_norm": 0.26953125, "learning_rate": 5.356425702811245e-06, "loss": 0.8612, "step": 925 }, { "epoch": 1.394578313253012, "grad_norm": 0.255859375, "learning_rate": 5.35140562248996e-06, "loss": 0.8914, "step": 926 }, { "epoch": 1.3960843373493976, "grad_norm": 0.2490234375, "learning_rate": 5.346385542168675e-06, "loss": 0.827, "step": 927 }, { "epoch": 1.3975903614457832, "grad_norm": 0.30078125, "learning_rate": 5.34136546184739e-06, "loss": 0.8406, "step": 928 }, { "epoch": 1.3990963855421688, "grad_norm": 0.244140625, "learning_rate": 5.336345381526104e-06, "loss": 0.8533, "step": 929 }, { "epoch": 1.4006024096385543, "grad_norm": 0.306640625, "learning_rate": 5.33132530120482e-06, "loss": 0.8169, "step": 930 }, { "epoch": 1.4021084337349397, "grad_norm": 0.2373046875, "learning_rate": 5.326305220883534e-06, "loss": 0.8398, "step": 931 }, { "epoch": 1.4036144578313254, "grad_norm": 0.234375, "learning_rate": 5.321285140562249e-06, "loss": 0.8215, "step": 932 }, { "epoch": 1.4051204819277108, "grad_norm": 0.232421875, "learning_rate": 5.316265060240964e-06, "loss": 0.8158, "step": 933 }, { "epoch": 1.4066265060240963, "grad_norm": 0.2412109375, "learning_rate": 5.3112449799196794e-06, "loss": 0.8637, "step": 934 }, { "epoch": 1.408132530120482, "grad_norm": 0.2470703125, "learning_rate": 5.306224899598395e-06, "loss": 0.8555, "step": 935 }, { "epoch": 1.4096385542168675, "grad_norm": 0.236328125, "learning_rate": 5.301204819277109e-06, "loss": 0.8455, "step": 936 }, { "epoch": 1.411144578313253, "grad_norm": 0.2431640625, "learning_rate": 5.296184738955824e-06, "loss": 0.8383, "step": 937 }, { "epoch": 1.4126506024096386, "grad_norm": 0.263671875, "learning_rate": 5.291164658634539e-06, "loss": 0.8776, "step": 938 }, { "epoch": 1.4141566265060241, "grad_norm": 0.2294921875, "learning_rate": 5.286144578313254e-06, "loss": 0.8137, "step": 939 }, { "epoch": 1.4156626506024097, "grad_norm": 0.2421875, "learning_rate": 5.281124497991968e-06, "loss": 0.8439, "step": 940 }, { "epoch": 1.4171686746987953, "grad_norm": 0.251953125, "learning_rate": 5.276104417670684e-06, "loss": 0.8199, "step": 941 }, { "epoch": 1.4186746987951806, "grad_norm": 0.25390625, "learning_rate": 5.271084337349398e-06, "loss": 0.817, "step": 942 }, { "epoch": 1.4201807228915664, "grad_norm": 0.25, "learning_rate": 5.266064257028113e-06, "loss": 0.8804, "step": 943 }, { "epoch": 1.4216867469879517, "grad_norm": 0.31640625, "learning_rate": 5.261044176706827e-06, "loss": 0.8316, "step": 944 }, { "epoch": 1.4231927710843373, "grad_norm": 0.2490234375, "learning_rate": 5.256024096385543e-06, "loss": 0.818, "step": 945 }, { "epoch": 1.4246987951807228, "grad_norm": 0.2392578125, "learning_rate": 5.251004016064257e-06, "loss": 0.8288, "step": 946 }, { "epoch": 1.4262048192771084, "grad_norm": 0.30859375, "learning_rate": 5.245983935742972e-06, "loss": 0.8284, "step": 947 }, { "epoch": 1.427710843373494, "grad_norm": 0.251953125, "learning_rate": 5.240963855421687e-06, "loss": 0.8349, "step": 948 }, { "epoch": 1.4292168674698795, "grad_norm": 0.2490234375, "learning_rate": 5.2359437751004016e-06, "loss": 0.8478, "step": 949 }, { "epoch": 1.430722891566265, "grad_norm": 0.2392578125, "learning_rate": 5.230923694779116e-06, "loss": 0.8442, "step": 950 }, { "epoch": 1.4322289156626506, "grad_norm": 0.248046875, "learning_rate": 5.2259036144578315e-06, "loss": 0.8122, "step": 951 }, { "epoch": 1.4337349397590362, "grad_norm": 0.244140625, "learning_rate": 5.220883534136547e-06, "loss": 0.8505, "step": 952 }, { "epoch": 1.4352409638554218, "grad_norm": 0.2734375, "learning_rate": 5.215863453815262e-06, "loss": 0.8088, "step": 953 }, { "epoch": 1.4367469879518073, "grad_norm": 0.2431640625, "learning_rate": 5.210843373493977e-06, "loss": 0.8251, "step": 954 }, { "epoch": 1.4382530120481927, "grad_norm": 0.2294921875, "learning_rate": 5.205823293172691e-06, "loss": 0.8449, "step": 955 }, { "epoch": 1.4397590361445782, "grad_norm": 0.244140625, "learning_rate": 5.200803212851407e-06, "loss": 0.8492, "step": 956 }, { "epoch": 1.4412650602409638, "grad_norm": 0.26953125, "learning_rate": 5.195783132530121e-06, "loss": 0.8801, "step": 957 }, { "epoch": 1.4427710843373494, "grad_norm": 0.396484375, "learning_rate": 5.190763052208836e-06, "loss": 0.852, "step": 958 }, { "epoch": 1.444277108433735, "grad_norm": 0.30859375, "learning_rate": 5.185742971887551e-06, "loss": 0.8074, "step": 959 }, { "epoch": 1.4457831325301205, "grad_norm": 0.251953125, "learning_rate": 5.180722891566266e-06, "loss": 0.8454, "step": 960 }, { "epoch": 1.447289156626506, "grad_norm": 0.244140625, "learning_rate": 5.17570281124498e-06, "loss": 0.8177, "step": 961 }, { "epoch": 1.4487951807228916, "grad_norm": 0.25, "learning_rate": 5.1706827309236955e-06, "loss": 0.8323, "step": 962 }, { "epoch": 1.4503012048192772, "grad_norm": 0.58984375, "learning_rate": 5.16566265060241e-06, "loss": 0.8549, "step": 963 }, { "epoch": 1.4518072289156627, "grad_norm": 0.255859375, "learning_rate": 5.1606425702811245e-06, "loss": 0.8367, "step": 964 }, { "epoch": 1.4533132530120483, "grad_norm": 0.298828125, "learning_rate": 5.155622489959839e-06, "loss": 0.7785, "step": 965 }, { "epoch": 1.4548192771084336, "grad_norm": 0.2431640625, "learning_rate": 5.1506024096385544e-06, "loss": 0.8564, "step": 966 }, { "epoch": 1.4563253012048194, "grad_norm": 0.2412109375, "learning_rate": 5.145582329317269e-06, "loss": 0.8333, "step": 967 }, { "epoch": 1.4578313253012047, "grad_norm": 0.259765625, "learning_rate": 5.1405622489959835e-06, "loss": 0.8065, "step": 968 }, { "epoch": 1.4593373493975903, "grad_norm": 0.23828125, "learning_rate": 5.135542168674699e-06, "loss": 0.8548, "step": 969 }, { "epoch": 1.4608433734939759, "grad_norm": 0.25, "learning_rate": 5.130522088353414e-06, "loss": 0.8517, "step": 970 }, { "epoch": 1.4623493975903614, "grad_norm": 0.2431640625, "learning_rate": 5.12550200803213e-06, "loss": 0.8342, "step": 971 }, { "epoch": 1.463855421686747, "grad_norm": 0.2431640625, "learning_rate": 5.120481927710844e-06, "loss": 0.869, "step": 972 }, { "epoch": 1.4653614457831325, "grad_norm": 0.267578125, "learning_rate": 5.115461847389559e-06, "loss": 0.828, "step": 973 }, { "epoch": 1.466867469879518, "grad_norm": 0.2421875, "learning_rate": 5.110441767068274e-06, "loss": 0.8533, "step": 974 }, { "epoch": 1.4683734939759037, "grad_norm": 0.259765625, "learning_rate": 5.1054216867469886e-06, "loss": 0.8118, "step": 975 }, { "epoch": 1.4698795180722892, "grad_norm": 0.2353515625, "learning_rate": 5.100401606425703e-06, "loss": 0.8488, "step": 976 }, { "epoch": 1.4713855421686746, "grad_norm": 0.2451171875, "learning_rate": 5.0953815261044185e-06, "loss": 0.819, "step": 977 }, { "epoch": 1.4728915662650603, "grad_norm": 0.267578125, "learning_rate": 5.090361445783133e-06, "loss": 0.8435, "step": 978 }, { "epoch": 1.4743975903614457, "grad_norm": 0.2451171875, "learning_rate": 5.0853413654618475e-06, "loss": 0.8111, "step": 979 }, { "epoch": 1.4759036144578312, "grad_norm": 0.232421875, "learning_rate": 5.080321285140563e-06, "loss": 0.8299, "step": 980 }, { "epoch": 1.4774096385542168, "grad_norm": 0.26171875, "learning_rate": 5.0753012048192774e-06, "loss": 0.8327, "step": 981 }, { "epoch": 1.4789156626506024, "grad_norm": 0.6328125, "learning_rate": 5.070281124497992e-06, "loss": 0.8596, "step": 982 }, { "epoch": 1.480421686746988, "grad_norm": 0.28515625, "learning_rate": 5.065261044176707e-06, "loss": 0.8573, "step": 983 }, { "epoch": 1.4819277108433735, "grad_norm": 0.244140625, "learning_rate": 5.060240963855422e-06, "loss": 0.856, "step": 984 }, { "epoch": 1.483433734939759, "grad_norm": 0.265625, "learning_rate": 5.055220883534136e-06, "loss": 0.806, "step": 985 }, { "epoch": 1.4849397590361446, "grad_norm": 0.2470703125, "learning_rate": 5.050200803212852e-06, "loss": 0.864, "step": 986 }, { "epoch": 1.4864457831325302, "grad_norm": 0.2421875, "learning_rate": 5.045180722891566e-06, "loss": 0.824, "step": 987 }, { "epoch": 1.4879518072289157, "grad_norm": 0.251953125, "learning_rate": 5.040160642570282e-06, "loss": 0.8229, "step": 988 }, { "epoch": 1.4894578313253013, "grad_norm": 0.259765625, "learning_rate": 5.035140562248997e-06, "loss": 0.8537, "step": 989 }, { "epoch": 1.4909638554216866, "grad_norm": 0.263671875, "learning_rate": 5.0301204819277116e-06, "loss": 0.8216, "step": 990 }, { "epoch": 1.4924698795180724, "grad_norm": 0.2353515625, "learning_rate": 5.025100401606426e-06, "loss": 0.8213, "step": 991 }, { "epoch": 1.4939759036144578, "grad_norm": 0.34375, "learning_rate": 5.0200803212851415e-06, "loss": 0.8163, "step": 992 }, { "epoch": 1.4954819277108433, "grad_norm": 0.265625, "learning_rate": 5.015060240963856e-06, "loss": 0.8538, "step": 993 }, { "epoch": 1.4969879518072289, "grad_norm": 0.267578125, "learning_rate": 5.0100401606425705e-06, "loss": 0.8702, "step": 994 }, { "epoch": 1.4984939759036144, "grad_norm": 0.26171875, "learning_rate": 5.005020080321286e-06, "loss": 0.8593, "step": 995 }, { "epoch": 1.5, "grad_norm": 0.240234375, "learning_rate": 5e-06, "loss": 0.84, "step": 996 }, { "epoch": 1.5015060240963856, "grad_norm": 0.2490234375, "learning_rate": 4.994979919678715e-06, "loss": 0.8475, "step": 997 }, { "epoch": 1.5030120481927711, "grad_norm": 0.2578125, "learning_rate": 4.98995983935743e-06, "loss": 0.7963, "step": 998 }, { "epoch": 1.5045180722891565, "grad_norm": 0.2412109375, "learning_rate": 4.984939759036145e-06, "loss": 0.7867, "step": 999 }, { "epoch": 1.5060240963855422, "grad_norm": 0.2421875, "learning_rate": 4.979919678714859e-06, "loss": 0.8513, "step": 1000 }, { "epoch": 1.5075301204819276, "grad_norm": 0.255859375, "learning_rate": 4.974899598393575e-06, "loss": 0.8002, "step": 1001 }, { "epoch": 1.5090361445783134, "grad_norm": 0.2734375, "learning_rate": 4.96987951807229e-06, "loss": 0.8288, "step": 1002 }, { "epoch": 1.5105421686746987, "grad_norm": 0.2890625, "learning_rate": 4.964859437751005e-06, "loss": 0.8212, "step": 1003 }, { "epoch": 1.5120481927710845, "grad_norm": 0.2412109375, "learning_rate": 4.959839357429719e-06, "loss": 0.7893, "step": 1004 }, { "epoch": 1.5135542168674698, "grad_norm": 0.2392578125, "learning_rate": 4.9548192771084345e-06, "loss": 0.8347, "step": 1005 }, { "epoch": 1.5150602409638554, "grad_norm": 0.2392578125, "learning_rate": 4.949799196787149e-06, "loss": 0.8223, "step": 1006 }, { "epoch": 1.516566265060241, "grad_norm": 0.248046875, "learning_rate": 4.944779116465864e-06, "loss": 0.8488, "step": 1007 }, { "epoch": 1.5180722891566265, "grad_norm": 0.2421875, "learning_rate": 4.939759036144578e-06, "loss": 0.8362, "step": 1008 }, { "epoch": 1.519578313253012, "grad_norm": 0.2578125, "learning_rate": 4.9347389558232935e-06, "loss": 0.8536, "step": 1009 }, { "epoch": 1.5210843373493976, "grad_norm": 0.248046875, "learning_rate": 4.929718875502009e-06, "loss": 0.8487, "step": 1010 }, { "epoch": 1.5225903614457832, "grad_norm": 0.251953125, "learning_rate": 4.924698795180723e-06, "loss": 0.8375, "step": 1011 }, { "epoch": 1.5240963855421685, "grad_norm": 0.27734375, "learning_rate": 4.919678714859438e-06, "loss": 0.8547, "step": 1012 }, { "epoch": 1.5256024096385543, "grad_norm": 0.236328125, "learning_rate": 4.914658634538153e-06, "loss": 0.8479, "step": 1013 }, { "epoch": 1.5271084337349397, "grad_norm": 0.2421875, "learning_rate": 4.909638554216868e-06, "loss": 0.8189, "step": 1014 }, { "epoch": 1.5286144578313254, "grad_norm": 0.2412109375, "learning_rate": 4.904618473895582e-06, "loss": 0.8481, "step": 1015 }, { "epoch": 1.5301204819277108, "grad_norm": 0.2412109375, "learning_rate": 4.899598393574298e-06, "loss": 0.8144, "step": 1016 }, { "epoch": 1.5316265060240963, "grad_norm": 0.263671875, "learning_rate": 4.894578313253012e-06, "loss": 0.8518, "step": 1017 }, { "epoch": 1.533132530120482, "grad_norm": 0.271484375, "learning_rate": 4.889558232931727e-06, "loss": 0.8335, "step": 1018 }, { "epoch": 1.5346385542168675, "grad_norm": 0.267578125, "learning_rate": 4.884538152610442e-06, "loss": 0.8357, "step": 1019 }, { "epoch": 1.536144578313253, "grad_norm": 0.248046875, "learning_rate": 4.8795180722891575e-06, "loss": 0.8756, "step": 1020 }, { "epoch": 1.5376506024096386, "grad_norm": 0.2412109375, "learning_rate": 4.874497991967872e-06, "loss": 0.8797, "step": 1021 }, { "epoch": 1.5391566265060241, "grad_norm": 0.314453125, "learning_rate": 4.8694779116465866e-06, "loss": 0.8475, "step": 1022 }, { "epoch": 1.5406626506024095, "grad_norm": 0.259765625, "learning_rate": 4.864457831325302e-06, "loss": 0.8364, "step": 1023 }, { "epoch": 1.5421686746987953, "grad_norm": 0.25, "learning_rate": 4.8594377510040165e-06, "loss": 0.8457, "step": 1024 }, { "epoch": 1.5436746987951806, "grad_norm": 0.251953125, "learning_rate": 4.854417670682731e-06, "loss": 0.8461, "step": 1025 }, { "epoch": 1.5451807228915664, "grad_norm": 0.2392578125, "learning_rate": 4.849397590361446e-06, "loss": 0.8383, "step": 1026 }, { "epoch": 1.5466867469879517, "grad_norm": 0.2412109375, "learning_rate": 4.844377510040161e-06, "loss": 0.8078, "step": 1027 }, { "epoch": 1.5481927710843375, "grad_norm": 0.234375, "learning_rate": 4.839357429718876e-06, "loss": 0.8192, "step": 1028 }, { "epoch": 1.5496987951807228, "grad_norm": 0.240234375, "learning_rate": 4.834337349397591e-06, "loss": 0.8267, "step": 1029 }, { "epoch": 1.5512048192771084, "grad_norm": 0.2451171875, "learning_rate": 4.829317269076305e-06, "loss": 0.8251, "step": 1030 }, { "epoch": 1.552710843373494, "grad_norm": 0.2451171875, "learning_rate": 4.824297188755021e-06, "loss": 0.8659, "step": 1031 }, { "epoch": 1.5542168674698795, "grad_norm": 0.265625, "learning_rate": 4.819277108433735e-06, "loss": 0.8991, "step": 1032 }, { "epoch": 1.555722891566265, "grad_norm": 0.255859375, "learning_rate": 4.81425702811245e-06, "loss": 0.8173, "step": 1033 }, { "epoch": 1.5572289156626506, "grad_norm": 0.271484375, "learning_rate": 4.809236947791165e-06, "loss": 0.8336, "step": 1034 }, { "epoch": 1.5587349397590362, "grad_norm": 0.35546875, "learning_rate": 4.80421686746988e-06, "loss": 0.8123, "step": 1035 }, { "epoch": 1.5602409638554215, "grad_norm": 0.26953125, "learning_rate": 4.799196787148594e-06, "loss": 0.8617, "step": 1036 }, { "epoch": 1.5617469879518073, "grad_norm": 0.255859375, "learning_rate": 4.7941767068273095e-06, "loss": 0.8619, "step": 1037 }, { "epoch": 1.5632530120481927, "grad_norm": 0.267578125, "learning_rate": 4.789156626506025e-06, "loss": 0.8172, "step": 1038 }, { "epoch": 1.5647590361445785, "grad_norm": 0.361328125, "learning_rate": 4.7841365461847394e-06, "loss": 0.7919, "step": 1039 }, { "epoch": 1.5662650602409638, "grad_norm": 0.294921875, "learning_rate": 4.779116465863454e-06, "loss": 0.8036, "step": 1040 }, { "epoch": 1.5677710843373494, "grad_norm": 0.298828125, "learning_rate": 4.774096385542169e-06, "loss": 0.8559, "step": 1041 }, { "epoch": 1.569277108433735, "grad_norm": 0.248046875, "learning_rate": 4.769076305220884e-06, "loss": 0.8318, "step": 1042 }, { "epoch": 1.5707831325301205, "grad_norm": 0.2431640625, "learning_rate": 4.764056224899598e-06, "loss": 0.7974, "step": 1043 }, { "epoch": 1.572289156626506, "grad_norm": 0.2490234375, "learning_rate": 4.759036144578314e-06, "loss": 0.79, "step": 1044 }, { "epoch": 1.5737951807228916, "grad_norm": 0.24609375, "learning_rate": 4.754016064257029e-06, "loss": 0.8467, "step": 1045 }, { "epoch": 1.5753012048192772, "grad_norm": 0.259765625, "learning_rate": 4.748995983935744e-06, "loss": 0.8233, "step": 1046 }, { "epoch": 1.5768072289156625, "grad_norm": 0.265625, "learning_rate": 4.743975903614458e-06, "loss": 0.8335, "step": 1047 }, { "epoch": 1.5783132530120483, "grad_norm": 0.255859375, "learning_rate": 4.7389558232931736e-06, "loss": 0.841, "step": 1048 }, { "epoch": 1.5798192771084336, "grad_norm": 0.2451171875, "learning_rate": 4.733935742971888e-06, "loss": 0.8403, "step": 1049 }, { "epoch": 1.5813253012048194, "grad_norm": 0.275390625, "learning_rate": 4.728915662650603e-06, "loss": 0.8581, "step": 1050 }, { "epoch": 1.5828313253012047, "grad_norm": 0.25, "learning_rate": 4.723895582329317e-06, "loss": 0.8689, "step": 1051 }, { "epoch": 1.5843373493975905, "grad_norm": 0.296875, "learning_rate": 4.7188755020080325e-06, "loss": 0.8653, "step": 1052 }, { "epoch": 1.5858433734939759, "grad_norm": 0.306640625, "learning_rate": 4.713855421686747e-06, "loss": 0.8706, "step": 1053 }, { "epoch": 1.5873493975903614, "grad_norm": 0.2412109375, "learning_rate": 4.7088353413654624e-06, "loss": 0.815, "step": 1054 }, { "epoch": 1.588855421686747, "grad_norm": 0.2451171875, "learning_rate": 4.703815261044177e-06, "loss": 0.8176, "step": 1055 }, { "epoch": 1.5903614457831325, "grad_norm": 0.26953125, "learning_rate": 4.698795180722892e-06, "loss": 0.8527, "step": 1056 }, { "epoch": 1.591867469879518, "grad_norm": 0.251953125, "learning_rate": 4.693775100401607e-06, "loss": 0.8204, "step": 1057 }, { "epoch": 1.5933734939759037, "grad_norm": 0.2373046875, "learning_rate": 4.688755020080321e-06, "loss": 0.8433, "step": 1058 }, { "epoch": 1.5948795180722892, "grad_norm": 