{ "best_metric": null, "best_model_checkpoint": null, "epoch": 8.0, "eval_steps": 500, "global_step": 8728, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 7.633587786259542e-08, "loss": 0.8707, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.5267175572519085e-07, "loss": 0.8989, "step": 2 }, { "epoch": 0.0, "learning_rate": 2.2900763358778629e-07, "loss": 0.8453, "step": 3 }, { "epoch": 0.0, "learning_rate": 3.053435114503817e-07, "loss": 0.8648, "step": 4 }, { "epoch": 0.0, "learning_rate": 3.8167938931297716e-07, "loss": 0.8816, "step": 5 }, { "epoch": 0.01, "learning_rate": 4.5801526717557257e-07, "loss": 0.9147, "step": 6 }, { "epoch": 0.01, "learning_rate": 5.34351145038168e-07, "loss": 0.8927, "step": 7 }, { "epoch": 0.01, "learning_rate": 6.106870229007634e-07, "loss": 0.9155, "step": 8 }, { "epoch": 0.01, "learning_rate": 6.870229007633589e-07, "loss": 0.8716, "step": 9 }, { "epoch": 0.01, "learning_rate": 7.633587786259543e-07, "loss": 0.8686, "step": 10 }, { "epoch": 0.01, "learning_rate": 8.396946564885497e-07, "loss": 0.9015, "step": 11 }, { "epoch": 0.01, "learning_rate": 9.160305343511451e-07, "loss": 0.8489, "step": 12 }, { "epoch": 0.01, "learning_rate": 9.923664122137404e-07, "loss": 0.7843, "step": 13 }, { "epoch": 0.01, "learning_rate": 1.068702290076336e-06, "loss": 0.8137, "step": 14 }, { "epoch": 0.01, "learning_rate": 1.1450381679389313e-06, "loss": 0.7844, "step": 15 }, { "epoch": 0.01, "learning_rate": 1.2213740458015268e-06, "loss": 0.6631, "step": 16 }, { "epoch": 0.02, "learning_rate": 1.297709923664122e-06, "loss": 0.6477, "step": 17 }, { "epoch": 0.02, "learning_rate": 1.3740458015267178e-06, "loss": 0.6598, "step": 18 }, { "epoch": 0.02, "learning_rate": 1.450381679389313e-06, "loss": 0.6032, "step": 19 }, { "epoch": 0.02, "learning_rate": 1.5267175572519086e-06, "loss": 0.62, "step": 20 }, { "epoch": 0.02, "learning_rate": 1.603053435114504e-06, "loss": 0.6099, "step": 21 }, { "epoch": 0.02, "learning_rate": 1.6793893129770995e-06, "loss": 0.5299, "step": 22 }, { "epoch": 0.02, "learning_rate": 1.7557251908396948e-06, "loss": 0.4973, "step": 23 }, { "epoch": 0.02, "learning_rate": 1.8320610687022903e-06, "loss": 0.4767, "step": 24 }, { "epoch": 0.02, "learning_rate": 1.908396946564886e-06, "loss": 0.4588, "step": 25 }, { "epoch": 0.02, "learning_rate": 1.984732824427481e-06, "loss": 0.488, "step": 26 }, { "epoch": 0.02, "learning_rate": 2.0610687022900764e-06, "loss": 0.465, "step": 27 }, { "epoch": 0.03, "learning_rate": 2.137404580152672e-06, "loss": 0.4603, "step": 28 }, { "epoch": 0.03, "learning_rate": 2.2137404580152674e-06, "loss": 0.4461, "step": 29 }, { "epoch": 0.03, "learning_rate": 2.2900763358778625e-06, "loss": 0.4482, "step": 30 }, { "epoch": 0.03, "learning_rate": 2.3664122137404585e-06, "loss": 0.4348, "step": 31 }, { "epoch": 0.03, "learning_rate": 2.4427480916030536e-06, "loss": 0.4126, "step": 32 }, { "epoch": 0.03, "learning_rate": 2.5190839694656487e-06, "loss": 0.4156, "step": 33 }, { "epoch": 0.03, "learning_rate": 2.595419847328244e-06, "loss": 0.4174, "step": 34 }, { "epoch": 0.03, "learning_rate": 2.67175572519084e-06, "loss": 0.4169, "step": 35 }, { "epoch": 0.03, "learning_rate": 2.7480916030534356e-06, "loss": 0.407, "step": 36 }, { "epoch": 0.03, "learning_rate": 2.824427480916031e-06, "loss": 0.4001, "step": 37 }, { "epoch": 0.03, "learning_rate": 2.900763358778626e-06, "loss": 0.3921, "step": 38 }, { "epoch": 0.04, "learning_rate": 2.9770992366412218e-06, "loss": 0.3954, "step": 