0.26171875, "learning_rate": 4.683734939759037e-06, "loss": 0.8504, "step": 1059 }, { "epoch": 1.5963855421686746, "grad_norm": 0.2431640625, "learning_rate": 4.678714859437751e-06, "loss": 0.826, "step": 1060 }, { "epoch": 1.5978915662650603, "grad_norm": 0.2431640625, "learning_rate": 4.673694779116466e-06, "loss": 0.8315, "step": 1061 }, { "epoch": 1.5993975903614457, "grad_norm": 0.435546875, "learning_rate": 4.668674698795181e-06, "loss": 0.8626, "step": 1062 }, { "epoch": 1.6009036144578315, "grad_norm": 0.248046875, "learning_rate": 4.6636546184738966e-06, "loss": 0.8512, "step": 1063 }, { "epoch": 1.6024096385542168, "grad_norm": 0.30859375, "learning_rate": 4.658634538152611e-06, "loss": 0.8396, "step": 1064 }, { "epoch": 1.6039156626506024, "grad_norm": 0.2490234375, "learning_rate": 4.653614457831326e-06, "loss": 0.8506, "step": 1065 }, { "epoch": 1.605421686746988, "grad_norm": 0.244140625, "learning_rate": 4.648594377510041e-06, "loss": 0.8385, "step": 1066 }, { "epoch": 1.6069277108433735, "grad_norm": 0.283203125, "learning_rate": 4.6435742971887555e-06, "loss": 0.8375, "step": 1067 }, { "epoch": 1.608433734939759, "grad_norm": 0.31640625, "learning_rate": 4.63855421686747e-06, "loss": 0.8321, "step": 1068 }, { "epoch": 1.6099397590361446, "grad_norm": 0.2431640625, "learning_rate": 4.633534136546185e-06, "loss": 0.8372, "step": 1069 }, { "epoch": 1.6114457831325302, "grad_norm": 0.251953125, "learning_rate": 4.6285140562249e-06, "loss": 0.8039, "step": 1070 }, { "epoch": 1.6129518072289155, "grad_norm": 0.23828125, "learning_rate": 4.6234939759036145e-06, "loss": 0.8069, "step": 1071 }, { "epoch": 1.6144578313253013, "grad_norm": 0.2431640625, "learning_rate": 4.61847389558233e-06, "loss": 0.8253, "step": 1072 }, { "epoch": 1.6159638554216866, "grad_norm": 0.2470703125, "learning_rate": 4.613453815261044e-06, "loss": 0.8381, "step": 1073 }, { "epoch": 1.6174698795180724, "grad_norm": 0.259765625, "learning_rate": 4.60843373493976e-06, "loss": 0.8577, "step": 1074 }, { "epoch": 1.6189759036144578, "grad_norm": 0.259765625, "learning_rate": 4.603413654618474e-06, "loss": 0.8593, "step": 1075 }, { "epoch": 1.6204819277108435, "grad_norm": 0.263671875, "learning_rate": 4.598393574297189e-06, "loss": 0.8296, "step": 1076 }, { "epoch": 1.6219879518072289, "grad_norm": 0.24609375, "learning_rate": 4.593373493975904e-06, "loss": 0.8405, "step": 1077 }, { "epoch": 1.6234939759036144, "grad_norm": 0.25, "learning_rate": 4.588353413654619e-06, "loss": 0.8307, "step": 1078 }, { "epoch": 1.625, "grad_norm": 0.2490234375, "learning_rate": 4.583333333333333e-06, "loss": 0.861, "step": 1079 }, { "epoch": 1.6265060240963856, "grad_norm": 0.2412109375, "learning_rate": 4.578313253012049e-06, "loss": 0.8103, "step": 1080 }, { "epoch": 1.6280120481927711, "grad_norm": 0.2451171875, "learning_rate": 4.573293172690764e-06, "loss": 0.8273, "step": 1081 }, { "epoch": 1.6295180722891565, "grad_norm": 0.248046875, "learning_rate": 4.5682730923694785e-06, "loss": 0.865, "step": 1082 }, { "epoch": 1.6310240963855422, "grad_norm": 0.2421875, "learning_rate": 4.563253012048193e-06, "loss": 0.8671, "step": 1083 }, { "epoch": 1.6325301204819276, "grad_norm": 0.2431640625, "learning_rate": 4.558232931726908e-06, "loss": 0.8694, "step": 1084 }, { "epoch": 1.6340361445783134, "grad_norm": 0.263671875, "learning_rate": 4.553212851405623e-06, "loss": 0.8569, "step": 1085 }, { "epoch": 1.6355421686746987, "grad_norm": 0.2470703125, "learning_rate": 4.5481927710843374e-06, "loss": 0.8425, "step": 1086 }, { "epoch": 1.6370481927710845, "grad_norm": 0.234375, "learning_rate": 4.543172690763053e-06, "loss": 0.8194, "step": 1087 }, { "epoch": 1.6385542168674698, "grad_norm": 0.244140625, "learning_rate": 4.538152610441767e-06, "loss": 0.8486, "step": 1088 }, { "epoch": 1.6400602409638554, "grad_norm": 0.265625, "learning_rate": 4.533132530120482e-06, "loss": 0.8194, "step": 1089 }, { "epoch": 1.641566265060241, "grad_norm": 0.236328125, "learning_rate": 4.528112449799197e-06, "loss": 0.8097, "step": 1090 }, { "epoch": 1.6430722891566265, "grad_norm": 0.251953125, "learning_rate": 4.523092369477913e-06, "loss": 0.8384, "step": 1091 }, { "epoch": 1.644578313253012, "grad_norm": 0.330078125, "learning_rate": 4.518072289156627e-06, "loss": 0.8274, "step": 1092 }, { "epoch": 1.6460843373493976, "grad_norm": 0.40625, "learning_rate": 4.513052208835342e-06, "loss": 0.8259, "step": 1093 }, { "epoch": 1.6475903614457832, "grad_norm": 0.25390625, "learning_rate": 4.508032128514056e-06, "loss": 0.8416, "step": 1094 }, { "epoch": 1.6490963855421685, "grad_norm": 0.25390625, "learning_rate": 4.5030120481927716e-06, "loss": 0.8419, "step": 1095 }, { "epoch": 1.6506024096385543, "grad_norm": 0.353515625, "learning_rate": 4.497991967871486e-06, "loss": 0.8516, "step": 1096 }, { "epoch": 1.6521084337349397, "grad_norm": 0.244140625, "learning_rate": 4.492971887550201e-06, "loss": 0.8404, "step": 1097 }, { "epoch": 1.6536144578313254, "grad_norm": 0.388671875, "learning_rate": 4.487951807228916e-06, "loss": 0.8662, "step": 1098 }, { "epoch": 1.6551204819277108, "grad_norm": 0.291015625, "learning_rate": 4.482931726907631e-06, "loss": 0.848, "step": 1099 }, { "epoch": 1.6566265060240963, "grad_norm": 0.248046875, "learning_rate": 4.477911646586346e-06, "loss": 0.8031, "step": 1100 }, { "epoch": 1.658132530120482, "grad_norm": 0.259765625, "learning_rate": 4.47289156626506e-06, "loss": 0.7839, "step": 1101 }, { "epoch": 1.6596385542168675, "grad_norm": 0.267578125, "learning_rate": 4.467871485943776e-06, "loss": 0.8417, "step": 1102 }, { "epoch": 1.661144578313253, "grad_norm": 0.25, "learning_rate": 4.46285140562249e-06, "loss": 0.8692, "step": 1103 }, { "epoch": 1.6626506024096386, "grad_norm": 0.259765625, "learning_rate": 4.457831325301205e-06, "loss": 0.8461, "step": 1104 }, { "epoch": 1.6641566265060241, "grad_norm": 0.451171875, "learning_rate": 4.45281124497992e-06, "loss": 0.8363, "step": 1105 }, { "epoch": 1.6656626506024095, "grad_norm": 0.2373046875, "learning_rate": 4.447791164658635e-06, "loss": 0.7898, "step": 1106 }, { "epoch": 1.6671686746987953, "grad_norm": 0.244140625, "learning_rate": 4.442771084337349e-06, "loss": 0.8118, "step": 1107 }, { "epoch": 1.6686746987951806, "grad_norm": 0.24609375, "learning_rate": 4.437751004016065e-06, "loss": 0.8024, "step": 1108 }, { "epoch": 1.6701807228915664, "grad_norm": 0.240234375, "learning_rate": 4.43273092369478e-06, "loss": 0.8303, "step": 1109 }, { "epoch": 1.6716867469879517, "grad_norm": 0.2451171875, "learning_rate": 4.4277108433734945e-06, "loss": 0.8321, "step": 1110 }, { "epoch": 1.6731927710843375, "grad_norm": 0.26953125, "learning_rate": 4.422690763052209e-06, "loss": 0.8086, "step": 1111 }, { "epoch": 1.6746987951807228, "grad_norm": 0.25, "learning_rate": 4.4176706827309244e-06, "loss": 0.8354, "step": 1112 }, { "epoch": 1.6762048192771084, "grad_norm": 1.7109375, "learning_rate": 4.412650602409639e-06, "loss": 0.8468, "step": 1113 }, { "epoch": 1.677710843373494, "grad_norm": 0.28125, "learning_rate": 4.4076305220883535e-06, "loss": 0.8348, "step": 1114 }, { "epoch": 1.6792168674698795, "grad_norm": 0.255859375, "learning_rate": 4.402610441767068e-06, "loss": 0.8157, "step": 1115 }, { "epoch": 1.680722891566265, "grad_norm": 0.255859375, "learning_rate": 4.397590361445783e-06, "loss": 0.8453, "step": 1116 }, { "epoch": 1.6822289156626506, "grad_norm": 0.2421875, "learning_rate": 4.392570281124499e-06, "loss": 0.8114, "step": 1117 }, { "epoch": 1.6837349397590362, "grad_norm": 0.26171875, "learning_rate": 4.387550200803213e-06, "loss": 0.8312, "step": 1118 }, { "epoch": 1.6852409638554215, "grad_norm": 0.25390625, "learning_rate": 4.382530120481928e-06, "loss": 0.8577, "step": 1119 }, { "epoch": 1.6867469879518073, "grad_norm": 0.275390625, "learning_rate": 4.377510040160643e-06, "loss": 0.8542, "step": 1120 }, { "epoch": 1.6882530120481927, "grad_norm": 0.26953125, "learning_rate": 4.372489959839358e-06, "loss": 0.8222, "step": 1121 }, { "epoch": 1.6897590361445785, "grad_norm": 0.251953125, "learning_rate": 4.367469879518072e-06, "loss": 0.8364, "step": 1122 }, { "epoch": 1.6912650602409638, "grad_norm": 0.251953125, "learning_rate": 4.362449799196788e-06, "loss": 0.8294, "step": 1123 }, { "epoch": 1.6927710843373494, "grad_norm": 0.2490234375, "learning_rate": 4.357429718875502e-06, "loss": 0.7772, "step": 1124 }, { "epoch": 1.694277108433735, "grad_norm": 0.240234375, "learning_rate": 4.3524096385542175e-06, "loss": 0.8393, "step": 1125 }, { "epoch": 1.6957831325301205, "grad_norm": 0.25390625, "learning_rate": 4.347389558232932e-06, "loss": 0.8308, "step": 1126 }, { "epoch": 1.697289156626506, "grad_norm": 0.23828125, "learning_rate": 4.3423694779116474e-06, "loss": 0.8126, "step": 1127 }, { "epoch": 1.6987951807228916, "grad_norm": 0.34375, "learning_rate": 4.337349397590362e-06, "loss": 0.8223, "step": 1128 }, { "epoch": 1.7003012048192772, "grad_norm": 0.2490234375, "learning_rate": 4.3323293172690765e-06, "loss": 0.8211, "step": 1129 }, { "epoch": 1.7018072289156625, "grad_norm": 0.333984375, "learning_rate": 4.327309236947792e-06, "loss": 0.7877, "step": 1130 }, { "epoch": 1.7033132530120483, "grad_norm": 0.26171875, "learning_rate": 4.322289156626506e-06, "loss": 0.8472, "step": 1131 }, { "epoch": 1.7048192771084336, "grad_norm": 0.2578125, "learning_rate": 4.317269076305221e-06, "loss": 0.8403, "step": 1132 }, { "epoch": 1.7063253012048194, "grad_norm": 0.3046875, "learning_rate": 4.312248995983936e-06, "loss": 0.8178, "step": 1133 }, { "epoch": 1.7078313253012047, "grad_norm": 0.24609375, "learning_rate": 4.307228915662651e-06, "loss": 0.8694, "step": 1134 }, { "epoch": 1.7093373493975905, "grad_norm": 0.25, "learning_rate": 4.302208835341366e-06, "loss": 0.8395, "step": 1135 }, { "epoch": 1.7108433734939759, "grad_norm": 0.23828125, "learning_rate": 4.297188755020081e-06, "loss": 0.8197, "step": 1136 }, { "epoch": 1.7123493975903614, "grad_norm": 0.248046875, "learning_rate": 4.292168674698795e-06, "loss": 0.8359, "step": 1137 }, { "epoch": 1.713855421686747, "grad_norm": 0.25, "learning_rate": 4.287148594377511e-06, "loss": 0.8918, "step": 1138 }, { "epoch": 1.7153614457831325, "grad_norm": 0.2421875, "learning_rate": 4.282128514056225e-06, "loss": 0.8438, "step": 1139 }, { "epoch": 1.716867469879518, "grad_norm": 0.2421875, "learning_rate": 4.27710843373494e-06, "loss": 0.8341, "step": 1140 }, { "epoch": 1.7183734939759037, "grad_norm": 0.244140625, "learning_rate": 4.272088353413655e-06, "loss": 0.8624, "step": 1141 }, { "epoch": 1.7198795180722892, "grad_norm": 0.2490234375, "learning_rate": 4.2670682730923696e-06, "loss": 0.8309, "step": 1142 }, { "epoch": 1.7213855421686746, "grad_norm": 0.283203125, "learning_rate": 4.262048192771085e-06, "loss": 0.8072, "step": 1143 }, { "epoch": 1.7228915662650603, "grad_norm": 0.275390625, "learning_rate": 4.2570281124497995e-06, "loss": 0.8504, "step": 1144 }, { "epoch": 1.7243975903614457, "grad_norm": 0.2470703125, "learning_rate": 4.252008032128515e-06, "loss": 0.8202, "step": 1145 }, { "epoch": 1.7259036144578315, "grad_norm": 0.236328125, "learning_rate": 4.246987951807229e-06, "loss": 0.8086, "step": 1146 }, { "epoch": 1.7274096385542168, "grad_norm": 0.625, "learning_rate": 4.241967871485944e-06, "loss": 0.8267, "step": 1147 }, { "epoch": 1.7289156626506024, "grad_norm": 0.259765625, "learning_rate": 4.236947791164659e-06, "loss": 0.7918, "step": 1148 }, { "epoch": 1.730421686746988, "grad_norm": 0.283203125, "learning_rate": 4.231927710843374e-06, "loss": 0.8269, "step": 1149 }, { "epoch": 1.7319277108433735, "grad_norm": 0.25, "learning_rate": 4.226907630522088e-06, "loss": 0.8417, "step": 1150 }, { "epoch": 1.733433734939759, "grad_norm": 0.251953125, "learning_rate": 4.221887550200804e-06, "loss": 0.8886, "step": 1151 }, { "epoch": 1.7349397590361446, "grad_norm": 0.25, "learning_rate": 4.216867469879519e-06, "loss": 0.8759, "step": 1152 }, { "epoch": 1.7364457831325302, "grad_norm": 0.2578125, "learning_rate": 4.211847389558234e-06, "loss": 0.8581, "step": 1153 }, { "epoch": 1.7379518072289155, "grad_norm": 0.25, "learning_rate": 4.206827309236948e-06, "loss": 0.831, "step": 1154 }, { "epoch": 1.7394578313253013, "grad_norm": 0.234375, "learning_rate": 4.201807228915663e-06, "loss": 0.7943, "step": 1155 }, { "epoch": 1.7409638554216866, "grad_norm": 0.244140625, "learning_rate": 4.196787148594378e-06, "loss": 0.8342, "step": 1156 }, { "epoch": 1.7424698795180724, "grad_norm": 0.259765625, "learning_rate": 4.1917670682730925e-06, "loss": 0.8298, "step": 1157 }, { "epoch": 1.7439759036144578, "grad_norm": 0.2490234375, "learning_rate": 4.186746987951807e-06, "loss": 0.8994, "step": 1158 }, { "epoch": 1.7454819277108435, "grad_norm": 0.46484375, "learning_rate": 4.1817269076305224e-06, "loss": 0.8625, "step": 1159 }, { "epoch": 1.7469879518072289, "grad_norm": 0.251953125, "learning_rate": 4.176706827309237e-06, "loss": 0.825, "step": 1160 }, { "epoch": 1.7484939759036144, "grad_norm": 0.298828125, "learning_rate": 4.171686746987952e-06, "loss": 0.8694, "step": 1161 }, { "epoch": 1.75, "grad_norm": 0.294921875, "learning_rate": 4.166666666666667e-06, "loss": 0.8062, "step": 1162 }, { "epoch": 1.7515060240963856, "grad_norm": 0.2353515625, "learning_rate": 4.161646586345382e-06, "loss": 0.8184, "step": 1163 }, { "epoch": 1.7530120481927711, "grad_norm": 0.279296875, "learning_rate": 4.156626506024097e-06, "loss": 0.8661, "step": 1164 }, { "epoch": 1.7545180722891565, "grad_norm": 0.25, "learning_rate": 4.151606425702811e-06, "loss": 0.8433, "step": 1165 }, { "epoch": 1.7560240963855422, "grad_norm": 0.2421875, "learning_rate": 4.146586345381527e-06, "loss": 0.8462, "step": 1166 }, { "epoch": 1.7575301204819276, "grad_norm": 0.26171875, "learning_rate": 4.141566265060241e-06, "loss": 0.8262, "step": 1167 }, { "epoch": 1.7590361445783134, "grad_norm": 0.279296875, "learning_rate": 4.136546184738956e-06, "loss": 0.8664, "step": 1168 }, { "epoch": 1.7605421686746987, "grad_norm": 0.2412109375, "learning_rate": 4.131526104417671e-06, "loss": 0.7963, "step": 1169 }, { "epoch": 1.7620481927710845, "grad_norm": 0.37109375, "learning_rate": 4.1265060240963865e-06, "loss": 0.8152, "step": 1170 }, { "epoch": 1.7635542168674698, "grad_norm": 0.2421875, "learning_rate": 4.121485943775101e-06, "loss": 0.8406, "step": 1171 }, { "epoch": 1.7650602409638554, "grad_norm": 0.24609375, "learning_rate": 4.1164658634538155e-06, "loss": 0.8281, "step": 1172 }, { "epoch": 1.766566265060241, "grad_norm": 0.28125, "learning_rate": 4.111445783132531e-06, "loss": 0.8513, "step": 1173 }, { "epoch": 1.7680722891566265, "grad_norm": 0.255859375, "learning_rate": 4.106425702811245e-06, "loss": 0.8063, "step": 1174 }, { "epoch": 1.769578313253012, "grad_norm": 0.251953125, "learning_rate": 4.10140562248996e-06, "loss": 0.8547, "step": 1175 }, { "epoch": 1.7710843373493976, "grad_norm": 0.251953125, "learning_rate": 4.096385542168675e-06, "loss": 0.8296, "step": 1176 }, { "epoch": 1.7725903614457832, "grad_norm": 0.275390625, "learning_rate": 4.09136546184739e-06, "loss": 0.8005, "step": 1177 }, { "epoch": 1.7740963855421685, "grad_norm": 0.2470703125, "learning_rate": 4.086345381526104e-06, "loss": 0.8157, "step": 1178 }, { "epoch": 1.7756024096385543, "grad_norm": 0.2578125, "learning_rate": 4.08132530120482e-06, "loss": 0.8326, "step": 1179 }, { "epoch": 1.7771084337349397, "grad_norm": 0.2734375, "learning_rate": 4.076305220883534e-06, "loss": 0.8409, "step": 1180 }, { "epoch": 1.7786144578313254, "grad_norm": 0.2431640625, "learning_rate": 4.07128514056225e-06, "loss": 0.8284, "step": 1181 }, { "epoch": 1.7801204819277108, "grad_norm": 0.2431640625, "learning_rate": 4.066265060240964e-06, "loss": 0.8712, "step": 1182 }, { "epoch": 1.7816265060240963, "grad_norm": 0.2490234375, "learning_rate": 4.061244979919679e-06, "loss": 0.8505, "step": 1183 }, { "epoch": 1.783132530120482, "grad_norm": 0.2490234375, "learning_rate": 4.056224899598394e-06, "loss": 0.8428, "step": 1184 }, { "epoch": 1.7846385542168675, "grad_norm": 0.2431640625, "learning_rate": 4.051204819277109e-06, "loss": 0.8004, "step": 1185 }, { "epoch": 1.786144578313253, "grad_norm": 0.458984375, "learning_rate": 4.046184738955823e-06, "loss": 0.7801, "step": 1186 }, { "epoch": 1.7876506024096386, "grad_norm": 0.236328125, "learning_rate": 4.0411646586345385e-06, "loss": 0.8512, "step": 1187 }, { "epoch": 1.7891566265060241, "grad_norm": 0.30078125, "learning_rate": 4.036144578313254e-06, "loss": 0.8197, "step": 1188 }, { "epoch": 1.7906626506024095, "grad_norm": 0.2421875, "learning_rate": 4.031124497991968e-06, "loss": 0.8148, "step": 1189 }, { "epoch": 1.7921686746987953, "grad_norm": 0.26171875, "learning_rate": 4.026104417670683e-06, "loss": 0.8221, "step": 1190 }, { "epoch": 1.7936746987951806, "grad_norm": 0.24609375, "learning_rate": 4.021084337349398e-06, "loss": 0.8351, "step": 1191 }, { "epoch": 1.7951807228915664, "grad_norm": 0.265625, "learning_rate": 4.016064257028113e-06, "loss": 0.8378, "step": 1192 }, { "epoch": 1.7966867469879517, "grad_norm": 0.244140625, "learning_rate": 4.011044176706827e-06, "loss": 0.7942, "step": 1193 }, { "epoch": 1.7981927710843375, "grad_norm": 0.26171875, "learning_rate": 4.006024096385543e-06, "loss": 0.8621, "step": 1194 }, { "epoch": 1.7996987951807228, "grad_norm": 0.26171875, "learning_rate": 4.001004016064257e-06, "loss": 0.8391, "step": 1195 }, { "epoch": 1.8012048192771084, "grad_norm": 0.259765625, "learning_rate": 3.995983935742972e-06, "loss": 0.8459, "step": 1196 }, { "epoch": 1.802710843373494, "grad_norm": 0.244140625, "learning_rate": 3.990963855421687e-06, "loss": 0.8199, "step": 1197 }, { "epoch": 1.8042168674698795, "grad_norm": 0.263671875, "learning_rate": 3.985943775100402e-06, "loss": 0.811, "step": 1198 }, { "epoch": 1.805722891566265, "grad_norm": 0.255859375, "learning_rate": 3.980923694779117e-06, "loss": 0.8212, "step": 1199 }, { "epoch": 1.8072289156626506, "grad_norm": 0.25, "learning_rate": 3.975903614457832e-06, "loss": 0.8371, "step": 1200 }, { "epoch": 1.8087349397590362, "grad_norm": 0.400390625, "learning_rate": 3.970883534136546e-06, "loss": 0.8311, "step": 1201 }, { "epoch": 1.8102409638554215, "grad_norm": 0.248046875, "learning_rate": 3.9658634538152615e-06, "loss": 0.8354, "step": 1202 }, { "epoch": 1.8117469879518073, "grad_norm": 0.26171875, "learning_rate": 3.960843373493976e-06, "loss": 0.8311, "step": 1203 }, { "epoch": 1.8132530120481927, "grad_norm": 0.267578125, "learning_rate": 3.9558232931726905e-06, "loss": 0.839, "step": 1204 }, { "epoch": 1.8147590361445785, "grad_norm": 0.3984375, "learning_rate": 3.950803212851406e-06, "loss": 0.8616, "step": 1205 }, { "epoch": 1.8162650602409638, "grad_norm": 0.25390625, "learning_rate": 3.945783132530121e-06, "loss": 0.8787, "step": 1206 }, { "epoch": 1.8177710843373494, "grad_norm": 0.24609375, "learning_rate": 3.940763052208836e-06, "loss": 0.8134, "step": 1207 }, { "epoch": 1.819277108433735, "grad_norm": 0.255859375, "learning_rate": 3.93574297188755e-06, "loss": 0.8187, "step": 1208 }, { "epoch": 1.8207831325301205, "grad_norm": 0.271484375, "learning_rate": 3.930722891566266e-06, "loss": 0.8037, "step": 1209 }, { "epoch": 1.822289156626506, "grad_norm": 0.255859375, "learning_rate": 3.92570281124498e-06, "loss": 0.8812, "step": 1210 }, { "epoch": 1.8237951807228916, "grad_norm": 0.275390625, "learning_rate": 3.920682730923695e-06, "loss": 0.8421, "step": 1211 }, { "epoch": 1.8253012048192772, "grad_norm": 0.251953125, "learning_rate": 3.91566265060241e-06, "loss": 0.8574, "step": 1212 }, { "epoch": 1.8268072289156625, "grad_norm": 0.2490234375, "learning_rate": 3.910642570281125e-06, "loss": 0.8229, "step": 1213 }, { "epoch": 1.8283132530120483, "grad_norm": 0.2421875, "learning_rate": 3.90562248995984e-06, "loss": 0.8348, "step": 1214 }, { "epoch": 1.8298192771084336, "grad_norm": 0.267578125, "learning_rate": 3.9006024096385546e-06, "loss": 0.8645, "step": 1215 }, { "epoch": 1.8313253012048194, "grad_norm": 0.384765625, "learning_rate": 3.89558232931727e-06, "loss": 0.8189, "step": 1216 }, { "epoch": 1.8328313253012047, "grad_norm": 0.341796875, "learning_rate": 3.8905622489959845e-06, "loss": 0.8252, "step": 1217 }, { "epoch": 1.8343373493975905, "grad_norm": 0.265625, "learning_rate": 3.885542168674699e-06, "loss": 0.8147, "step": 1218 }, { "epoch": 1.8358433734939759, "grad_norm": 0.2412109375, "learning_rate": 3.8805220883534135e-06, "loss": 0.8303, "step": 1219 }, { "epoch": 1.8373493975903614, "grad_norm": 0.2578125, "learning_rate": 3.875502008032129e-06, "loss": 0.8398, "step": 1220 }, { "epoch": 1.838855421686747, "grad_norm": 0.2578125, "learning_rate": 3.870481927710843e-06, "loss": 0.8297, "step": 1221 }, { "epoch": 1.8403614457831325, "grad_norm": 0.2421875, "learning_rate": 3.865461847389558e-06, "loss": 0.8509, "step": 1222 }, { "epoch": 1.841867469879518, "grad_norm": 0.255859375, "learning_rate": 3.860441767068273e-06, "loss": 0.8157, "step": 1223 }, { "epoch": 1.8433734939759037, "grad_norm": 0.2421875, "learning_rate": 3.855421686746989e-06, "loss": 0.8403, "step": 1224 }, { "epoch": 1.8448795180722892, "grad_norm": 0.248046875, "learning_rate": 3.850401606425703e-06, "loss": 0.8228, "step": 1225 }, { "epoch": 1.8463855421686746, "grad_norm": 0.27734375, "learning_rate": 3.845381526104418e-06, "loss": 0.8486, "step": 1226 }, { "epoch": 1.8478915662650603, "grad_norm": 0.97265625, "learning_rate": 3.840361445783133e-06, "loss": 0.7986, "step": 1227 }, { "epoch": 1.8493975903614457, "grad_norm": 0.29296875, "learning_rate": 3.835341365461848e-06, "loss": 0.8566, "step": 1228 }, { "epoch": 1.8509036144578315, "grad_norm": 0.291015625, "learning_rate": 3.830321285140562e-06, "loss": 0.8394, "step": 1229 }, { "epoch": 1.8524096385542168, "grad_norm": 0.2421875, "learning_rate": 3.8253012048192775e-06, "loss": 0.8427, "step": 1230 }, { "epoch": 1.8539156626506024, "grad_norm": 0.25, "learning_rate": 3.820281124497992e-06, "loss": 0.8763, "step": 1231 }, { "epoch": 1.855421686746988, "grad_norm": 0.25390625, "learning_rate": 3.8152610441767074e-06, "loss": 0.8084, "step": 1232 }, { "epoch": 1.8569277108433735, "grad_norm": 0.296875, "learning_rate": 3.8102409638554224e-06, "loss": 0.8208, "step": 1233 }, { "epoch": 1.858433734939759, "grad_norm": 0.26171875, "learning_rate": 3.805220883534137e-06, "loss": 0.8636, "step": 1234 }, { "epoch": 1.8599397590361446, "grad_norm": 0.244140625, "learning_rate": 3.800200803212852e-06, "loss": 0.8129, "step": 1235 }, { "epoch": 1.8614457831325302, "grad_norm": 0.326171875, "learning_rate": 3.7951807228915664e-06, "loss": 0.8492, "step": 1236 }, { "epoch": 1.8629518072289155, "grad_norm": 0.25390625, "learning_rate": 3.7901606425702813e-06, "loss": 0.8217, "step": 1237 }, { "epoch": 1.8644578313253013, "grad_norm": 0.265625, "learning_rate": 3.7851405622489963e-06, "loss": 0.8472, "step": 1238 }, { "epoch": 1.8659638554216866, "grad_norm": 0.248046875, "learning_rate": 3.780120481927711e-06, "loss": 0.8314, "step": 1239 }, { "epoch": 1.8674698795180724, "grad_norm": 0.248046875, "learning_rate": 3.7751004016064258e-06, "loss": 0.8633, "step": 1240 }, { "epoch": 1.8689759036144578, "grad_norm": 0.251953125, "learning_rate": 3.770080321285141e-06, "loss": 0.8082, "step": 1241 }, { "epoch": 1.8704819277108435, "grad_norm": 0.2890625, "learning_rate": 3.765060240963856e-06, "loss": 0.8496, "step": 1242 }, { "epoch": 1.8719879518072289, "grad_norm": 0.2431640625, "learning_rate": 3.7600401606425706e-06, "loss": 0.7969, "step": 1243 }, { "epoch": 1.8734939759036144, "grad_norm": 0.248046875, "learning_rate": 3.7550200803212856e-06, "loss": 0.8643, "step": 1244 }, { "epoch": 1.875, "grad_norm": 0.248046875, "learning_rate": 3.7500000000000005e-06, "loss": 0.8567, "step": 1245 }, { "epoch": 1.8765060240963856, "grad_norm": 0.251953125, "learning_rate": 3.744979919678715e-06, "loss": 0.8472, "step": 1246 }, { "epoch": 1.8780120481927711, "grad_norm": 0.2451171875, "learning_rate": 3.73995983935743e-06, "loss": 0.8344, "step": 1247 }, { "epoch": 1.8795180722891565, "grad_norm": 0.2470703125, "learning_rate": 3.7349397590361445e-06, "loss": 0.8398, "step": 1248 }, { "epoch": 1.8810240963855422, "grad_norm": 0.25390625, "learning_rate": 3.7299196787148595e-06, "loss": 0.7925, "step": 1249 }, { "epoch": 1.8825301204819276, "grad_norm": 0.30078125, "learning_rate": 3.724899598393575e-06, "loss": 0.8315, "step": 1250 }, { "epoch": 1.8840361445783134, "grad_norm": 0.353515625, "learning_rate": 3.71987951807229e-06, "loss": 0.8367, "step": 1251 }, { "epoch": 1.8855421686746987, "grad_norm": 0.248046875, "learning_rate": 3.7148594377510043e-06, "loss": 0.8636, "step": 1252 }, { "epoch": 1.8870481927710845, "grad_norm": 0.2373046875, "learning_rate": 3.7098393574297193e-06, "loss": 0.8073, "step": 1253 }, { "epoch": 1.8885542168674698, "grad_norm": 0.25, "learning_rate": 3.7048192771084342e-06, "loss": 0.8244, "step": 1254 }, { "epoch": 1.8900602409638554, "grad_norm": 0.2890625, "learning_rate": 3.6997991967871487e-06, "loss": 0.8432, "step": 1255 }, { "epoch": 1.891566265060241, "grad_norm": 0.259765625, "learning_rate": 3.6947791164658637e-06, "loss": 0.8356, "step": 1256 }, { "epoch": 1.8930722891566265, "grad_norm": 0.251953125, "learning_rate": 3.6897590361445782e-06, "loss": 0.8718, "step": 1257 }, { "epoch": 1.894578313253012, "grad_norm": 0.265625, "learning_rate": 3.684738955823293e-06, "loss": 0.8886, "step": 1258 }, { "epoch": 1.8960843373493976, "grad_norm": 0.251953125, "learning_rate": 3.6797188755020085e-06, "loss": 0.8099, "step": 1259 }, { "epoch": 1.8975903614457832, "grad_norm": 0.26953125, "learning_rate": 3.6746987951807235e-06, "loss": 0.8038, "step": 1260 }, { "epoch": 1.8990963855421685, "grad_norm": 0.25390625, "learning_rate": 3.669678714859438e-06, "loss": 0.8352, "step": 1261 }, { "epoch": 1.9006024096385543, "grad_norm": 0.24609375, "learning_rate": 3.664658634538153e-06, "loss": 0.8331, "step": 1262 }, { "epoch": 1.9021084337349397, "grad_norm": 0.2470703125, "learning_rate": 3.659638554216868e-06, "loss": 0.8455, "step": 1263 }, { "epoch": 1.9036144578313254, "grad_norm": 0.26171875, "learning_rate": 3.6546184738955825e-06, "loss": 0.8132, "step": 1264 }, { "epoch": 1.9051204819277108, "grad_norm": 0.2470703125, "learning_rate": 3.6495983935742974e-06, "loss": 0.8112, "step": 1265 }, { "epoch": 1.9066265060240963, "grad_norm": 0.25390625, "learning_rate": 3.6445783132530124e-06, "loss": 0.859, "step": 1266 }, { "epoch": 1.908132530120482, "grad_norm": 0.244140625, "learning_rate": 3.639558232931727e-06, "loss": 0.815, "step": 1267 }, { "epoch": 1.9096385542168675, "grad_norm": 0.302734375, "learning_rate": 3.6345381526104423e-06, "loss": 0.8502, "step": 1268 }, { "epoch": 1.911144578313253, "grad_norm": 0.2578125, "learning_rate": 3.629518072289157e-06, "loss": 0.8612, "step": 1269 }, { "epoch": 1.9126506024096386, "grad_norm": 0.2490234375, "learning_rate": 3.6244979919678717e-06, "loss": 0.8045, "step": 1270 }, { "epoch": 1.9141566265060241, "grad_norm": 0.2578125, "learning_rate": 3.6194779116465867e-06, "loss": 0.8985, "step": 1271 }, { "epoch": 1.9156626506024095, "grad_norm": 0.2578125, "learning_rate": 3.6144578313253016e-06, "loss": 0.8093, "step": 1272 }, { "epoch": 1.9171686746987953, "grad_norm": 0.318359375, "learning_rate": 3.609437751004016e-06, "loss": 0.827, "step": 1273 }, { "epoch": 1.9186746987951806, "grad_norm": 0.37890625, "learning_rate": 3.604417670682731e-06, "loss": 0.8189, "step": 1274 }, { "epoch": 1.9201807228915664, "grad_norm": 0.279296875, "learning_rate": 3.599397590361446e-06, "loss": 0.8751, "step": 1275 }, { "epoch": 1.9216867469879517, "grad_norm": 0.2451171875, "learning_rate": 3.5943775100401606e-06, "loss": 0.8191, "step": 1276 }, { "epoch": 1.9231927710843375, "grad_norm": 0.25, "learning_rate": 3.589357429718876e-06, "loss": 0.8225, "step": 1277 }, { "epoch": 1.9246987951807228, "grad_norm": 0.302734375, "learning_rate": 3.584337349397591e-06, "loss": 0.8273, "step": 1278 }, { "epoch": 1.9262048192771084, "grad_norm": 0.2578125, "learning_rate": 3.5793172690763054e-06, "loss": 0.8654, "step": 1279 }, { "epoch": 1.927710843373494, "grad_norm": 0.2490234375, "learning_rate": 3.5742971887550204e-06, "loss": 0.8307, "step": 1280 }, { "epoch": 1.9292168674698795, "grad_norm": 0.251953125, "learning_rate": 3.5692771084337353e-06, "loss": 0.8342, "step": 1281 }, { "epoch": 1.930722891566265, "grad_norm": 0.2470703125, "learning_rate": 3.56425702811245e-06, "loss": 0.8523, "step": 1282 }, { "epoch": 1.9322289156626506, "grad_norm": 0.25390625, "learning_rate": 3.559236947791165e-06, "loss": 0.8286, "step": 1283 }, { "epoch": 1.9337349397590362, "grad_norm": 0.2373046875, "learning_rate": 3.5542168674698798e-06, "loss": 0.8106, "step": 1284 }, { "epoch": 1.9352409638554215, "grad_norm": 0.40234375, "learning_rate": 3.5491967871485943e-06, "loss": 0.8219, "step": 1285 }, { "epoch": 1.9367469879518073, "grad_norm": 0.255859375, "learning_rate": 3.5441767068273097e-06, "loss": 0.8174, "step": 1286 }, { "epoch": 1.9382530120481927, "grad_norm": 0.251953125, "learning_rate": 3.5391566265060246e-06, "loss": 0.8321, "step": 1287 }, { "epoch": 1.9397590361445785, "grad_norm": 0.25390625, "learning_rate": 3.534136546184739e-06, "loss": 0.8448, "step": 1288 }, { "epoch": 1.9412650602409638, "grad_norm": 0.2578125, "learning_rate": 3.529116465863454e-06, "loss": 0.8645, "step": 1289 }, { "epoch": 1.9427710843373494, "grad_norm": 0.330078125, "learning_rate": 3.524096385542169e-06, "loss": 0.8054, "step": 1290 }, { "epoch": 1.944277108433735, "grad_norm": 0.25, "learning_rate": 3.5190763052208836e-06, "loss": 0.8266, "step": 1291 }, { "epoch": 1.9457831325301205, "grad_norm": 0.248046875, "learning_rate": 3.5140562248995985e-06, "loss": 0.8322, "step": 1292 }, { "epoch": 1.947289156626506, "grad_norm": 0.248046875, "learning_rate": 3.5090361445783135e-06, "loss": 0.8359, "step": 1293 }, { "epoch": 1.9487951807228916, "grad_norm": 0.25, "learning_rate": 3.504016064257029e-06, "loss": 0.8347, "step": 1294 }, { "epoch": 1.9503012048192772, "grad_norm": 0.267578125, "learning_rate": 3.4989959839357434e-06, "loss": 0.8278, "step": 1295 }, { "epoch": 1.9518072289156625, "grad_norm": 0.32421875, "learning_rate": 3.4939759036144583e-06, "loss": 0.8317, "step": 1296 }, { "epoch": 1.9533132530120483, "grad_norm": 0.2431640625, "learning_rate": 3.4889558232931733e-06, "loss": 0.8315, "step": 1297 }, { "epoch": 1.9548192771084336, "grad_norm": 0.25, "learning_rate": 3.4839357429718878e-06, "loss": 0.8557, "step": 1298 }, { "epoch": 1.9563253012048194, "grad_norm": 0.2734375, "learning_rate": 3.4789156626506027e-06, "loss": 0.8528, "step": 1299 }, { "epoch": 1.9578313253012047, "grad_norm": 0.2490234375, "learning_rate": 3.4738955823293173e-06, "loss": 0.8201, "step": 1300 }, { "epoch": 1.9593373493975905, "grad_norm": 0.2392578125, "learning_rate": 3.4688755020080322e-06, "loss": 0.7992, "step": 1301 }, { "epoch": 1.9608433734939759, "grad_norm": 0.24609375, "learning_rate": 3.463855421686747e-06, "loss": 0.8172, "step": 1302 }, { "epoch": 1.9623493975903614, "grad_norm": 0.2470703125, "learning_rate": 3.4588353413654625e-06, "loss": 0.8352, "step": 1303 }, { "epoch": 1.963855421686747, "grad_norm": 0.2412109375, "learning_rate": 3.453815261044177e-06, "loss": 0.7697, "step": 1304 }, { "epoch": 1.9653614457831325, "grad_norm": 0.375, "learning_rate": 3.448795180722892e-06, "loss": 0.7819, "step": 1305 }, { "epoch": 1.966867469879518, "grad_norm": 0.2578125, "learning_rate": 3.443775100401607e-06, "loss": 0.844, "step": 1306 }, { "epoch": 1.9683734939759037, "grad_norm": 0.2490234375, "learning_rate": 3.4387550200803215e-06, "loss": 0.8195, "step": 1307 }, { "epoch": 1.9698795180722892, "grad_norm": 0.251953125, "learning_rate": 3.4337349397590364e-06, "loss": 0.8009, "step": 1308 }, { "epoch": 1.9713855421686746, "grad_norm": 0.2578125, "learning_rate": 3.428714859437751e-06, "loss": 0.8361, "step": 1309 }, { "epoch": 1.9728915662650603, "grad_norm": 0.2578125, "learning_rate": 3.423694779116466e-06, "loss": 0.8461, "step": 1310 }, { "epoch": 1.9743975903614457, "grad_norm": 0.271484375, "learning_rate": 3.418674698795181e-06, "loss": 0.8058, "step": 1311 }, { "epoch": 1.9759036144578315, "grad_norm": 0.30859375, "learning_rate": 3.4136546184738962e-06, "loss": 0.8582, "step": 1312 }, { "epoch": 1.9774096385542168, "grad_norm": 0.267578125, "learning_rate": 3.4086345381526108e-06, "loss": 0.8424, "step": 1313 }, { "epoch": 1.9789156626506024, "grad_norm": 0.28125, "learning_rate": 3.4036144578313257e-06, "loss": 0.8489, "step": 1314 }, { "epoch": 1.980421686746988, "grad_norm": 0.24609375, "learning_rate": 3.3985943775100407e-06, "loss": 0.8361, "step": 1315 }, { "epoch": 1.9819277108433735, "grad_norm": 0.271484375, "learning_rate": 3.393574297188755e-06, "loss": 0.8448, "step": 1316 }, { "epoch": 1.983433734939759, "grad_norm": 0.2490234375, "learning_rate": 3.38855421686747e-06, "loss": 0.8548, "step": 1317 }, { "epoch": 1.9849397590361446, "grad_norm": 0.26953125, "learning_rate": 3.383534136546185e-06, "loss": 0.7868, "step": 1318 }, { "epoch": 1.9864457831325302, "grad_norm": 0.259765625, "learning_rate": 3.3785140562248996e-06, "loss": 0.8704, "step": 1319 }, { "epoch": 1.9879518072289155, "grad_norm": 0.26953125, "learning_rate": 3.3734939759036146e-06, "loss": 0.8583, "step": 1320 }, { "epoch": 1.9894578313253013, "grad_norm": 0.2578125, "learning_rate": 3.36847389558233e-06, "loss": 0.8498, "step": 1321 }, { "epoch": 1.9909638554216866, "grad_norm": 0.255859375, "learning_rate": 3.3634538152610445e-06, "loss": 0.8428, "step": 1322 }, { "epoch": 1.9924698795180724, "grad_norm": 0.2890625, "learning_rate": 3.3584337349397594e-06, "loss": 0.8749, "step": 1323 }, { "epoch": 1.9939759036144578, "grad_norm": 0.2412109375, "learning_rate": 3.3534136546184744e-06, "loss": 0.8024, "step": 1324 }, { "epoch": 1.9954819277108435, "grad_norm": 0.2470703125, "learning_rate": 3.348393574297189e-06, "loss": 0.8157, "step": 1325 }, { "epoch": 1.9969879518072289, "grad_norm": 0.298828125, "learning_rate": 3.343373493975904e-06, "loss": 0.8042, "step": 1326 }, { "epoch": 1.9984939759036144, "grad_norm": 0.251953125, "learning_rate": 3.338353413654619e-06, "loss": 0.8558, "step": 1327 }, { "epoch": 2.0, "grad_norm": 0.27734375, "learning_rate": 3.3333333333333333e-06, "loss": 0.8152, "step": 1328 }, { "epoch": 2.0015060240963853, "grad_norm": 0.25, "learning_rate": 3.3283132530120483e-06, "loss": 0.8599, "step": 1329 }, { "epoch": 2.003012048192771, "grad_norm": 0.29296875, "learning_rate": 3.3232931726907636e-06, "loss": 0.8219, "step": 1330 }, { "epoch": 2.0045180722891565, "grad_norm": 0.271484375, "learning_rate": 3.318273092369478e-06, "loss": 0.797, "step": 1331 }, { "epoch": 2.0060240963855422, "grad_norm": 0.251953125, "learning_rate": 3.313253012048193e-06, "loss": 0.8524, "step": 1332 }, { "epoch": 2.0075301204819276, "grad_norm": 0.2470703125, "learning_rate": 3.308232931726908e-06, "loss": 0.8576, "step": 1333 }, { "epoch": 2.0090361445783134, "grad_norm": 0.2451171875, "learning_rate": 3.3032128514056226e-06, "loss": 0.8095, "step": 1334 }, { "epoch": 2.0105421686746987, "grad_norm": 0.302734375, "learning_rate": 3.2981927710843376e-06, "loss": 0.7863, "step": 1335 }, { "epoch": 2.0120481927710845, "grad_norm": 0.25390625, "learning_rate": 3.2931726907630525e-06, "loss": 0.8314, "step": 1336 }, { "epoch": 2.01355421686747, "grad_norm": 0.240234375, "learning_rate": 3.288152610441767e-06, "loss": 0.8071, "step": 1337 }, { "epoch": 2.0150602409638556, "grad_norm": 0.251953125, "learning_rate": 3.283132530120482e-06, "loss": 0.8313, "step": 1338 }, { "epoch": 2.016566265060241, "grad_norm": 0.251953125, "learning_rate": 3.2781124497991974e-06, "loss": 0.8506, "step": 1339 }, { "epoch": 2.0180722891566263, "grad_norm": 0.2431640625, "learning_rate": 3.2730923694779123e-06, "loss": 0.8458, "step": 1340 }, { "epoch": 2.019578313253012, "grad_norm": 0.2734375, "learning_rate": 3.268072289156627e-06, "loss": 0.815, "step": 1341 }, { "epoch": 2.0210843373493974, "grad_norm": 0.26171875, "learning_rate": 3.2630522088353418e-06, "loss": 0.7832, "step": 1342 }, { "epoch": 2.022590361445783, "grad_norm": 0.25, "learning_rate": 3.2580321285140563e-06, "loss": 0.814, "step": 1343 }, { "epoch": 2.0240963855421685, "grad_norm": 0.244140625, "learning_rate": 3.2530120481927713e-06, "loss": 0.7797, "step": 1344 }, { "epoch": 2.0256024096385543, "grad_norm": 0.248046875, "learning_rate": 3.247991967871486e-06, "loss": 0.833, "step": 1345 }, { "epoch": 2.0271084337349397, "grad_norm": 0.2490234375, "learning_rate": 3.2429718875502007e-06, "loss": 0.8174, "step": 1346 }, { "epoch": 2.0286144578313254, "grad_norm": 0.275390625, "learning_rate": 3.2379518072289157e-06, "loss": 0.7923, "step": 1347 }, { "epoch": 2.0301204819277108, "grad_norm": 0.26171875, "learning_rate": 3.232931726907631e-06, "loss": 0.8257, "step": 1348 }, { "epoch": 2.0316265060240966, "grad_norm": 0.25390625, "learning_rate": 3.227911646586346e-06, "loss": 0.7905, "step": 1349 }, { "epoch": 2.033132530120482, "grad_norm": 0.28515625, "learning_rate": 3.2228915662650605e-06, "loss": 0.8212, "step": 1350 }, { "epoch": 2.0346385542168677, "grad_norm": 0.267578125, "learning_rate": 3.2178714859437755e-06, "loss": 0.8241, "step": 1351 }, { "epoch": 2.036144578313253, "grad_norm": 0.255859375, "learning_rate": 3.21285140562249e-06, "loss": 0.8441, "step": 1352 }, { "epoch": 2.0376506024096384, "grad_norm": 0.365234375, "learning_rate": 3.207831325301205e-06, "loss": 0.8498, "step": 1353 }, { "epoch": 2.039156626506024, "grad_norm": 0.380859375, "learning_rate": 3.20281124497992e-06, "loss": 0.7872, "step": 1354 }, { "epoch": 2.0406626506024095, "grad_norm": 0.2734375, "learning_rate": 3.1977911646586344e-06, "loss": 0.7995, "step": 1355 }, { "epoch": 2.0421686746987953, "grad_norm": 0.37109375, "learning_rate": 3.1927710843373494e-06, "loss": 0.8254, "step": 1356 }, { "epoch": 2.0436746987951806, "grad_norm": 0.2490234375, "learning_rate": 3.1877510040160648e-06, "loss": 0.8435, "step": 1357 }, { "epoch": 2.0451807228915664, "grad_norm": 0.25390625, "learning_rate": 3.1827309236947797e-06, "loss": 0.8255, "step": 1358 }, { "epoch": 2.0466867469879517, "grad_norm": 0.27734375, "learning_rate": 3.1777108433734942e-06, "loss": 0.8286, "step": 1359 }, { "epoch": 2.0481927710843375, "grad_norm": 0.263671875, "learning_rate": 3.172690763052209e-06, "loss": 0.8103, "step": 1360 }, { "epoch": 2.049698795180723, "grad_norm": 0.265625, "learning_rate": 3.167670682730924e-06, "loss": 0.8598, "step": 1361 }, { "epoch": 2.0512048192771086, "grad_norm": 0.41015625, "learning_rate": 3.1626506024096387e-06, "loss": 0.8017, "step": 1362 }, { "epoch": 2.052710843373494, "grad_norm": 0.25390625, "learning_rate": 3.1576305220883536e-06, "loss": 0.7972, "step": 1363 }, { "epoch": 2.0542168674698793, "grad_norm": 0.2490234375, "learning_rate": 3.152610441767068e-06, "loss": 0.8275, "step": 1364 }, { "epoch": 2.055722891566265, "grad_norm": 0.25, "learning_rate": 3.147590361445783e-06, "loss": 0.8322, "step": 1365 }, { "epoch": 2.0572289156626504, "grad_norm": 0.27734375, "learning_rate": 3.1425702811244985e-06, "loss": 0.8144, "step": 1366 }, { "epoch": 2.058734939759036, "grad_norm": 0.24609375, "learning_rate": 3.1375502008032134e-06, "loss": 0.8361, "step": 1367 }, { "epoch": 2.0602409638554215, "grad_norm": 0.3828125, "learning_rate": 3.132530120481928e-06, "loss": 0.8421, "step": 1368 }, { "epoch": 2.0617469879518073, "grad_norm": 0.2470703125, "learning_rate": 3.127510040160643e-06, "loss": 0.8666, "step": 1369 }, { "epoch": 2.0632530120481927, "grad_norm": 0.263671875, "learning_rate": 3.122489959839358e-06, "loss": 0.8316, "step": 1370 }, { "epoch": 2.0647590361445785, "grad_norm": 0.29296875, "learning_rate": 3.1174698795180724e-06, "loss": 0.8172, "step": 1371 }, { "epoch": 2.066265060240964, "grad_norm": 0.255859375, "learning_rate": 3.1124497991967873e-06, "loss": 0.8143, "step": 1372 }, { "epoch": 2.0677710843373496, "grad_norm": 0.318359375, "learning_rate": 3.107429718875502e-06, "loss": 0.8356, "step": 1373 }, { "epoch": 2.069277108433735, "grad_norm": 0.25390625, "learning_rate": 3.1024096385542172e-06, "loss": 0.845, "step": 1374 }, { "epoch": 2.0707831325301207, "grad_norm": 0.326171875, "learning_rate": 3.097389558232932e-06, "loss": 0.8451, "step": 1375 }, { "epoch": 2.072289156626506, "grad_norm": 0.25390625, "learning_rate": 3.092369477911647e-06, "loss": 0.8001, "step": 1376 }, { "epoch": 2.0737951807228914, "grad_norm": 0.310546875, "learning_rate": 3.0873493975903616e-06, "loss": 0.8438, "step": 1377 }, { "epoch": 2.075301204819277, "grad_norm": 0.2578125, "learning_rate": 3.0823293172690766e-06, "loss": 0.8377, "step": 1378 }, { "epoch": 2.0768072289156625, "grad_norm": 0.25390625, "learning_rate": 3.0773092369477915e-06, "loss": 0.85, "step": 1379 }, { "epoch": 2.0783132530120483, "grad_norm": 0.26171875, "learning_rate": 3.072289156626506e-06, "loss": 0.7974, "step": 1380 }, { "epoch": 2.0798192771084336, "grad_norm": 0.283203125, "learning_rate": 3.067269076305221e-06, "loss": 0.8209, "step": 1381 }, { "epoch": 2.0813253012048194, "grad_norm": 0.2578125, "learning_rate": 3.062248995983936e-06, "loss": 0.862, "step": 1382 }, { "epoch": 2.0828313253012047, "grad_norm": 0.2470703125, "learning_rate": 3.057228915662651e-06, "loss": 0.8241, "step": 1383 }, { "epoch": 2.0843373493975905, "grad_norm": 0.609375, "learning_rate": 3.052208835341366e-06, "loss": 0.8213, "step": 1384 }, { "epoch": 2.085843373493976, "grad_norm": 0.255859375, "learning_rate": 3.047188755020081e-06, "loss": 0.7966, "step": 1385 }, { "epoch": 2.0873493975903616, "grad_norm": 0.255859375, "learning_rate": 3.0421686746987953e-06, "loss": 0.8465, "step": 1386 }, { "epoch": 2.088855421686747, "grad_norm": 0.2578125, "learning_rate": 3.0371485943775103e-06, "loss": 0.8446, "step": 1387 }, { "epoch": 2.0903614457831323, "grad_norm": 0.25390625, "learning_rate": 3.0321285140562252e-06, "loss": 0.8235, "step": 1388 }, { "epoch": 2.091867469879518, "grad_norm": 0.255859375, "learning_rate": 3.0271084337349398e-06, "loss": 0.8165, "step": 1389 }, { "epoch": 2.0933734939759034, "grad_norm": 0.2451171875, "learning_rate": 3.0220883534136547e-06, "loss": 0.8426, "step": 1390 }, { "epoch": 2.0948795180722892, "grad_norm": 0.26953125, "learning_rate": 3.0170682730923697e-06, "loss": 0.8709, "step": 1391 }, { "epoch": 2.0963855421686746, "grad_norm": 0.2373046875, "learning_rate": 3.012048192771085e-06, "loss": 0.8314, "step": 1392 }, { "epoch": 2.0978915662650603, "grad_norm": 0.400390625, "learning_rate": 3.0070281124497996e-06, "loss": 0.8489, "step": 1393 }, { "epoch": 2.0993975903614457, "grad_norm": 0.255859375, "learning_rate": 3.0020080321285145e-06, "loss": 0.8062, "step": 1394 }, { "epoch": 2.1009036144578315, "grad_norm": 0.2490234375, "learning_rate": 2.996987951807229e-06, "loss": 0.8305, "step": 1395 }, { "epoch": 2.102409638554217, "grad_norm": 1.9921875, "learning_rate": 2.991967871485944e-06, "loss": 0.7818, "step": 1396 }, { "epoch": 2.1039156626506026, "grad_norm": 0.25390625, "learning_rate": 2.986947791164659e-06, "loss": 0.8169, "step": 1397 }, { "epoch": 2.105421686746988, "grad_norm": 0.25390625, "learning_rate": 2.9819277108433735e-06, "loss": 0.8366, "step": 1398 }, { "epoch": 2.1069277108433733, "grad_norm": 0.259765625, "learning_rate": 2.9769076305220884e-06, "loss": 0.8403, "step": 1399 }, { "epoch": 2.108433734939759, "grad_norm": 0.265625, "learning_rate": 2.9718875502008034e-06, "loss": 0.787, "step": 1400 }, { "epoch": 2.1099397590361444, "grad_norm": 0.25390625, "learning_rate": 2.9668674698795187e-06, "loss": 0.835, "step": 1401 }, { "epoch": 2.11144578313253, "grad_norm": 0.251953125, "learning_rate": 2.9618473895582333e-06, "loss": 0.8116, "step": 1402 }, { "epoch": 2.1129518072289155, "grad_norm": 0.255859375, "learning_rate": 2.9568273092369482e-06, "loss": 0.834, "step": 1403 }, { "epoch": 2.1144578313253013, "grad_norm": 0.27734375, "learning_rate": 2.9518072289156627e-06, "loss": 0.8434, "step": 1404 }, { "epoch": 2.1159638554216866, "grad_norm": 0.3046875, "learning_rate": 2.9467871485943777e-06, "loss": 0.7913, "step": 1405 }, { "epoch": 2.1174698795180724, "grad_norm": 0.240234375, "learning_rate": 2.9417670682730926e-06, "loss": 0.8063, "step": 1406 }, { "epoch": 2.1189759036144578, "grad_norm": 2.546875, "learning_rate": 2.936746987951807e-06, "loss": 0.8765, "step": 1407 }, { "epoch": 2.1204819277108435, "grad_norm": 0.2578125, "learning_rate": 2.931726907630522e-06, "loss": 0.8255, "step": 1408 }, { "epoch": 2.121987951807229, "grad_norm": 2.421875, "learning_rate": 2.926706827309237e-06, "loss": 0.8194, "step": 1409 }, { "epoch": 2.1234939759036147, "grad_norm": 0.2470703125, "learning_rate": 2.9216867469879524e-06, "loss": 0.7879, "step": 1410 }, { "epoch": 2.125, "grad_norm": 0.2578125, "learning_rate": 2.916666666666667e-06, "loss": 0.8168, "step": 1411 }, { "epoch": 2.1265060240963853, "grad_norm": 0.255859375, "learning_rate": 2.911646586345382e-06, "loss": 0.8523, "step": 1412 }, { "epoch": 2.128012048192771, "grad_norm": 0.240234375, "learning_rate": 2.906626506024097e-06, "loss": 0.8373, "step": 1413 }, { "epoch": 2.1295180722891565, "grad_norm": 0.263671875, "learning_rate": 2.9016064257028114e-06, "loss": 0.8937, "step": 1414 }, { "epoch": 2.1310240963855422, "grad_norm": 0.318359375, "learning_rate": 2.8965863453815264e-06, "loss": 0.8285, "step": 1415 }, { "epoch": 2.1325301204819276, "grad_norm": 0.248046875, "learning_rate": 2.891566265060241e-06, "loss": 0.8213, "step": 1416 }, { "epoch": 2.1340361445783134, "grad_norm": 0.275390625, "learning_rate": 2.886546184738956e-06, "loss": 0.8345, "step": 1417 }, { "epoch": 2.1355421686746987, "grad_norm": 0.25390625, "learning_rate": 2.8815261044176708e-06, "loss": 0.795, "step": 1418 }, { "epoch": 2.1370481927710845, "grad_norm": 0.271484375, "learning_rate": 2.876506024096386e-06, "loss": 0.8317, "step": 1419 }, { "epoch": 2.13855421686747, "grad_norm": 0.255859375, "learning_rate": 2.8714859437751007e-06, "loss": 0.8304, "step": 1420 }, { "epoch": 2.1400602409638556, "grad_norm": 0.259765625, "learning_rate": 2.8664658634538156e-06, "loss": 0.8673, "step": 1421 }, { "epoch": 2.141566265060241, "grad_norm": 0.2431640625, "learning_rate": 2.8614457831325306e-06, "loss": 0.823, "step": 1422 }, { "epoch": 2.1430722891566267, "grad_norm": 0.2578125, "learning_rate": 2.856425702811245e-06, "loss": 0.8444, "step": 1423 }, { "epoch": 2.144578313253012, "grad_norm": 0.248046875, "learning_rate": 2.85140562248996e-06, "loss": 0.8019, "step": 1424 }, { "epoch": 2.1460843373493974, "grad_norm": 0.267578125, "learning_rate": 2.846385542168675e-06, "loss": 0.8201, "step": 1425 }, { "epoch": 2.147590361445783, "grad_norm": 0.2470703125, "learning_rate": 2.8413654618473895e-06, "loss": 0.8205, "step": 1426 }, { "epoch": 2.1490963855421685, "grad_norm": 0.25, "learning_rate": 2.8363453815261045e-06, "loss": 0.8231, "step": 1427 }, { "epoch": 2.1506024096385543, "grad_norm": 0.265625, "learning_rate": 2.83132530120482e-06, "loss": 0.8652, "step": 1428 }, { "epoch": 2.1521084337349397, "grad_norm": 0.255859375, "learning_rate": 2.8263052208835344e-06, "loss": 0.8423, "step": 1429 }, { "epoch": 2.1536144578313254, "grad_norm": 0.265625, "learning_rate": 2.8212851405622493e-06, "loss": 0.8415, "step": 1430 }, { "epoch": 2.1551204819277108, "grad_norm": 0.279296875, "learning_rate": 2.8162650602409643e-06, "loss": 0.8124, "step": 1431 }, { "epoch": 2.1566265060240966, "grad_norm": 0.263671875, "learning_rate": 2.811244979919679e-06, "loss": 0.7975, "step": 1432 }, { "epoch": 2.158132530120482, "grad_norm": 0.328125, "learning_rate": 2.8062248995983938e-06, "loss": 0.8338, "step": 1433 }, { "epoch": 2.1596385542168672, "grad_norm": 0.2470703125, "learning_rate": 2.8012048192771087e-06, "loss": 0.8465, "step": 1434 }, { "epoch": 2.161144578313253, "grad_norm": 0.2412109375, "learning_rate": 2.7961847389558232e-06, "loss": 0.7937, "step": 1435 }, { "epoch": 2.1626506024096384, "grad_norm": 0.265625, "learning_rate": 2.791164658634538e-06, "loss": 0.8284, "step": 1436 }, { "epoch": 2.164156626506024, "grad_norm": 0.318359375, "learning_rate": 2.7861445783132536e-06, "loss": 0.8559, "step": 1437 }, { "epoch": 2.1656626506024095, "grad_norm": 0.318359375, "learning_rate": 2.781124497991968e-06, "loss": 0.7641, "step": 1438 }, { "epoch": 2.1671686746987953, "grad_norm": 0.255859375, "learning_rate": 2.776104417670683e-06, "loss": 0.8307, "step": 1439 }, { "epoch": 2.1686746987951806, "grad_norm": 0.255859375, "learning_rate": 2.771084337349398e-06, "loss": 0.8304, "step": 1440 }, { "epoch": 2.1701807228915664, "grad_norm": 0.25390625, "learning_rate": 2.7660642570281125e-06, "loss": 0.8063, "step": 1441 }, { "epoch": 2.1716867469879517, "grad_norm": 0.267578125, "learning_rate": 2.7610441767068275e-06, "loss": 0.8315, "step": 1442 }, { "epoch": 2.1731927710843375, "grad_norm": 0.279296875, "learning_rate": 2.7560240963855424e-06, "loss": 0.8299, "step": 1443 }, { "epoch": 2.174698795180723, "grad_norm": 0.265625, "learning_rate": 2.751004016064257e-06, "loss": 0.8731, "step": 1444 }, { "epoch": 2.1762048192771086, "grad_norm": 0.248046875, "learning_rate": 2.745983935742972e-06, "loss": 0.7707, "step": 1445 }, { "epoch": 2.177710843373494, "grad_norm": 0.3046875, "learning_rate": 2.7409638554216873e-06, "loss": 0.8098, "step": 1446 }, { "epoch": 2.1792168674698793, "grad_norm": 0.248046875, "learning_rate": 2.735943775100402e-06, "loss": 0.8302, "step": 1447 }, { "epoch": 2.180722891566265, "grad_norm": 0.265625, "learning_rate": 2.7309236947791167e-06, "loss": 0.8517, "step": 1448 }, { "epoch": 2.1822289156626504, "grad_norm": 0.25, "learning_rate": 2.7259036144578317e-06, "loss": 0.8566, "step": 1449 }, { "epoch": 2.183734939759036, "grad_norm": 0.2451171875, "learning_rate": 2.7208835341365462e-06, "loss": 0.8245, "step": 1450 }, { "epoch": 2.1852409638554215, "grad_norm": 0.255859375, "learning_rate": 2.715863453815261e-06, "loss": 0.8796, "step": 1451 }, { "epoch": 2.1867469879518073, "grad_norm": 0.244140625, "learning_rate": 2.710843373493976e-06, "loss": 0.8091, "step": 1452 }, { "epoch": 2.1882530120481927, "grad_norm": 0.255859375, "learning_rate": 2.7058232931726906e-06, "loss": 0.8074, "step": 1453 }, { "epoch": 2.1897590361445785, "grad_norm": 0.265625, "learning_rate": 2.700803212851406e-06, "loss": 0.8406, "step": 1454 }, { "epoch": 2.191265060240964, "grad_norm": 0.3125, "learning_rate": 2.695783132530121e-06, "loss": 0.8162, "step": 1455 }, { "epoch": 2.1927710843373496, "grad_norm": 0.265625, "learning_rate": 2.690763052208836e-06, "loss": 0.831, "step": 1456 }, { "epoch": 2.194277108433735, "grad_norm": 0.2470703125, "learning_rate": 2.6857429718875504e-06, "loss": 0.8082, "step": 1457 }, { "epoch": 2.1957831325301207, "grad_norm": 0.296875, "learning_rate": 2.6807228915662654e-06, "loss": 0.7967, "step": 1458 }, { "epoch": 2.197289156626506, "grad_norm": 0.251953125, "learning_rate": 2.67570281124498e-06, "loss": 0.8136, "step": 1459 }, { "epoch": 2.1987951807228914, "grad_norm": 0.2890625, "learning_rate": 2.670682730923695e-06, "loss": 0.7868, "step": 1460 }, { "epoch": 2.200301204819277, "grad_norm": 0.263671875, "learning_rate": 2.66566265060241e-06, "loss": 0.8331, "step": 1461 }, { "epoch": 2.2018072289156625, "grad_norm": 0.26171875, "learning_rate": 2.6606425702811243e-06, "loss": 0.8287, "step": 1462 }, { "epoch": 2.2033132530120483, "grad_norm": 0.302734375, "learning_rate": 2.6556224899598397e-06, "loss": 0.8057, "step": 1463 }, { "epoch": 2.2048192771084336, "grad_norm": 0.248046875, "learning_rate": 2.6506024096385547e-06, "loss": 0.8285, "step": 1464 }, { "epoch": 2.2063253012048194, "grad_norm": 0.26171875, "learning_rate": 2.6455823293172696e-06, "loss": 0.8403, "step": 1465 }, { "epoch": 2.2078313253012047, "grad_norm": 0.318359375, "learning_rate": 2.640562248995984e-06, "loss": 0.7968, "step": 1466 }, { "epoch": 2.2093373493975905, "grad_norm": 0.2421875, "learning_rate": 2.635542168674699e-06, "loss": 0.76, "step": 1467 }, { "epoch": 2.210843373493976, "grad_norm": 0.2490234375, "learning_rate": 2.6305220883534136e-06, "loss": 0.8426, "step": 1468 }, { "epoch": 2.2123493975903616, "grad_norm": 0.255859375, "learning_rate": 2.6255020080321286e-06, "loss": 0.8206, "step": 1469 }, { "epoch": 2.213855421686747, "grad_norm": 0.33203125, "learning_rate": 2.6204819277108435e-06, "loss": 0.8403, "step": 1470 }, { "epoch": 2.2153614457831328, "grad_norm": 0.263671875, "learning_rate": 2.615461847389558e-06, "loss": 0.854, "step": 1471 }, { "epoch": 2.216867469879518, "grad_norm": 0.2578125, "learning_rate": 2.6104417670682734e-06, "loss": 0.8559, "step": 1472 }, { "epoch": 2.2183734939759034, "grad_norm": 0.2490234375, "learning_rate": 2.6054216867469884e-06, "loss": 0.8363, "step": 1473 }, { "epoch": 2.2198795180722892, "grad_norm": 0.26171875, "learning_rate": 2.6004016064257033e-06, "loss": 0.8623, "step": 1474 }, { "epoch": 2.2213855421686746, "grad_norm": 0.2421875, "learning_rate": 2.595381526104418e-06, "loss": 0.8058, "step": 1475 }, { "epoch": 2.2228915662650603, "grad_norm": 0.251953125, "learning_rate": 2.590361445783133e-06, "loss": 0.7957, "step": 1476 }, { "epoch": 2.2243975903614457, "grad_norm": 0.361328125, "learning_rate": 2.5853413654618477e-06, "loss": 0.7873, "step": 1477 }, { "epoch": 2.2259036144578315, "grad_norm": 0.29296875, "learning_rate": 2.5803212851405623e-06, "loss": 0.8426, "step": 1478 }, { "epoch": 2.227409638554217, "grad_norm": 0.25, "learning_rate": 2.5753012048192772e-06, "loss": 0.8493, "step": 1479 }, { "epoch": 2.2289156626506026, "grad_norm": 0.283203125, "learning_rate": 2.5702811244979918e-06, "loss": 0.8385, "step": 1480 }, { "epoch": 2.230421686746988, "grad_norm": 0.2470703125, "learning_rate": 2.565261044176707e-06, "loss": 0.8073, "step": 1481 }, { "epoch": 2.2319277108433733, "grad_norm": 0.25390625, "learning_rate": 2.560240963855422e-06, "loss": 0.8171, "step": 1482 }, { "epoch": 2.233433734939759, "grad_norm": 0.259765625, "learning_rate": 2.555220883534137e-06, "loss": 0.8243, "step": 1483 }, { "epoch": 2.2349397590361444, "grad_norm": 0.322265625, "learning_rate": 2.5502008032128516e-06, "loss": 0.854, "step": 1484 }, { "epoch": 2.23644578313253, "grad_norm": 0.251953125, "learning_rate": 2.5451807228915665e-06, "loss": 0.8245, "step": 1485 }, { "epoch": 2.2379518072289155, "grad_norm": 0.265625, "learning_rate": 2.5401606425702815e-06, "loss": 0.8159, "step": 1486 }, { "epoch": 2.2394578313253013, "grad_norm": 0.267578125, "learning_rate": 2.535140562248996e-06, "loss": 0.7954, "step": 1487 }, { "epoch": 2.2409638554216866, "grad_norm": 0.25, "learning_rate": 2.530120481927711e-06, "loss": 0.8355, "step": 1488 }, { "epoch": 2.2424698795180724, "grad_norm": 0.25390625, "learning_rate": 2.525100401606426e-06, "loss": 0.8334, "step": 1489 }, { "epoch": 2.2439759036144578, "grad_norm": 0.255859375, "learning_rate": 2.520080321285141e-06, "loss": 0.831, "step": 1490 }, { "epoch": 2.2454819277108435, "grad_norm": 0.2451171875, "learning_rate": 2.5150602409638558e-06, "loss": 0.8488, "step": 1491 }, { "epoch": 2.246987951807229, "grad_norm": 0.2578125, "learning_rate": 2.5100401606425707e-06, "loss": 0.854, "step": 1492 }, { "epoch": 2.2484939759036147, "grad_norm": 0.361328125, "learning_rate": 2.5050200803212853e-06, "loss": 0.8075, "step": 1493 }, { "epoch": 2.25, "grad_norm": 0.3125, "learning_rate": 2.5e-06, "loss": 0.7717, "step": 1494 }, { "epoch": 2.2515060240963853, "grad_norm": 0.25390625, "learning_rate": 2.494979919678715e-06, "loss": 0.883, "step": 1495 }, { "epoch": 2.253012048192771, "grad_norm": 0.2470703125, "learning_rate": 2.4899598393574297e-06, "loss": 0.8207, "step": 1496 }, { "epoch": 2.2545180722891565, "grad_norm": 0.27734375, "learning_rate": 2.484939759036145e-06, "loss": 0.857, "step": 1497 }, { "epoch": 2.2560240963855422, "grad_norm": 0.2470703125, "learning_rate": 2.4799196787148596e-06, "loss": 0.8525, "step": 1498 }, { "epoch": 2.2575301204819276, "grad_norm": 0.26953125, "learning_rate": 2.4748995983935745e-06, "loss": 0.7866, "step": 1499 }, { "epoch": 2.2590361445783134, "grad_norm": 0.25390625, "learning_rate": 2.469879518072289e-06, "loss": 0.8253, "step": 1500 }, { "epoch": 2.2605421686746987, "grad_norm": 0.337890625, "learning_rate": 2.4648594377510044e-06, "loss": 0.8845, "step": 1501 }, { "epoch": 2.2620481927710845, "grad_norm": 0.263671875, "learning_rate": 2.459839357429719e-06, "loss": 0.8498, "step": 1502 }, { "epoch": 2.26355421686747, "grad_norm": 0.294921875, "learning_rate": 2.454819277108434e-06, "loss": 0.8183, "step": 1503 }, { "epoch": 2.2650602409638556, "grad_norm": 0.259765625, "learning_rate": 2.449799196787149e-06, "loss": 0.8185, "step": 1504 }, { "epoch": 2.266566265060241, "grad_norm": 0.267578125, "learning_rate": 2.4447791164658634e-06, "loss": 0.8032, "step": 1505 }, { "epoch": 2.2680722891566267, "grad_norm": 0.251953125, "learning_rate": 2.4397590361445788e-06, "loss": 0.8281, "step": 1506 }, { "epoch": 2.269578313253012, "grad_norm": 0.35546875, "learning_rate": 2.4347389558232933e-06, "loss": 0.8279, "step": 1507 }, { "epoch": 2.2710843373493974, "grad_norm": 0.259765625, "learning_rate": 2.4297188755020082e-06, "loss": 0.7928, "step": 1508 }, { "epoch": 2.272590361445783, "grad_norm": 0.255859375, "learning_rate": 2.424698795180723e-06, "loss": 0.8162, "step": 1509 }, { "epoch": 2.2740963855421685, "grad_norm": 0.24609375, "learning_rate": 2.419678714859438e-06, "loss": 0.8364, "step": 1510 }, { "epoch": 2.2756024096385543, "grad_norm": 0.2578125, "learning_rate": 2.4146586345381527e-06, "loss": 0.8215, "step": 1511 }, { "epoch": 2.2771084337349397, "grad_norm": 0.2470703125, "learning_rate": 2.4096385542168676e-06, "loss": 0.8264, "step": 1512 }, { "epoch": 2.2786144578313254, "grad_norm": 0.263671875, "learning_rate": 2.4046184738955826e-06, "loss": 0.8341, "step": 1513 }, { "epoch": 2.2801204819277108, "grad_norm": 0.2470703125, "learning_rate": 2.399598393574297e-06, "loss": 0.8302, "step": 1514 }, { "epoch": 2.2816265060240966, "grad_norm": 0.341796875, "learning_rate": 2.3945783132530125e-06, "loss": 0.8434, "step": 1515 }, { "epoch": 2.283132530120482, "grad_norm": 0.265625, "learning_rate": 2.389558232931727e-06, "loss": 0.8033, "step": 1516 }, { "epoch": 2.2846385542168672, "grad_norm": 0.259765625, "learning_rate": 2.384538152610442e-06, "loss": 0.8206, "step": 1517 }, { "epoch": 2.286144578313253, "grad_norm": 0.390625, "learning_rate": 2.379518072289157e-06, "loss": 0.8131, "step": 1518 }, { "epoch": 2.287650602409639, "grad_norm": 0.34375, "learning_rate": 2.374497991967872e-06, "loss": 0.7945, "step": 1519 }, { "epoch": 2.289156626506024, "grad_norm": 0.255859375, "learning_rate": 2.3694779116465868e-06, "loss": 0.8116, "step": 1520 }, { "epoch": 2.2906626506024095, "grad_norm": 0.251953125, "learning_rate": 2.3644578313253013e-06, "loss": 0.8346, "step": 1521 }, { "epoch": 2.2921686746987953, "grad_norm": 0.255859375, "learning_rate": 2.3594377510040163e-06, "loss": 0.8331, "step": 1522 }, { "epoch": 2.2936746987951806, "grad_norm": 0.255859375, "learning_rate": 2.3544176706827312e-06, "loss": 0.8383, "step": 1523 }, { "epoch": 2.2951807228915664, "grad_norm": 0.271484375, "learning_rate": 2.349397590361446e-06, "loss": 0.8305, "step": 1524 }, { "epoch": 2.2966867469879517, "grad_norm": 0.2431640625, "learning_rate": 2.3443775100401607e-06, "loss": 0.8216, "step": 1525 }, { "epoch": 2.2981927710843375, "grad_norm": 0.283203125, "learning_rate": 2.3393574297188756e-06, "loss": 0.8375, "step": 1526 }, { "epoch": 2.299698795180723, "grad_norm": 0.25, "learning_rate": 2.3343373493975906e-06, "loss": 0.809, "step": 1527 }, { "epoch": 2.3012048192771086, "grad_norm": 0.248046875, "learning_rate": 2.3293172690763055e-06, "loss": 0.7866, "step": 1528 }, { "epoch": 2.302710843373494, "grad_norm": 0.26953125, "learning_rate": 2.3242971887550205e-06, "loss": 0.8124, "step": 1529 }, { "epoch": 2.3042168674698793, "grad_norm": 0.24609375, "learning_rate": 2.319277108433735e-06, "loss": 0.8204, "step": 1530 }, { "epoch": 2.305722891566265, "grad_norm": 0.271484375, "learning_rate": 2.31425702811245e-06, "loss": 0.845, "step": 1531 }, { "epoch": 2.3072289156626504, "grad_norm": 0.2470703125, "learning_rate": 2.309236947791165e-06, "loss": 0.8331, "step": 1532 }, { "epoch": 2.308734939759036, "grad_norm": 0.34765625, "learning_rate": 2.30421686746988e-06, "loss": 0.793, "step": 1533 }, { "epoch": 2.3102409638554215, "grad_norm": 0.26171875, "learning_rate": 2.2991967871485944e-06, "loss": 0.7699, "step": 1534 }, { "epoch": 2.3117469879518073, "grad_norm": 0.296875, "learning_rate": 2.2941767068273093e-06, "loss": 0.8053, "step": 1535 }, { "epoch": 2.3132530120481927, "grad_norm": 0.251953125, "learning_rate": 2.2891566265060243e-06, "loss": 0.8409, "step": 1536 }, { "epoch": 2.3147590361445785, "grad_norm": 0.2578125, "learning_rate": 2.2841365461847392e-06, "loss": 0.8386, "step": 1537 }, { "epoch": 2.316265060240964, "grad_norm": 0.265625, "learning_rate": 2.279116465863454e-06, "loss": 0.8244, "step": 1538 }, { "epoch": 2.3177710843373496, "grad_norm": 0.251953125, "learning_rate": 2.2740963855421687e-06, "loss": 0.867, "step": 1539 }, { "epoch": 2.319277108433735, "grad_norm": 0.2578125, "learning_rate": 2.2690763052208837e-06, "loss": 0.8556, "step": 1540 }, { "epoch": 2.3207831325301207, "grad_norm": 0.298828125, "learning_rate": 2.2640562248995986e-06, "loss": 0.7904, "step": 1541 }, { "epoch": 2.322289156626506, "grad_norm": 0.26953125, "learning_rate": 2.2590361445783136e-06, "loss": 0.7821, "step": 1542 }, { "epoch": 2.3237951807228914, "grad_norm": 0.27734375, "learning_rate": 2.254016064257028e-06, "loss": 0.8536, "step": 1543 }, { "epoch": 2.325301204819277, "grad_norm": 0.2490234375, "learning_rate": 2.248995983935743e-06, "loss": 0.8445, "step": 1544 }, { "epoch": 2.3268072289156625, "grad_norm": 0.24609375, "learning_rate": 2.243975903614458e-06, "loss": 0.8285, "step": 1545 }, { "epoch": 2.3283132530120483, "grad_norm": 0.259765625, "learning_rate": 2.238955823293173e-06, "loss": 0.8422, "step": 1546 }, { "epoch": 2.3298192771084336, "grad_norm": 0.275390625, "learning_rate": 2.233935742971888e-06, "loss": 0.8269, "step": 1547 }, { "epoch": 2.3313253012048194, "grad_norm": 0.2470703125, "learning_rate": 2.2289156626506024e-06, "loss": 0.8403, "step": 1548 }, { "epoch": 2.3328313253012047, "grad_norm": 0.24609375, "learning_rate": 2.2238955823293174e-06, "loss": 0.791, "step": 1549 }, { "epoch": 2.3343373493975905, "grad_norm": 0.25390625, "learning_rate": 2.2188755020080323e-06, "loss": 0.8221, "step": 1550 }, { "epoch": 2.335843373493976, "grad_norm": 0.2578125, "learning_rate": 2.2138554216867473e-06, "loss": 0.8114, "step": 1551 }, { "epoch": 2.337349397590361, "grad_norm": 0.251953125, "learning_rate": 2.2088353413654622e-06, "loss": 0.8309, "step": 1552 }, { "epoch": 2.338855421686747, "grad_norm": 0.294921875, "learning_rate": 2.2038152610441768e-06, "loss": 0.8163, "step": 1553 }, { "epoch": 2.3403614457831328, "grad_norm": 0.255859375, "learning_rate": 2.1987951807228917e-06, "loss": 0.8246, "step": 1554 }, { "epoch": 2.341867469879518, "grad_norm": 0.24609375, "learning_rate": 2.1937751004016067e-06, "loss": 0.8442, "step": 1555 }, { "epoch": 2.3433734939759034, "grad_norm": 0.2490234375, "learning_rate": 2.1887550200803216e-06, "loss": 0.8313, "step": 1556 }, { "epoch": 2.3448795180722892, "grad_norm": 0.25390625, "learning_rate": 2.183734939759036e-06, "loss": 0.8318, "step": 1557 }, { "epoch": 2.3463855421686746, "grad_norm": 0.251953125, "learning_rate": 2.178714859437751e-06, "loss": 0.7915, "step": 1558 }, { "epoch": 2.3478915662650603, "grad_norm": 0.271484375, "learning_rate": 2.173694779116466e-06, "loss": 0.7789, "step": 1559 }, { "epoch": 2.3493975903614457, "grad_norm": 0.314453125, "learning_rate": 2.168674698795181e-06, "loss": 0.8059, "step": 1560 }, { "epoch": 2.3509036144578315, "grad_norm": 0.25390625, "learning_rate": 2.163654618473896e-06, "loss": 0.8519, "step": 1561 }, { "epoch": 2.352409638554217, "grad_norm": 0.265625, "learning_rate": 2.1586345381526105e-06, "loss": 0.8363, "step": 1562 }, { "epoch": 2.3539156626506026, "grad_norm": 0.2734375, "learning_rate": 2.1536144578313254e-06, "loss": 0.857, "step": 1563 }, { "epoch": 2.355421686746988, "grad_norm": 0.26171875, "learning_rate": 2.1485943775100404e-06, "loss": 0.7926, "step": 1564 }, { "epoch": 2.3569277108433733, "grad_norm": 0.25390625, "learning_rate": 2.1435742971887553e-06, "loss": 0.8396, "step": 1565 }, { "epoch": 2.358433734939759, "grad_norm": 0.26171875, "learning_rate": 2.13855421686747e-06, "loss": 0.8449, "step": 1566 }, { "epoch": 2.3599397590361444, "grad_norm": 0.255859375, "learning_rate": 2.1335341365461848e-06, "loss": 0.8545, "step": 1567 }, { "epoch": 2.36144578313253, "grad_norm": 0.279296875, "learning_rate": 2.1285140562248997e-06, "loss": 0.8286, "step": 1568 }, { "epoch": 2.3629518072289155, "grad_norm": 0.263671875, "learning_rate": 2.1234939759036147e-06, "loss": 0.8382, "step": 1569 }, { "epoch": 2.3644578313253013, "grad_norm": 0.3671875, "learning_rate": 2.1184738955823296e-06, "loss": 0.8071, "step": 1570 }, { "epoch": 2.3659638554216866, "grad_norm": 0.345703125, "learning_rate": 2.113453815261044e-06, "loss": 0.8489, "step": 1571 }, { "epoch": 2.3674698795180724, "grad_norm": 0.26953125, "learning_rate": 2.1084337349397595e-06, "loss": 0.8384, "step": 1572 }, { "epoch": 2.3689759036144578, "grad_norm": 0.2431640625, "learning_rate": 2.103413654618474e-06, "loss": 0.7965, "step": 1573 }, { "epoch": 2.3704819277108435, "grad_norm": 0.251953125, "learning_rate": 2.098393574297189e-06, "loss": 0.8172, "step": 1574 }, { "epoch": 2.371987951807229, "grad_norm": 0.267578125, "learning_rate": 2.0933734939759035e-06, "loss": 0.8559, "step": 1575 }, { "epoch": 2.3734939759036147, "grad_norm": 0.267578125, "learning_rate": 2.0883534136546185e-06, "loss": 0.833, "step": 1576 }, { "epoch": 2.375, "grad_norm": 0.251953125, "learning_rate": 2.0833333333333334e-06, "loss": 0.8312, "step": 1577 }, { "epoch": 2.3765060240963853, "grad_norm": 0.2578125, "learning_rate": 2.0783132530120484e-06, "loss": 0.8352, "step": 1578 }, { "epoch": 2.378012048192771, "grad_norm": 0.26171875, "learning_rate": 2.0732931726907633e-06, "loss": 0.8366, "step": 1579 }, { "epoch": 2.3795180722891565, "grad_norm": 0.2470703125, "learning_rate": 2.068273092369478e-06, "loss": 0.822, "step": 1580 }, { "epoch": 2.3810240963855422, "grad_norm": 0.2890625, "learning_rate": 2.0632530120481932e-06, "loss": 0.8241, "step": 1581 }, { "epoch": 2.3825301204819276, "grad_norm": 0.2470703125, "learning_rate": 2.0582329317269078e-06, "loss": 0.7911, "step": 1582 }, { "epoch": 2.3840361445783134, "grad_norm": 0.251953125, "learning_rate": 2.0532128514056227e-06, "loss": 0.8544, "step": 1583 }, { "epoch": 2.3855421686746987, "grad_norm": 0.25, "learning_rate": 2.0481927710843377e-06, "loss": 0.8135, "step": 1584 }, { "epoch": 2.3870481927710845, "grad_norm": 0.271484375, "learning_rate": 2.043172690763052e-06, "loss": 0.8376, "step": 1585 }, { "epoch": 2.38855421686747, "grad_norm": 0.259765625, "learning_rate": 2.038152610441767e-06, "loss": 0.8358, "step": 1586 }, { "epoch": 2.3900602409638556, "grad_norm": 0.2734375, "learning_rate": 2.033132530120482e-06, "loss": 0.8305, "step": 1587 }, { "epoch": 2.391566265060241, "grad_norm": 0.2578125, "learning_rate": 2.028112449799197e-06, "loss": 0.7977, "step": 1588 }, { "epoch": 2.3930722891566267, "grad_norm": 0.271484375, "learning_rate": 2.0230923694779116e-06, "loss": 0.8877, "step": 1589 }, { "epoch": 2.394578313253012, "grad_norm": 0.283203125, "learning_rate": 2.018072289156627e-06, "loss": 0.8527, "step": 1590 }, { "epoch": 2.3960843373493974, "grad_norm": 0.27734375, "learning_rate": 2.0130522088353415e-06, "loss": 0.8511, "step": 1591 }, { "epoch": 2.397590361445783, "grad_norm": 0.255859375, "learning_rate": 2.0080321285140564e-06, "loss": 0.805, "step": 1592 }, { "epoch": 2.3990963855421685, "grad_norm": 0.248046875, "learning_rate": 2.0030120481927714e-06, "loss": 0.814, "step": 1593 }, { "epoch": 2.4006024096385543, "grad_norm": 0.349609375, "learning_rate": 1.997991967871486e-06, "loss": 0.8013, "step": 1594 }, { "epoch": 2.4021084337349397, "grad_norm": 0.25, "learning_rate": 1.992971887550201e-06, "loss": 0.8337, "step": 1595 }, { "epoch": 2.4036144578313254, "grad_norm": 0.34375, "learning_rate": 1.987951807228916e-06, "loss": 0.8315, "step": 1596 }, { "epoch": 2.4051204819277108, "grad_norm": 0.39453125, "learning_rate": 1.9829317269076307e-06, "loss": 0.7854, "step": 1597 }, { "epoch": 2.4066265060240966, "grad_norm": 0.25, "learning_rate": 1.9779116465863453e-06, "loss": 0.8416, "step": 1598 }, { "epoch": 2.408132530120482, "grad_norm": 0.25390625, "learning_rate": 1.9728915662650606e-06, "loss": 0.8264, "step": 1599 }, { "epoch": 2.4096385542168672, "grad_norm": 0.25, "learning_rate": 1.967871485943775e-06, "loss": 0.8211, "step": 1600 }, { "epoch": 2.411144578313253, "grad_norm": 0.26953125, "learning_rate": 1.96285140562249e-06, "loss": 0.7783, "step": 1601 }, { "epoch": 2.412650602409639, "grad_norm": 0.26171875, "learning_rate": 1.957831325301205e-06, "loss": 0.8469, "step": 1602 }, { "epoch": 2.414156626506024, "grad_norm": 0.298828125, "learning_rate": 1.95281124497992e-06, "loss": 0.8244, "step": 1603 }, { "epoch": 2.4156626506024095, "grad_norm": 0.26171875, "learning_rate": 1.947791164658635e-06, "loss": 0.8106, "step": 1604 }, { "epoch": 2.4171686746987953, "grad_norm": 0.51171875, "learning_rate": 1.9427710843373495e-06, "loss": 0.8445, "step": 1605 }, { "epoch": 2.4186746987951806, "grad_norm": 0.25390625, "learning_rate": 1.9377510040160644e-06, "loss": 0.8256, "step": 1606 }, { "epoch": 2.4201807228915664, "grad_norm": 0.2490234375, "learning_rate": 1.932730923694779e-06, "loss": 0.8256, "step": 1607 }, { "epoch": 2.4216867469879517, "grad_norm": 0.32421875, "learning_rate": 1.9277108433734943e-06, "loss": 0.8066, "step": 1608 }, { "epoch": 2.4231927710843375, "grad_norm": 0.25390625, "learning_rate": 1.922690763052209e-06, "loss": 0.8234, "step": 1609 }, { "epoch": 2.424698795180723, "grad_norm": 0.28125, "learning_rate": 1.917670682730924e-06, "loss": 0.8058, "step": 1610 }, { "epoch": 2.4262048192771086, "grad_norm": 0.30078125, "learning_rate": 1.9126506024096388e-06, "loss": 0.7898, "step": 1611 }, { "epoch": 2.427710843373494, "grad_norm": 0.3828125, "learning_rate": 1.9076305220883537e-06, "loss": 0.8439, "step": 1612 }, { "epoch": 2.4292168674698793, "grad_norm": 0.255859375, "learning_rate": 1.9026104417670685e-06, "loss": 0.8261, "step": 1613 }, { "epoch": 2.430722891566265, "grad_norm": 0.423828125, "learning_rate": 1.8975903614457832e-06, "loss": 0.803, "step": 1614 }, { "epoch": 2.4322289156626504, "grad_norm": 0.28125, "learning_rate": 1.8925702811244981e-06, "loss": 0.8604, "step": 1615 }, { "epoch": 2.433734939759036, "grad_norm": 0.263671875, "learning_rate": 1.8875502008032129e-06, "loss": 0.8425, "step": 1616 }, { "epoch": 2.4352409638554215, "grad_norm": 0.251953125, "learning_rate": 1.882530120481928e-06, "loss": 0.8107, "step": 1617 }, { "epoch": 2.4367469879518073, "grad_norm": 0.244140625, "learning_rate": 1.8775100401606428e-06, "loss": 0.817, "step": 1618 }, { "epoch": 2.4382530120481927, "grad_norm": 1.578125, "learning_rate": 1.8724899598393575e-06, "loss": 0.8064, "step": 1619 }, { "epoch": 2.4397590361445785, "grad_norm": 0.271484375, "learning_rate": 1.8674698795180723e-06, "loss": 0.8333, "step": 1620 }, { "epoch": 2.441265060240964, "grad_norm": 0.283203125, "learning_rate": 1.8624497991967874e-06, "loss": 0.8296, "step": 1621 }, { "epoch": 2.4427710843373496, "grad_norm": 0.263671875, "learning_rate": 1.8574297188755022e-06, "loss": 0.855, "step": 1622 }, { "epoch": 2.444277108433735, "grad_norm": 0.263671875, "learning_rate": 1.8524096385542171e-06, "loss": 0.8318, "step": 1623 }, { "epoch": 2.4457831325301207, "grad_norm": 0.25, "learning_rate": 1.8473895582329318e-06, "loss": 0.7919, "step": 1624 }, { "epoch": 2.447289156626506, "grad_norm": 0.2421875, "learning_rate": 1.8423694779116466e-06, "loss": 0.8181, "step": 1625 }, { "epoch": 2.4487951807228914, "grad_norm": 0.2578125, "learning_rate": 1.8373493975903617e-06, "loss": 0.8756, "step": 1626 }, { "epoch": 2.450301204819277, "grad_norm": 0.294921875, "learning_rate": 1.8323293172690765e-06, "loss": 0.8552, "step": 1627 }, { "epoch": 2.4518072289156625, "grad_norm": 0.25390625, "learning_rate": 1.8273092369477912e-06, "loss": 0.8366, "step": 1628 }, { "epoch": 2.4533132530120483, "grad_norm": 0.26171875, "learning_rate": 1.8222891566265062e-06, "loss": 0.8277, "step": 1629 }, { "epoch": 2.4548192771084336, "grad_norm": 0.255859375, "learning_rate": 1.8172690763052211e-06, "loss": 0.7646, "step": 1630 }, { "epoch": 2.4563253012048194, "grad_norm": 0.2578125, "learning_rate": 1.8122489959839359e-06, "loss": 0.7943, "step": 1631 }, { "epoch": 2.4578313253012047, "grad_norm": 0.2412109375, "learning_rate": 1.8072289156626508e-06, "loss": 0.7999, "step": 1632 }, { "epoch": 2.4593373493975905, "grad_norm": 0.25, "learning_rate": 1.8022088353413656e-06, "loss": 0.8173, "step": 1633 }, { "epoch": 2.460843373493976, "grad_norm": 0.248046875, "learning_rate": 1.7971887550200803e-06, "loss": 0.8341, "step": 1634 }, { "epoch": 2.462349397590361, "grad_norm": 0.25390625, "learning_rate": 1.7921686746987955e-06, "loss": 0.8697, "step": 1635 }, { "epoch": 2.463855421686747, "grad_norm": 0.259765625, "learning_rate": 1.7871485943775102e-06, "loss": 0.8821, "step": 1636 }, { "epoch": 2.4653614457831328, "grad_norm": 0.24609375, "learning_rate": 1.782128514056225e-06, "loss": 0.7958, "step": 1637 }, { "epoch": 2.466867469879518, "grad_norm": 0.30859375, "learning_rate": 1.7771084337349399e-06, "loss": 0.8011, "step": 1638 }, { "epoch": 2.4683734939759034, "grad_norm": 0.267578125, "learning_rate": 1.7720883534136548e-06, "loss": 0.8087, "step": 1639 }, { "epoch": 2.4698795180722892, "grad_norm": 0.275390625, "learning_rate": 1.7670682730923696e-06, "loss": 0.8183, "step": 1640 }, { "epoch": 2.4713855421686746, "grad_norm": 0.263671875, "learning_rate": 1.7620481927710845e-06, "loss": 0.8348, "step": 1641 }, { "epoch": 2.4728915662650603, "grad_norm": 0.263671875, "learning_rate": 1.7570281124497993e-06, "loss": 0.8727, "step": 1642 }, { "epoch": 2.4743975903614457, "grad_norm": 0.27734375, "learning_rate": 1.7520080321285144e-06, "loss": 0.8211, "step": 1643 }, { "epoch": 2.4759036144578315, "grad_norm": 0.26953125, "learning_rate": 1.7469879518072292e-06, "loss": 0.8151, "step": 1644 }, { "epoch": 2.477409638554217, "grad_norm": 0.40234375, "learning_rate": 1.7419678714859439e-06, "loss": 0.8112, "step": 1645 }, { "epoch": 2.4789156626506026, "grad_norm": 0.443359375, "learning_rate": 1.7369477911646586e-06, "loss": 0.8526, "step": 1646 }, { "epoch": 2.480421686746988, "grad_norm": 0.2470703125, "learning_rate": 1.7319277108433736e-06, "loss": 0.8339, "step": 1647 }, { "epoch": 2.4819277108433733, "grad_norm": 0.259765625, "learning_rate": 1.7269076305220885e-06, "loss": 0.8275, "step": 1648 }, { "epoch": 2.483433734939759, "grad_norm": 0.25390625, "learning_rate": 1.7218875502008035e-06, "loss": 0.8192, "step": 1649 }, { "epoch": 2.4849397590361444, "grad_norm": 0.2578125, "learning_rate": 1.7168674698795182e-06, "loss": 0.8188, "step": 1650 }, { "epoch": 2.48644578313253, "grad_norm": 0.294921875, "learning_rate": 1.711847389558233e-06, "loss": 0.8378, "step": 1651 }, { "epoch": 2.4879518072289155, "grad_norm": 0.25, "learning_rate": 1.7068273092369481e-06, "loss": 0.807, "step": 1652 }, { "epoch": 2.4894578313253013, "grad_norm": 0.251953125, "learning_rate": 1.7018072289156629e-06, "loss": 0.8047, "step": 1653 }, { "epoch": 2.4909638554216866, "grad_norm": 0.251953125, "learning_rate": 1.6967871485943776e-06, "loss": 0.8345, "step": 1654 }, { "epoch": 2.4924698795180724, "grad_norm": 0.25, "learning_rate": 1.6917670682730925e-06, "loss": 0.8099, "step": 1655 }, { "epoch": 2.4939759036144578, "grad_norm": 0.24609375, "learning_rate": 1.6867469879518073e-06, "loss": 0.8614, "step": 1656 }, { "epoch": 2.4954819277108435, "grad_norm": 0.36328125, "learning_rate": 1.6817269076305222e-06, "loss": 0.8303, "step": 1657 }, { "epoch": 2.496987951807229, "grad_norm": 0.265625, "learning_rate": 1.6767068273092372e-06, "loss": 0.8313, "step": 1658 }, { "epoch": 2.4984939759036147, "grad_norm": 0.2470703125, "learning_rate": 1.671686746987952e-06, "loss": 0.8125, "step": 1659 }, { "epoch": 2.5, "grad_norm": 0.310546875, "learning_rate": 1.6666666666666667e-06, "loss": 0.8111, "step": 1660 }, { "epoch": 2.5015060240963853, "grad_norm": 0.29296875, "learning_rate": 1.6616465863453818e-06, "loss": 0.8395, "step": 1661 }, { "epoch": 2.503012048192771, "grad_norm": 0.2734375, "learning_rate": 1.6566265060240966e-06, "loss": 0.8006, "step": 1662 }, { "epoch": 2.5045180722891565, "grad_norm": 0.2578125, "learning_rate": 1.6516064257028113e-06, "loss": 0.798, "step": 1663 }, { "epoch": 2.5060240963855422, "grad_norm": 0.2490234375, "learning_rate": 1.6465863453815263e-06, "loss": 0.8363, "step": 1664 }, { "epoch": 2.5075301204819276, "grad_norm": 0.2431640625, "learning_rate": 1.641566265060241e-06, "loss": 0.8229, "step": 1665 }, { "epoch": 2.5090361445783134, "grad_norm": 0.25, "learning_rate": 1.6365461847389562e-06, "loss": 0.8713, "step": 1666 }, { "epoch": 2.5105421686746987, "grad_norm": 0.29296875, "learning_rate": 1.6315261044176709e-06, "loss": 0.849, "step": 1667 }, { "epoch": 2.5120481927710845, "grad_norm": 0.25390625, "learning_rate": 1.6265060240963856e-06, "loss": 0.8117, "step": 1668 }, { "epoch": 2.51355421686747, "grad_norm": 0.310546875, "learning_rate": 1.6214859437751004e-06, "loss": 0.8378, "step": 1669 }, { "epoch": 2.515060240963855, "grad_norm": 0.2490234375, "learning_rate": 1.6164658634538155e-06, "loss": 0.7969, "step": 1670 }, { "epoch": 2.516566265060241, "grad_norm": 0.357421875, "learning_rate": 1.6114457831325303e-06, "loss": 0.8051, "step": 1671 }, { "epoch": 2.5180722891566267, "grad_norm": 0.251953125, "learning_rate": 1.606425702811245e-06, "loss": 0.8303, "step": 1672 }, { "epoch": 2.519578313253012, "grad_norm": 0.25390625, "learning_rate": 1.60140562248996e-06, "loss": 0.8436, "step": 1673 }, { "epoch": 2.5210843373493974, "grad_norm": 0.318359375, "learning_rate": 1.5963855421686747e-06, "loss": 0.8039, "step": 1674 }, { "epoch": 2.522590361445783, "grad_norm": 0.291015625, "learning_rate": 1.5913654618473899e-06, "loss": 0.8093, "step": 1675 }, { "epoch": 2.5240963855421685, "grad_norm": 0.32421875, "learning_rate": 1.5863453815261046e-06, "loss": 0.8133, "step": 1676 }, { "epoch": 2.5256024096385543, "grad_norm": 0.375, "learning_rate": 1.5813253012048193e-06, "loss": 0.8275, "step": 1677 }, { "epoch": 2.5271084337349397, "grad_norm": 0.25390625, "learning_rate": 1.576305220883534e-06, "loss": 0.8171, "step": 1678 }, { "epoch": 2.5286144578313254, "grad_norm": 0.73046875, "learning_rate": 1.5712851405622492e-06, "loss": 0.8346, "step": 1679 }, { "epoch": 2.5301204819277108, "grad_norm": 0.259765625, "learning_rate": 1.566265060240964e-06, "loss": 0.8059, "step": 1680 }, { "epoch": 2.5316265060240966, "grad_norm": 0.333984375, "learning_rate": 1.561244979919679e-06, "loss": 0.7883, "step": 1681 }, { "epoch": 2.533132530120482, "grad_norm": 0.25390625, "learning_rate": 1.5562248995983937e-06, "loss": 0.8394, "step": 1682 }, { "epoch": 2.5346385542168672, "grad_norm": 0.24609375, "learning_rate": 1.5512048192771086e-06, "loss": 0.7901, "step": 1683 }, { "epoch": 2.536144578313253, "grad_norm": 0.25390625, "learning_rate": 1.5461847389558236e-06, "loss": 0.8435, "step": 1684 }, { "epoch": 2.537650602409639, "grad_norm": 0.251953125, "learning_rate": 1.5411646586345383e-06, "loss": 0.855, "step": 1685 }, { "epoch": 2.539156626506024, "grad_norm": 0.26953125, "learning_rate": 1.536144578313253e-06, "loss": 0.827, "step": 1686 }, { "epoch": 2.5406626506024095, "grad_norm": 0.275390625, "learning_rate": 1.531124497991968e-06, "loss": 0.8064, "step": 1687 }, { "epoch": 2.5421686746987953, "grad_norm": 0.25390625, "learning_rate": 1.526104417670683e-06, "loss": 0.8361, "step": 1688 }, { "epoch": 2.5436746987951806, "grad_norm": 0.240234375, "learning_rate": 1.5210843373493977e-06, "loss": 0.8024, "step": 1689 }, { "epoch": 2.5451807228915664, "grad_norm": 0.291015625, "learning_rate": 1.5160642570281126e-06, "loss": 0.85, "step": 1690 }, { "epoch": 2.5466867469879517, "grad_norm": 0.255859375, "learning_rate": 1.5110441767068274e-06, "loss": 0.8629, "step": 1691 }, { "epoch": 2.5481927710843375, "grad_norm": 0.25390625, "learning_rate": 1.5060240963855425e-06, "loss": 0.8087, "step": 1692 }, { "epoch": 2.549698795180723, "grad_norm": 0.3203125, "learning_rate": 1.5010040160642573e-06, "loss": 0.8221, "step": 1693 }, { "epoch": 2.5512048192771086, "grad_norm": 0.2490234375, "learning_rate": 1.495983935742972e-06, "loss": 0.8288, "step": 1694 }, { "epoch": 2.552710843373494, "grad_norm": 0.25390625, "learning_rate": 1.4909638554216867e-06, "loss": 0.8021, "step": 1695 }, { "epoch": 2.5542168674698793, "grad_norm": 0.25390625, "learning_rate": 1.4859437751004017e-06, "loss": 0.7927, "step": 1696 }, { "epoch": 2.555722891566265, "grad_norm": 0.310546875, "learning_rate": 1.4809236947791166e-06, "loss": 0.8006, "step": 1697 }, { "epoch": 2.557228915662651, "grad_norm": 0.2578125, "learning_rate": 1.4759036144578314e-06, "loss": 0.8648, "step": 1698 }, { "epoch": 2.558734939759036, "grad_norm": 0.27734375, "learning_rate": 1.4708835341365463e-06, "loss": 0.8477, "step": 1699 }, { "epoch": 2.5602409638554215, "grad_norm": 0.259765625, "learning_rate": 1.465863453815261e-06, "loss": 0.8493, "step": 1700 }, { "epoch": 2.5617469879518073, "grad_norm": 0.2578125, "learning_rate": 1.4608433734939762e-06, "loss": 0.8075, "step": 1701 }, { "epoch": 2.5632530120481927, "grad_norm": 0.26171875, "learning_rate": 1.455823293172691e-06, "loss": 0.8186, "step": 1702 }, { "epoch": 2.5647590361445785, "grad_norm": 0.248046875, "learning_rate": 1.4508032128514057e-06, "loss": 0.8207, "step": 1703 }, { "epoch": 2.566265060240964, "grad_norm": 0.265625, "learning_rate": 1.4457831325301204e-06, "loss": 0.8613, "step": 1704 }, { "epoch": 2.567771084337349, "grad_norm": 0.24609375, "learning_rate": 1.4407630522088354e-06, "loss": 0.8082, "step": 1705 }, { "epoch": 2.569277108433735, "grad_norm": 0.25390625, "learning_rate": 1.4357429718875503e-06, "loss": 0.8138, "step": 1706 }, { "epoch": 2.5707831325301207, "grad_norm": 0.2470703125, "learning_rate": 1.4307228915662653e-06, "loss": 0.8356, "step": 1707 }, { "epoch": 2.572289156626506, "grad_norm": 0.361328125, "learning_rate": 1.42570281124498e-06, "loss": 0.833, "step": 1708 }, { "epoch": 2.5737951807228914, "grad_norm": 0.265625, "learning_rate": 1.4206827309236948e-06, "loss": 0.8497, "step": 1709 }, { "epoch": 2.575301204819277, "grad_norm": 0.251953125, "learning_rate": 1.41566265060241e-06, "loss": 0.8375, "step": 1710 }, { "epoch": 2.5768072289156625, "grad_norm": 0.26171875, "learning_rate": 1.4106425702811247e-06, "loss": 0.8334, "step": 1711 }, { "epoch": 2.5783132530120483, "grad_norm": 0.3203125, "learning_rate": 1.4056224899598394e-06, "loss": 0.7976, "step": 1712 }, { "epoch": 2.5798192771084336, "grad_norm": 0.28125, "learning_rate": 1.4006024096385544e-06, "loss": 0.8529, "step": 1713 }, { "epoch": 2.5813253012048194, "grad_norm": 0.2490234375, "learning_rate": 1.395582329317269e-06, "loss": 0.801, "step": 1714 }, { "epoch": 2.5828313253012047, "grad_norm": 0.2490234375, "learning_rate": 1.390562248995984e-06, "loss": 0.8061, "step": 1715 }, { "epoch": 2.5843373493975905, "grad_norm": 0.259765625, "learning_rate": 1.385542168674699e-06, "loss": 0.857, "step": 1716 }, { "epoch": 2.585843373493976, "grad_norm": 0.259765625, "learning_rate": 1.3805220883534137e-06, "loss": 0.8119, "step": 1717 }, { "epoch": 2.587349397590361, "grad_norm": 0.259765625, "learning_rate": 1.3755020080321285e-06, "loss": 0.8448, "step": 1718 }, { "epoch": 2.588855421686747, "grad_norm": 0.30859375, "learning_rate": 1.3704819277108436e-06, "loss": 0.8255, "step": 1719 }, { "epoch": 2.5903614457831328, "grad_norm": 0.271484375, "learning_rate": 1.3654618473895584e-06, "loss": 0.823, "step": 1720 }, { "epoch": 2.591867469879518, "grad_norm": 0.2451171875, "learning_rate": 1.3604417670682731e-06, "loss": 0.829, "step": 1721 }, { "epoch": 2.5933734939759034, "grad_norm": 0.27734375, "learning_rate": 1.355421686746988e-06, "loss": 0.8282, "step": 1722 }, { "epoch": 2.5948795180722892, "grad_norm": 0.265625, "learning_rate": 1.350401606425703e-06, "loss": 0.8208, "step": 1723 }, { "epoch": 2.5963855421686746, "grad_norm": 0.2578125, "learning_rate": 1.345381526104418e-06, "loss": 0.8508, "step": 1724 }, { "epoch": 2.5978915662650603, "grad_norm": 0.2421875, "learning_rate": 1.3403614457831327e-06, "loss": 0.8128, "step": 1725 }, { "epoch": 2.5993975903614457, "grad_norm": 0.244140625, "learning_rate": 1.3353413654618474e-06, "loss": 0.8364, "step": 1726 }, { "epoch": 2.6009036144578315, "grad_norm": 0.265625, "learning_rate": 1.3303212851405622e-06, "loss": 0.8108, "step": 1727 }, { "epoch": 2.602409638554217, "grad_norm": 0.32421875, "learning_rate": 1.3253012048192773e-06, "loss": 0.8254, "step": 1728 }, { "epoch": 2.6039156626506026, "grad_norm": 0.251953125, "learning_rate": 1.320281124497992e-06, "loss": 0.8386, "step": 1729 }, { "epoch": 2.605421686746988, "grad_norm": 0.279296875, "learning_rate": 1.3152610441767068e-06, "loss": 0.8335, "step": 1730 }, { "epoch": 2.6069277108433733, "grad_norm": 0.251953125, "learning_rate": 1.3102409638554218e-06, "loss": 0.8167, "step": 1731 }, { "epoch": 2.608433734939759, "grad_norm": 0.259765625, "learning_rate": 1.3052208835341367e-06, "loss": 0.865, "step": 1732 }, { "epoch": 2.609939759036145, "grad_norm": 0.251953125, "learning_rate": 1.3002008032128517e-06, "loss": 0.8334, "step": 1733 }, { "epoch": 2.61144578313253, "grad_norm": 0.25390625, "learning_rate": 1.2951807228915664e-06, "loss": 0.8197, "step": 1734 }, { "epoch": 2.6129518072289155, "grad_norm": 0.2353515625, "learning_rate": 1.2901606425702811e-06, "loss": 0.8057, "step": 1735 }, { "epoch": 2.6144578313253013, "grad_norm": 0.26171875, "learning_rate": 1.2851405622489959e-06, "loss": 0.8103, "step": 1736 }, { "epoch": 2.6159638554216866, "grad_norm": 0.255859375, "learning_rate": 1.280120481927711e-06, "loss": 0.7884, "step": 1737 }, { "epoch": 2.6174698795180724, "grad_norm": 0.2490234375, "learning_rate": 1.2751004016064258e-06, "loss": 0.8285, "step": 1738 }, { "epoch": 2.6189759036144578, "grad_norm": 0.255859375, "learning_rate": 1.2700803212851407e-06, "loss": 0.825, "step": 1739 }, { "epoch": 2.6204819277108435, "grad_norm": 0.263671875, "learning_rate": 1.2650602409638555e-06, "loss": 0.8473, "step": 1740 }, { "epoch": 2.621987951807229, "grad_norm": 0.2578125, "learning_rate": 1.2600401606425704e-06, "loss": 0.8194, "step": 1741 }, { "epoch": 2.6234939759036147, "grad_norm": 0.248046875, "learning_rate": 1.2550200803212854e-06, "loss": 0.8411, "step": 1742 }, { "epoch": 2.625, "grad_norm": 0.26171875, "learning_rate": 1.25e-06, "loss": 0.8427, "step": 1743 }, { "epoch": 2.6265060240963853, "grad_norm": 0.578125, "learning_rate": 1.2449799196787148e-06, "loss": 0.8666, "step": 1744 }, { "epoch": 2.628012048192771, "grad_norm": 0.2490234375, "learning_rate": 1.2399598393574298e-06, "loss": 0.7911, "step": 1745 }, { "epoch": 2.6295180722891565, "grad_norm": 0.404296875, "learning_rate": 1.2349397590361445e-06, "loss": 0.8038, "step": 1746 }, { "epoch": 2.6310240963855422, "grad_norm": 0.259765625, "learning_rate": 1.2299196787148595e-06, "loss": 0.8449, "step": 1747 }, { "epoch": 2.6325301204819276, "grad_norm": 0.2490234375, "learning_rate": 1.2248995983935744e-06, "loss": 0.8454, "step": 1748 }, { "epoch": 2.6340361445783134, "grad_norm": 0.251953125, "learning_rate": 1.2198795180722894e-06, "loss": 0.8515, "step": 1749 }, { "epoch": 2.6355421686746987, "grad_norm": 0.263671875, "learning_rate": 1.2148594377510041e-06, "loss": 0.8346, "step": 1750 }, { "epoch": 2.6370481927710845, "grad_norm": 0.34765625, "learning_rate": 1.209839357429719e-06, "loss": 0.8103, "step": 1751 }, { "epoch": 2.63855421686747, "grad_norm": 0.2578125, "learning_rate": 1.2048192771084338e-06, "loss": 0.8678, "step": 1752 }, { "epoch": 2.640060240963855, "grad_norm": 0.25, "learning_rate": 1.1997991967871485e-06, "loss": 0.8271, "step": 1753 }, { "epoch": 2.641566265060241, "grad_norm": 0.2578125, "learning_rate": 1.1947791164658635e-06, "loss": 0.8366, "step": 1754 }, { "epoch": 2.6430722891566267, "grad_norm": 0.255859375, "learning_rate": 1.1897590361445784e-06, "loss": 0.8083, "step": 1755 }, { "epoch": 2.644578313253012, "grad_norm": 0.2578125, "learning_rate": 1.1847389558232934e-06, "loss": 0.8228, "step": 1756 }, { "epoch": 2.6460843373493974, "grad_norm": 0.25, "learning_rate": 1.1797188755020081e-06, "loss": 0.8339, "step": 1757 }, { "epoch": 2.647590361445783, "grad_norm": 0.34375, "learning_rate": 1.174698795180723e-06, "loss": 0.8494, "step": 1758 }, { "epoch": 2.6490963855421685, "grad_norm": 0.255859375, "learning_rate": 1.1696787148594378e-06, "loss": 0.8067, "step": 1759 }, { "epoch": 2.6506024096385543, "grad_norm": 0.2470703125, "learning_rate": 1.1646586345381528e-06, "loss": 0.8041, "step": 1760 }, { "epoch": 2.6521084337349397, "grad_norm": 0.267578125, "learning_rate": 1.1596385542168675e-06, "loss": 0.8032, "step": 1761 }, { "epoch": 2.6536144578313254, "grad_norm": 0.255859375, "learning_rate": 1.1546184738955825e-06, "loss": 0.866, "step": 1762 }, { "epoch": 2.6551204819277108, "grad_norm": 0.24609375, "learning_rate": 1.1495983935742972e-06, "loss": 0.8008, "step": 1763 }, { "epoch": 2.6566265060240966, "grad_norm": 0.2451171875, "learning_rate": 1.1445783132530121e-06, "loss": 0.786, "step": 1764 }, { "epoch": 2.658132530120482, "grad_norm": 0.26953125, "learning_rate": 1.139558232931727e-06, "loss": 0.8447, "step": 1765 }, { "epoch": 2.6596385542168672, "grad_norm": 0.251953125, "learning_rate": 1.1345381526104418e-06, "loss": 0.8256, "step": 1766 }, { "epoch": 2.661144578313253, "grad_norm": 0.2578125, "learning_rate": 1.1295180722891568e-06, "loss": 0.8104, "step": 1767 }, { "epoch": 2.662650602409639, "grad_norm": 0.26171875, "learning_rate": 1.1244979919678715e-06, "loss": 0.8768, "step": 1768 }, { "epoch": 2.664156626506024, "grad_norm": 0.2412109375, "learning_rate": 1.1194779116465865e-06, "loss": 0.826, "step": 1769 }, { "epoch": 2.6656626506024095, "grad_norm": 0.255859375, "learning_rate": 1.1144578313253012e-06, "loss": 0.7851, "step": 1770 }, { "epoch": 2.6671686746987953, "grad_norm": 0.25390625, "learning_rate": 1.1094377510040162e-06, "loss": 0.8454, "step": 1771 }, { "epoch": 2.6686746987951806, "grad_norm": 0.255859375, "learning_rate": 1.1044176706827311e-06, "loss": 0.8298, "step": 1772 }, { "epoch": 2.6701807228915664, "grad_norm": 0.28515625, "learning_rate": 1.0993975903614459e-06, "loss": 0.838, "step": 1773 }, { "epoch": 2.6716867469879517, "grad_norm": 0.251953125, "learning_rate": 1.0943775100401608e-06, "loss": 0.7625, "step": 1774 }, { "epoch": 2.6731927710843375, "grad_norm": 0.251953125, "learning_rate": 1.0893574297188755e-06, "loss": 0.8068, "step": 1775 }, { "epoch": 2.674698795180723, "grad_norm": 0.267578125, "learning_rate": 1.0843373493975905e-06, "loss": 0.8349, "step": 1776 }, { "epoch": 2.6762048192771086, "grad_norm": 0.2470703125, "learning_rate": 1.0793172690763052e-06, "loss": 0.823, "step": 1777 }, { "epoch": 2.677710843373494, "grad_norm": 0.25, "learning_rate": 1.0742971887550202e-06, "loss": 0.8179, "step": 1778 }, { "epoch": 2.6792168674698793, "grad_norm": 0.255859375, "learning_rate": 1.069277108433735e-06, "loss": 0.8884, "step": 1779 }, { "epoch": 2.680722891566265, "grad_norm": 0.244140625, "learning_rate": 1.0642570281124499e-06, "loss": 0.8052, "step": 1780 }, { "epoch": 2.682228915662651, "grad_norm": 0.2451171875, "learning_rate": 1.0592369477911648e-06, "loss": 0.8107, "step": 1781 }, { "epoch": 2.683734939759036, "grad_norm": 0.2578125, "learning_rate": 1.0542168674698798e-06, "loss": 0.8451, "step": 1782 }, { "epoch": 2.6852409638554215, "grad_norm": 0.3046875, "learning_rate": 1.0491967871485945e-06, "loss": 0.8231, "step": 1783 }, { "epoch": 2.6867469879518073, "grad_norm": 0.25, "learning_rate": 1.0441767068273092e-06, "loss": 0.8128, "step": 1784 }, { "epoch": 2.6882530120481927, "grad_norm": 0.259765625, "learning_rate": 1.0391566265060242e-06, "loss": 0.7989, "step": 1785 }, { "epoch": 2.6897590361445785, "grad_norm": 0.26171875, "learning_rate": 1.034136546184739e-06, "loss": 0.8493, "step": 1786 }, { "epoch": 2.691265060240964, "grad_norm": 0.35546875, "learning_rate": 1.0291164658634539e-06, "loss": 0.7807, "step": 1787 }, { "epoch": 2.692771084337349, "grad_norm": 0.25390625, "learning_rate": 1.0240963855421688e-06, "loss": 0.7937, "step": 1788 }, { "epoch": 2.694277108433735, "grad_norm": 0.251953125, "learning_rate": 1.0190763052208836e-06, "loss": 0.8017, "step": 1789 }, { "epoch": 2.6957831325301207, "grad_norm": 0.259765625, "learning_rate": 1.0140562248995985e-06, "loss": 0.8523, "step": 1790 }, { "epoch": 2.697289156626506, "grad_norm": 0.24609375, "learning_rate": 1.0090361445783135e-06, "loss": 0.8305, "step": 1791 }, { "epoch": 2.6987951807228914, "grad_norm": 0.259765625, "learning_rate": 1.0040160642570282e-06, "loss": 0.8167, "step": 1792 }, { "epoch": 2.700301204819277, "grad_norm": 0.271484375, "learning_rate": 9.98995983935743e-07, "loss": 0.8184, "step": 1793 }, { "epoch": 2.7018072289156625, "grad_norm": 0.2490234375, "learning_rate": 9.93975903614458e-07, "loss": 0.8445, "step": 1794 }, { "epoch": 2.7033132530120483, "grad_norm": 0.2734375, "learning_rate": 9.889558232931726e-07, "loss": 0.8241, "step": 1795 }, { "epoch": 2.7048192771084336, "grad_norm": 0.255859375, "learning_rate": 9.839357429718876e-07, "loss": 0.7923, "step": 1796 }, { "epoch": 2.7063253012048194, "grad_norm": 0.244140625, "learning_rate": 9.789156626506025e-07, "loss": 0.8499, "step": 1797 }, { "epoch": 2.7078313253012047, "grad_norm": 0.2412109375, "learning_rate": 9.738955823293175e-07, "loss": 0.7868, "step": 1798 }, { "epoch": 2.7093373493975905, "grad_norm": 0.255859375, "learning_rate": 9.688755020080322e-07, "loss": 0.7927, "step": 1799 }, { "epoch": 2.710843373493976, "grad_norm": 0.34375, "learning_rate": 9.638554216867472e-07, "loss": 0.7868, "step": 1800 }, { "epoch": 2.712349397590361, "grad_norm": 0.267578125, "learning_rate": 9.58835341365462e-07, "loss": 0.8084, "step": 1801 }, { "epoch": 2.713855421686747, "grad_norm": 0.25, "learning_rate": 9.538152610441769e-07, "loss": 0.8296, "step": 1802 }, { "epoch": 2.7153614457831328, "grad_norm": 0.2470703125, "learning_rate": 9.487951807228916e-07, "loss": 0.8379, "step": 1803 }, { "epoch": 2.716867469879518, "grad_norm": 0.2431640625, "learning_rate": 9.437751004016064e-07, "loss": 0.7967, "step": 1804 }, { "epoch": 2.7183734939759034, "grad_norm": 0.271484375, "learning_rate": 9.387550200803214e-07, "loss": 0.8241, "step": 1805 }, { "epoch": 2.7198795180722892, "grad_norm": 0.2451171875, "learning_rate": 9.337349397590361e-07, "loss": 0.8072, "step": 1806 }, { "epoch": 2.7213855421686746, "grad_norm": 0.2890625, "learning_rate": 9.287148594377511e-07, "loss": 0.8153, "step": 1807 }, { "epoch": 2.7228915662650603, "grad_norm": 0.24609375, "learning_rate": 9.236947791164659e-07, "loss": 0.8186, "step": 1808 }, { "epoch": 2.7243975903614457, "grad_norm": 0.279296875, "learning_rate": 9.186746987951809e-07, "loss": 0.8089, "step": 1809 }, { "epoch": 2.7259036144578315, "grad_norm": 0.2578125, "learning_rate": 9.136546184738956e-07, "loss": 0.8134, "step": 1810 }, { "epoch": 2.727409638554217, "grad_norm": 0.2470703125, "learning_rate": 9.086345381526106e-07, "loss": 0.796, "step": 1811 }, { "epoch": 2.7289156626506026, "grad_norm": 0.24609375, "learning_rate": 9.036144578313254e-07, "loss": 0.8123, "step": 1812 }, { "epoch": 2.730421686746988, "grad_norm": 0.259765625, "learning_rate": 8.985943775100401e-07, "loss": 0.831, "step": 1813 }, { "epoch": 2.7319277108433733, "grad_norm": 0.259765625, "learning_rate": 8.935742971887551e-07, "loss": 0.7875, "step": 1814 }, { "epoch": 2.733433734939759, "grad_norm": 0.306640625, "learning_rate": 8.885542168674699e-07, "loss": 0.8541, "step": 1815 }, { "epoch": 2.734939759036145, "grad_norm": 0.25390625, "learning_rate": 8.835341365461848e-07, "loss": 0.839, "step": 1816 }, { "epoch": 2.73644578313253, "grad_norm": 0.279296875, "learning_rate": 8.785140562248996e-07, "loss": 0.8184, "step": 1817 }, { "epoch": 2.7379518072289155, "grad_norm": 0.267578125, "learning_rate": 8.734939759036146e-07, "loss": 0.7933, "step": 1818 }, { "epoch": 2.7394578313253013, "grad_norm": 0.3046875, "learning_rate": 8.684738955823293e-07, "loss": 0.8605, "step": 1819 }, { "epoch": 2.7409638554216866, "grad_norm": 0.24609375, "learning_rate": 8.634538152610443e-07, "loss": 0.8345, "step": 1820 }, { "epoch": 2.7424698795180724, "grad_norm": 0.263671875, "learning_rate": 8.584337349397591e-07, "loss": 0.8134, "step": 1821 }, { "epoch": 2.7439759036144578, "grad_norm": 0.294921875, "learning_rate": 8.534136546184741e-07, "loss": 0.8179, "step": 1822 }, { "epoch": 2.7454819277108435, "grad_norm": 0.25390625, "learning_rate": 8.483935742971888e-07, "loss": 0.8108, "step": 1823 }, { "epoch": 2.746987951807229, "grad_norm": 0.25390625, "learning_rate": 8.433734939759036e-07, "loss": 0.784, "step": 1824 }, { "epoch": 2.7484939759036147, "grad_norm": 0.3359375, "learning_rate": 8.383534136546186e-07, "loss": 0.8592, "step": 1825 }, { "epoch": 2.75, "grad_norm": 0.341796875, "learning_rate": 8.333333333333333e-07, "loss": 0.852, "step": 1826 }, { "epoch": 2.7515060240963853, "grad_norm": 0.3046875, "learning_rate": 8.283132530120483e-07, "loss": 0.8471, "step": 1827 }, { "epoch": 2.753012048192771, "grad_norm": 0.2490234375, "learning_rate": 8.232931726907631e-07, "loss": 0.8165, "step": 1828 }, { "epoch": 2.7545180722891565, "grad_norm": 0.30078125, "learning_rate": 8.182730923694781e-07, "loss": 0.8338, "step": 1829 }, { "epoch": 2.7560240963855422, "grad_norm": 0.26171875, "learning_rate": 8.132530120481928e-07, "loss": 0.8212, "step": 1830 }, { "epoch": 2.7575301204819276, "grad_norm": 0.263671875, "learning_rate": 8.082329317269078e-07, "loss": 0.8424, "step": 1831 }, { "epoch": 2.7590361445783134, "grad_norm": 0.2451171875, "learning_rate": 8.032128514056225e-07, "loss": 0.8033, "step": 1832 }, { "epoch": 2.7605421686746987, "grad_norm": 0.267578125, "learning_rate": 7.981927710843373e-07, "loss": 0.8168, "step": 1833 }, { "epoch": 2.7620481927710845, "grad_norm": 0.251953125, "learning_rate": 7.931726907630523e-07, "loss": 0.8373, "step": 1834 }, { "epoch": 2.76355421686747, "grad_norm": 0.244140625, "learning_rate": 7.88152610441767e-07, "loss": 0.8538, "step": 1835 }, { "epoch": 2.765060240963855, "grad_norm": 0.2490234375, "learning_rate": 7.83132530120482e-07, "loss": 0.7706, "step": 1836 }, { "epoch": 2.766566265060241, "grad_norm": 0.275390625, "learning_rate": 7.781124497991968e-07, "loss": 0.7958, "step": 1837 }, { "epoch": 2.7680722891566267, "grad_norm": 0.2578125, "learning_rate": 7.730923694779118e-07, "loss": 0.8079, "step": 1838 }, { "epoch": 2.769578313253012, "grad_norm": 0.267578125, "learning_rate": 7.680722891566265e-07, "loss": 0.8111, "step": 1839 }, { "epoch": 2.7710843373493974, "grad_norm": 0.27734375, "learning_rate": 7.630522088353415e-07, "loss": 0.8244, "step": 1840 }, { "epoch": 2.772590361445783, "grad_norm": 0.248046875, "learning_rate": 7.580321285140563e-07, "loss": 0.8465, "step": 1841 }, { "epoch": 2.7740963855421685, "grad_norm": 0.283203125, "learning_rate": 7.530120481927713e-07, "loss": 0.8335, "step": 1842 }, { "epoch": 2.7756024096385543, "grad_norm": 0.24609375, "learning_rate": 7.47991967871486e-07, "loss": 0.8423, "step": 1843 }, { "epoch": 2.7771084337349397, "grad_norm": 0.259765625, "learning_rate": 7.429718875502008e-07, "loss": 0.8214, "step": 1844 }, { "epoch": 2.7786144578313254, "grad_norm": 0.31640625, "learning_rate": 7.379518072289157e-07, "loss": 0.8088, "step": 1845 }, { "epoch": 2.7801204819277108, "grad_norm": 0.296875, "learning_rate": 7.329317269076305e-07, "loss": 0.8272, "step": 1846 }, { "epoch": 2.7816265060240966, "grad_norm": 0.255859375, "learning_rate": 7.279116465863455e-07, "loss": 0.8394, "step": 1847 }, { "epoch": 2.783132530120482, "grad_norm": 0.2421875, "learning_rate": 7.228915662650602e-07, "loss": 0.8196, "step": 1848 }, { "epoch": 2.7846385542168672, "grad_norm": 0.337890625, "learning_rate": 7.178714859437752e-07, "loss": 0.8129, "step": 1849 }, { "epoch": 2.786144578313253, "grad_norm": 0.25, "learning_rate": 7.1285140562249e-07, "loss": 0.8246, "step": 1850 }, { "epoch": 2.787650602409639, "grad_norm": 0.279296875, "learning_rate": 7.07831325301205e-07, "loss": 0.7964, "step": 1851 }, { "epoch": 2.789156626506024, "grad_norm": 0.26171875, "learning_rate": 7.028112449799197e-07, "loss": 0.8505, "step": 1852 }, { "epoch": 2.7906626506024095, "grad_norm": 0.25390625, "learning_rate": 6.977911646586345e-07, "loss": 0.8438, "step": 1853 }, { "epoch": 2.7921686746987953, "grad_norm": 0.341796875, "learning_rate": 6.927710843373495e-07, "loss": 0.8454, "step": 1854 }, { "epoch": 2.7936746987951806, "grad_norm": 0.251953125, "learning_rate": 6.877510040160642e-07, "loss": 0.8331, "step": 1855 }, { "epoch": 2.7951807228915664, "grad_norm": 0.251953125, "learning_rate": 6.827309236947792e-07, "loss": 0.8375, "step": 1856 }, { "epoch": 2.7966867469879517, "grad_norm": 0.263671875, "learning_rate": 6.77710843373494e-07, "loss": 0.8485, "step": 1857 }, { "epoch": 2.7981927710843375, "grad_norm": 0.291015625, "learning_rate": 6.72690763052209e-07, "loss": 0.7911, "step": 1858 }, { "epoch": 2.799698795180723, "grad_norm": 0.24609375, "learning_rate": 6.676706827309237e-07, "loss": 0.8403, "step": 1859 }, { "epoch": 2.8012048192771086, "grad_norm": 0.359375, "learning_rate": 6.626506024096387e-07, "loss": 0.8208, "step": 1860 }, { "epoch": 2.802710843373494, "grad_norm": 0.265625, "learning_rate": 6.576305220883534e-07, "loss": 0.8136, "step": 1861 }, { "epoch": 2.8042168674698793, "grad_norm": 0.2734375, "learning_rate": 6.526104417670684e-07, "loss": 0.8143, "step": 1862 }, { "epoch": 2.805722891566265, "grad_norm": 0.2470703125, "learning_rate": 6.475903614457832e-07, "loss": 0.8188, "step": 1863 }, { "epoch": 2.807228915662651, "grad_norm": 0.251953125, "learning_rate": 6.425702811244979e-07, "loss": 0.8108, "step": 1864 }, { "epoch": 2.808734939759036, "grad_norm": 0.248046875, "learning_rate": 6.375502008032129e-07, "loss": 0.7955, "step": 1865 }, { "epoch": 2.8102409638554215, "grad_norm": 0.255859375, "learning_rate": 6.325301204819277e-07, "loss": 0.8428, "step": 1866 }, { "epoch": 2.8117469879518073, "grad_norm": 0.30078125, "learning_rate": 6.275100401606427e-07, "loss": 0.8638, "step": 1867 }, { "epoch": 2.8132530120481927, "grad_norm": 0.26171875, "learning_rate": 6.224899598393574e-07, "loss": 0.8189, "step": 1868 }, { "epoch": 2.8147590361445785, "grad_norm": 0.96875, "learning_rate": 6.174698795180723e-07, "loss": 0.8421, "step": 1869 }, { "epoch": 2.816265060240964, "grad_norm": 0.2490234375, "learning_rate": 6.124497991967872e-07, "loss": 0.8365, "step": 1870 }, { "epoch": 2.817771084337349, "grad_norm": 0.255859375, "learning_rate": 6.074297188755021e-07, "loss": 0.8349, "step": 1871 }, { "epoch": 2.819277108433735, "grad_norm": 0.255859375, "learning_rate": 6.024096385542169e-07, "loss": 0.816, "step": 1872 }, { "epoch": 2.8207831325301207, "grad_norm": 0.26953125, "learning_rate": 5.973895582329317e-07, "loss": 0.8065, "step": 1873 }, { "epoch": 2.822289156626506, "grad_norm": 0.265625, "learning_rate": 5.923694779116467e-07, "loss": 0.8645, "step": 1874 }, { "epoch": 2.8237951807228914, "grad_norm": 0.287109375, "learning_rate": 5.873493975903615e-07, "loss": 0.8041, "step": 1875 }, { "epoch": 2.825301204819277, "grad_norm": 0.279296875, "learning_rate": 5.823293172690764e-07, "loss": 0.8323, "step": 1876 }, { "epoch": 2.8268072289156625, "grad_norm": 0.259765625, "learning_rate": 5.773092369477912e-07, "loss": 0.8449, "step": 1877 }, { "epoch": 2.8283132530120483, "grad_norm": 0.318359375, "learning_rate": 5.722891566265061e-07, "loss": 0.7993, "step": 1878 }, { "epoch": 2.8298192771084336, "grad_norm": 0.345703125, "learning_rate": 5.672690763052209e-07, "loss": 0.816, "step": 1879 }, { "epoch": 2.8313253012048194, "grad_norm": 0.25, "learning_rate": 5.622489959839358e-07, "loss": 0.825, "step": 1880 }, { "epoch": 2.8328313253012047, "grad_norm": 0.2578125, "learning_rate": 5.572289156626506e-07, "loss": 0.8534, "step": 1881 }, { "epoch": 2.8343373493975905, "grad_norm": 0.251953125, "learning_rate": 5.522088353413656e-07, "loss": 0.8248, "step": 1882 }, { "epoch": 2.835843373493976, "grad_norm": 0.2578125, "learning_rate": 5.471887550200804e-07, "loss": 0.802, "step": 1883 }, { "epoch": 2.837349397590361, "grad_norm": 0.265625, "learning_rate": 5.421686746987952e-07, "loss": 0.844, "step": 1884 }, { "epoch": 2.838855421686747, "grad_norm": 0.25390625, "learning_rate": 5.371485943775101e-07, "loss": 0.8301, "step": 1885 }, { "epoch": 2.8403614457831328, "grad_norm": 0.2490234375, "learning_rate": 5.321285140562249e-07, "loss": 0.832, "step": 1886 }, { "epoch": 2.841867469879518, "grad_norm": 0.251953125, "learning_rate": 5.271084337349399e-07, "loss": 0.8331, "step": 1887 }, { "epoch": 2.8433734939759034, "grad_norm": 0.271484375, "learning_rate": 5.220883534136546e-07, "loss": 0.8352, "step": 1888 }, { "epoch": 2.8448795180722892, "grad_norm": 0.322265625, "learning_rate": 5.170682730923695e-07, "loss": 0.8475, "step": 1889 }, { "epoch": 2.8463855421686746, "grad_norm": 0.25390625, "learning_rate": 5.120481927710844e-07, "loss": 0.8381, "step": 1890 }, { "epoch": 2.8478915662650603, "grad_norm": 0.267578125, "learning_rate": 5.070281124497993e-07, "loss": 0.8627, "step": 1891 }, { "epoch": 2.8493975903614457, "grad_norm": 0.32421875, "learning_rate": 5.020080321285141e-07, "loss": 0.8673, "step": 1892 }, { "epoch": 2.8509036144578315, "grad_norm": 0.3046875, "learning_rate": 4.96987951807229e-07, "loss": 0.8489, "step": 1893 }, { "epoch": 2.852409638554217, "grad_norm": 0.251953125, "learning_rate": 4.919678714859438e-07, "loss": 0.8727, "step": 1894 }, { "epoch": 2.8539156626506026, "grad_norm": 0.26171875, "learning_rate": 4.869477911646587e-07, "loss": 0.8496, "step": 1895 }, { "epoch": 2.855421686746988, "grad_norm": 0.25, "learning_rate": 4.819277108433736e-07, "loss": 0.861, "step": 1896 }, { "epoch": 2.8569277108433733, "grad_norm": 0.25, "learning_rate": 4.769076305220884e-07, "loss": 0.8313, "step": 1897 }, { "epoch": 2.858433734939759, "grad_norm": 0.259765625, "learning_rate": 4.718875502008032e-07, "loss": 0.8168, "step": 1898 }, { "epoch": 2.859939759036145, "grad_norm": 0.255859375, "learning_rate": 4.6686746987951807e-07, "loss": 0.8521, "step": 1899 }, { "epoch": 2.86144578313253, "grad_norm": 0.2470703125, "learning_rate": 4.6184738955823296e-07, "loss": 0.8424, "step": 1900 }, { "epoch": 2.8629518072289155, "grad_norm": 0.2412109375, "learning_rate": 4.568273092369478e-07, "loss": 0.8096, "step": 1901 }, { "epoch": 2.8644578313253013, "grad_norm": 0.255859375, "learning_rate": 4.518072289156627e-07, "loss": 0.813, "step": 1902 }, { "epoch": 2.8659638554216866, "grad_norm": 0.306640625, "learning_rate": 4.4678714859437755e-07, "loss": 0.8201, "step": 1903 }, { "epoch": 2.8674698795180724, "grad_norm": 0.3203125, "learning_rate": 4.417670682730924e-07, "loss": 0.8023, "step": 1904 }, { "epoch": 2.8689759036144578, "grad_norm": 0.251953125, "learning_rate": 4.367469879518073e-07, "loss": 0.8045, "step": 1905 }, { "epoch": 2.8704819277108435, "grad_norm": 0.255859375, "learning_rate": 4.3172690763052213e-07, "loss": 0.8902, "step": 1906 }, { "epoch": 2.871987951807229, "grad_norm": 0.2431640625, "learning_rate": 4.2670682730923703e-07, "loss": 0.8481, "step": 1907 }, { "epoch": 2.8734939759036147, "grad_norm": 0.2578125, "learning_rate": 4.216867469879518e-07, "loss": 0.8317, "step": 1908 }, { "epoch": 2.875, "grad_norm": 0.26953125, "learning_rate": 4.1666666666666667e-07, "loss": 0.8454, "step": 1909 }, { "epoch": 2.8765060240963853, "grad_norm": 0.25390625, "learning_rate": 4.1164658634538156e-07, "loss": 0.8225, "step": 1910 }, { "epoch": 2.878012048192771, "grad_norm": 0.2431640625, "learning_rate": 4.066265060240964e-07, "loss": 0.8392, "step": 1911 }, { "epoch": 2.8795180722891565, "grad_norm": 0.2578125, "learning_rate": 4.0160642570281125e-07, "loss": 0.8196, "step": 1912 }, { "epoch": 2.8810240963855422, "grad_norm": 0.251953125, "learning_rate": 3.9658634538152615e-07, "loss": 0.8479, "step": 1913 }, { "epoch": 2.8825301204819276, "grad_norm": 0.25390625, "learning_rate": 3.91566265060241e-07, "loss": 0.8375, "step": 1914 }, { "epoch": 2.8840361445783134, "grad_norm": 0.26171875, "learning_rate": 3.865461847389559e-07, "loss": 0.8355, "step": 1915 }, { "epoch": 2.8855421686746987, "grad_norm": 0.2421875, "learning_rate": 3.8152610441767073e-07, "loss": 0.8221, "step": 1916 }, { "epoch": 2.8870481927710845, "grad_norm": 0.265625, "learning_rate": 3.7650602409638563e-07, "loss": 0.8244, "step": 1917 }, { "epoch": 2.88855421686747, "grad_norm": 0.25390625, "learning_rate": 3.714859437751004e-07, "loss": 0.8393, "step": 1918 }, { "epoch": 2.890060240963855, "grad_norm": 0.2431640625, "learning_rate": 3.6646586345381527e-07, "loss": 0.7962, "step": 1919 }, { "epoch": 2.891566265060241, "grad_norm": 0.333984375, "learning_rate": 3.614457831325301e-07, "loss": 0.8598, "step": 1920 }, { "epoch": 2.8930722891566267, "grad_norm": 0.265625, "learning_rate": 3.56425702811245e-07, "loss": 0.8031, "step": 1921 }, { "epoch": 2.894578313253012, "grad_norm": 0.25, "learning_rate": 3.5140562248995985e-07, "loss": 0.8402, "step": 1922 }, { "epoch": 2.8960843373493974, "grad_norm": 0.283203125, "learning_rate": 3.4638554216867475e-07, "loss": 0.8228, "step": 1923 }, { "epoch": 2.897590361445783, "grad_norm": 0.259765625, "learning_rate": 3.413654618473896e-07, "loss": 0.8208, "step": 1924 }, { "epoch": 2.8990963855421685, "grad_norm": 0.291015625, "learning_rate": 3.363453815261045e-07, "loss": 0.8058, "step": 1925 }, { "epoch": 2.9006024096385543, "grad_norm": 0.2490234375, "learning_rate": 3.3132530120481933e-07, "loss": 0.8476, "step": 1926 }, { "epoch": 2.9021084337349397, "grad_norm": 0.244140625, "learning_rate": 3.263052208835342e-07, "loss": 0.8182, "step": 1927 }, { "epoch": 2.9036144578313254, "grad_norm": 0.271484375, "learning_rate": 3.2128514056224897e-07, "loss": 0.8137, "step": 1928 }, { "epoch": 2.9051204819277108, "grad_norm": 0.2734375, "learning_rate": 3.1626506024096387e-07, "loss": 0.8213, "step": 1929 }, { "epoch": 2.9066265060240966, "grad_norm": 0.259765625, "learning_rate": 3.112449799196787e-07, "loss": 0.7966, "step": 1930 }, { "epoch": 2.908132530120482, "grad_norm": 0.24609375, "learning_rate": 3.062248995983936e-07, "loss": 0.8136, "step": 1931 }, { "epoch": 2.9096385542168672, "grad_norm": 0.251953125, "learning_rate": 3.0120481927710845e-07, "loss": 0.7971, "step": 1932 }, { "epoch": 2.911144578313253, "grad_norm": 0.25390625, "learning_rate": 2.9618473895582335e-07, "loss": 0.8534, "step": 1933 }, { "epoch": 2.912650602409639, "grad_norm": 0.279296875, "learning_rate": 2.911646586345382e-07, "loss": 0.8266, "step": 1934 }, { "epoch": 2.914156626506024, "grad_norm": 0.25390625, "learning_rate": 2.8614457831325304e-07, "loss": 0.8523, "step": 1935 }, { "epoch": 2.9156626506024095, "grad_norm": 0.26953125, "learning_rate": 2.811244979919679e-07, "loss": 0.7599, "step": 1936 }, { "epoch": 2.9171686746987953, "grad_norm": 0.255859375, "learning_rate": 2.761044176706828e-07, "loss": 0.8455, "step": 1937 }, { "epoch": 2.9186746987951806, "grad_norm": 0.255859375, "learning_rate": 2.710843373493976e-07, "loss": 0.8573, "step": 1938 }, { "epoch": 2.9201807228915664, "grad_norm": 0.25, "learning_rate": 2.6606425702811247e-07, "loss": 0.8451, "step": 1939 }, { "epoch": 2.9216867469879517, "grad_norm": 0.31640625, "learning_rate": 2.610441767068273e-07, "loss": 0.8156, "step": 1940 }, { "epoch": 2.9231927710843375, "grad_norm": 0.28125, "learning_rate": 2.560240963855422e-07, "loss": 0.8538, "step": 1941 }, { "epoch": 2.924698795180723, "grad_norm": 0.255859375, "learning_rate": 2.5100401606425705e-07, "loss": 0.8537, "step": 1942 }, { "epoch": 2.9262048192771086, "grad_norm": 0.263671875, "learning_rate": 2.459839357429719e-07, "loss": 0.8495, "step": 1943 }, { "epoch": 2.927710843373494, "grad_norm": 0.2470703125, "learning_rate": 2.409638554216868e-07, "loss": 0.8476, "step": 1944 }, { "epoch": 2.9292168674698793, "grad_norm": 0.29296875, "learning_rate": 2.359437751004016e-07, "loss": 0.8278, "step": 1945 }, { "epoch": 2.930722891566265, "grad_norm": 0.263671875, "learning_rate": 2.3092369477911648e-07, "loss": 0.836, "step": 1946 }, { "epoch": 2.932228915662651, "grad_norm": 0.248046875, "learning_rate": 2.2590361445783135e-07, "loss": 0.8444, "step": 1947 }, { "epoch": 2.933734939759036, "grad_norm": 0.400390625, "learning_rate": 2.208835341365462e-07, "loss": 0.8387, "step": 1948 }, { "epoch": 2.9352409638554215, "grad_norm": 0.283203125, "learning_rate": 2.1586345381526107e-07, "loss": 0.8221, "step": 1949 }, { "epoch": 2.9367469879518073, "grad_norm": 0.255859375, "learning_rate": 2.108433734939759e-07, "loss": 0.8264, "step": 1950 }, { "epoch": 2.9382530120481927, "grad_norm": 0.248046875, "learning_rate": 2.0582329317269078e-07, "loss": 0.823, "step": 1951 }, { "epoch": 2.9397590361445785, "grad_norm": 0.2490234375, "learning_rate": 2.0080321285140563e-07, "loss": 0.8194, "step": 1952 }, { "epoch": 2.941265060240964, "grad_norm": 0.25, "learning_rate": 1.957831325301205e-07, "loss": 0.795, "step": 1953 }, { "epoch": 2.942771084337349, "grad_norm": 0.275390625, "learning_rate": 1.9076305220883537e-07, "loss": 0.8232, "step": 1954 }, { "epoch": 2.944277108433735, "grad_norm": 0.2470703125, "learning_rate": 1.857429718875502e-07, "loss": 0.7856, "step": 1955 }, { "epoch": 2.9457831325301207, "grad_norm": 0.25, "learning_rate": 1.8072289156626505e-07, "loss": 0.794, "step": 1956 }, { "epoch": 2.947289156626506, "grad_norm": 0.251953125, "learning_rate": 1.7570281124497993e-07, "loss": 0.8237, "step": 1957 }, { "epoch": 2.9487951807228914, "grad_norm": 0.251953125, "learning_rate": 1.706827309236948e-07, "loss": 0.8137, "step": 1958 }, { "epoch": 2.950301204819277, "grad_norm": 0.328125, "learning_rate": 1.6566265060240967e-07, "loss": 0.8237, "step": 1959 }, { "epoch": 2.9518072289156625, "grad_norm": 0.30859375, "learning_rate": 1.6064257028112448e-07, "loss": 0.7784, "step": 1960 }, { "epoch": 2.9533132530120483, "grad_norm": 0.26171875, "learning_rate": 1.5562248995983936e-07, "loss": 0.8296, "step": 1961 }, { "epoch": 2.9548192771084336, "grad_norm": 0.2890625, "learning_rate": 1.5060240963855423e-07, "loss": 0.842, "step": 1962 }, { "epoch": 2.9563253012048194, "grad_norm": 0.2734375, "learning_rate": 1.455823293172691e-07, "loss": 0.8427, "step": 1963 }, { "epoch": 2.9578313253012047, "grad_norm": 0.2578125, "learning_rate": 1.4056224899598394e-07, "loss": 0.8516, "step": 1964 }, { "epoch": 2.9593373493975905, "grad_norm": 0.271484375, "learning_rate": 1.355421686746988e-07, "loss": 0.8447, "step": 1965 }, { "epoch": 2.960843373493976, "grad_norm": 0.255859375, "learning_rate": 1.3052208835341366e-07, "loss": 0.8231, "step": 1966 }, { "epoch": 2.962349397590361, "grad_norm": 0.255859375, "learning_rate": 1.2550200803212853e-07, "loss": 0.8556, "step": 1967 }, { "epoch": 2.963855421686747, "grad_norm": 0.2490234375, "learning_rate": 1.204819277108434e-07, "loss": 0.8321, "step": 1968 }, { "epoch": 2.9653614457831328, "grad_norm": 0.259765625, "learning_rate": 1.1546184738955824e-07, "loss": 0.8828, "step": 1969 }, { "epoch": 2.966867469879518, "grad_norm": 0.2578125, "learning_rate": 1.104417670682731e-07, "loss": 0.85, "step": 1970 }, { "epoch": 2.9683734939759034, "grad_norm": 0.2421875, "learning_rate": 1.0542168674698796e-07, "loss": 0.7893, "step": 1971 }, { "epoch": 2.9698795180722892, "grad_norm": 0.27734375, "learning_rate": 1.0040160642570281e-07, "loss": 0.8171, "step": 1972 }, { "epoch": 2.9713855421686746, "grad_norm": 0.248046875, "learning_rate": 9.538152610441768e-08, "loss": 0.8456, "step": 1973 }, { "epoch": 2.9728915662650603, "grad_norm": 0.2578125, "learning_rate": 9.036144578313253e-08, "loss": 0.8313, "step": 1974 }, { "epoch": 2.9743975903614457, "grad_norm": 0.244140625, "learning_rate": 8.53413654618474e-08, "loss": 0.7815, "step": 1975 }, { "epoch": 2.9759036144578315, "grad_norm": 0.255859375, "learning_rate": 8.032128514056224e-08, "loss": 0.8597, "step": 1976 }, { "epoch": 2.977409638554217, "grad_norm": 0.25, "learning_rate": 7.530120481927711e-08, "loss": 0.8327, "step": 1977 }, { "epoch": 2.9789156626506026, "grad_norm": 0.267578125, "learning_rate": 7.028112449799197e-08, "loss": 0.796, "step": 1978 }, { "epoch": 2.980421686746988, "grad_norm": 0.248046875, "learning_rate": 6.526104417670683e-08, "loss": 0.8186, "step": 1979 }, { "epoch": 2.9819277108433733, "grad_norm": 0.2490234375, "learning_rate": 6.02409638554217e-08, "loss": 0.8155, "step": 1980 }, { "epoch": 2.983433734939759, "grad_norm": 0.263671875, "learning_rate": 5.522088353413655e-08, "loss": 0.8357, "step": 1981 }, { "epoch": 2.984939759036145, "grad_norm": 0.2490234375, "learning_rate": 5.0200803212851406e-08, "loss": 0.8499, "step": 1982 }, { "epoch": 2.98644578313253, "grad_norm": 0.283203125, "learning_rate": 4.5180722891566264e-08, "loss": 0.8692, "step": 1983 }, { "epoch": 2.9879518072289155, "grad_norm": 0.25390625, "learning_rate": 4.016064257028112e-08, "loss": 0.7951, "step": 1984 }, { "epoch": 2.9894578313253013, "grad_norm": 0.263671875, "learning_rate": 3.5140562248995985e-08, "loss": 0.839, "step": 1985 }, { "epoch": 2.9909638554216866, "grad_norm": 0.32421875, "learning_rate": 3.012048192771085e-08, "loss": 0.847, "step": 1986 }, { "epoch": 2.9924698795180724, "grad_norm": 0.2578125, "learning_rate": 2.5100401606425703e-08, "loss": 0.8279, "step": 1987 }, { "epoch": 2.9939759036144578, "grad_norm": 0.2431640625, "learning_rate": 2.008032128514056e-08, "loss": 0.8342, "step": 1988 }, { "epoch": 2.9954819277108435, "grad_norm": 0.271484375, "learning_rate": 1.5060240963855425e-08, "loss": 0.8189, "step": 1989 }, { "epoch": 2.996987951807229, "grad_norm": 0.25390625, "learning_rate": 1.004016064257028e-08, "loss": 0.8671, "step": 1990 }, { "epoch": 2.9984939759036147, "grad_norm": 0.28515625, "learning_rate": 5.02008032128514e-09, "loss": 0.7814, "step": 1991 }, { "epoch": 3.0, "grad_norm": 0.2490234375, "learning_rate": 0.0, "loss": 0.8393, "step": 1992 } ], "logging_steps": 1.0, "max_steps": 1992, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 0, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 6.387390132206436e+18, "train_batch_size": 1, "trial_name": null, "trial_params": null }