39 }, { "epoch": 0.04, "learning_rate": 3.0534351145038173e-06, "loss": 0.3985, "step": 40 }, { "epoch": 0.04, "learning_rate": 3.129770992366413e-06, "loss": 0.3798, "step": 41 }, { "epoch": 0.04, "learning_rate": 3.206106870229008e-06, "loss": 0.384, "step": 42 }, { "epoch": 0.04, "learning_rate": 3.2824427480916034e-06, "loss": 0.3733, "step": 43 }, { "epoch": 0.04, "learning_rate": 3.358778625954199e-06, "loss": 0.3734, "step": 44 }, { "epoch": 0.04, "learning_rate": 3.4351145038167944e-06, "loss": 0.3743, "step": 45 }, { "epoch": 0.04, "learning_rate": 3.5114503816793895e-06, "loss": 0.3608, "step": 46 }, { "epoch": 0.04, "learning_rate": 3.587786259541985e-06, "loss": 0.3522, "step": 47 }, { "epoch": 0.04, "learning_rate": 3.6641221374045806e-06, "loss": 0.3479, "step": 48 }, { "epoch": 0.04, "learning_rate": 3.740458015267176e-06, "loss": 0.349, "step": 49 }, { "epoch": 0.05, "learning_rate": 3.816793893129772e-06, "loss": 0.3527, "step": 50 }, { "epoch": 0.05, "learning_rate": 3.893129770992366e-06, "loss": 0.3445, "step": 51 }, { "epoch": 0.05, "learning_rate": 3.969465648854962e-06, "loss": 0.3422, "step": 52 }, { "epoch": 0.05, "learning_rate": 4.045801526717557e-06, "loss": 0.3425, "step": 53 }, { "epoch": 0.05, "learning_rate": 4.122137404580153e-06, "loss": 0.3463, "step": 54 }, { "epoch": 0.05, "learning_rate": 4.198473282442748e-06, "loss": 0.3323, "step": 55 }, { "epoch": 0.05, "learning_rate": 4.274809160305344e-06, "loss": 0.3357, "step": 56 }, { "epoch": 0.05, "learning_rate": 4.351145038167939e-06, "loss": 0.332, "step": 57 }, { "epoch": 0.05, "learning_rate": 4.427480916030535e-06, "loss": 0.3156, "step": 58 }, { "epoch": 0.05, "learning_rate": 4.5038167938931296e-06, "loss": 0.3137, "step": 59 }, { "epoch": 0.05, "learning_rate": 4.580152671755725e-06, "loss": 0.3141, "step": 60 }, { "epoch": 0.06, "learning_rate": 4.656488549618321e-06, "loss": 0.2954, "step": 61 }, { "epoch": 0.06, "learning_rate": 4.732824427480917e-06, "loss": 0.2965, "step": 62 }, { "epoch": 0.06, "learning_rate": 4.8091603053435125e-06, "loss": 0.3031, "step": 63 }, { "epoch": 0.06, "learning_rate": 4.885496183206107e-06, "loss": 0.2668, "step": 64 }, { "epoch": 0.06, "learning_rate": 4.961832061068703e-06, "loss": 0.2639, "step": 65 }, { "epoch": 0.06, "learning_rate": 5.038167938931297e-06, "loss": 0.2765, "step": 66 }, { "epoch": 0.06, "learning_rate": 5.114503816793893e-06, "loss": 0.2673, "step": 67 }, { "epoch": 0.06, "learning_rate": 5.190839694656488e-06, "loss": 0.2526, "step": 68 }, { "epoch": 0.06, "learning_rate": 5.267175572519084e-06, "loss": 0.2544, "step": 69 }, { "epoch": 0.06, "learning_rate": 5.34351145038168e-06, "loss": 0.2532, "step": 70 }, { "epoch": 0.07, "learning_rate": 5.419847328244276e-06, "loss": 0.2458, "step": 71 }, { "epoch": 0.07, "learning_rate": 5.496183206106871e-06, "loss": 0.2397, "step": 72 }, { "epoch": 0.07, "learning_rate": 5.572519083969467e-06, "loss": 0.2456, "step": 73 }, { "epoch": 0.07, "learning_rate": 5.648854961832062e-06, "loss": 0.2363, "step": 74 }, { "epoch": 0.07, "learning_rate": 5.725190839694656e-06, "loss": 0.2361, "step": 75 }, { "epoch": 0.07, "learning_rate": 5.801526717557252e-06, "loss": 0.2348, "step": 76 }, { "epoch": 0.07, "learning_rate": 5.877862595419848e-06, "loss": 0.2378, "step": 77 }, { "epoch": 0.07, "learning_rate": 5.9541984732824435e-06, "loss": 0.2317, "step": 78 }, { "epoch": 0.07, "learning_rate": 6.030534351145039e-06, "loss": 0.2354, "step": 79 }, { "epoch": 0.07, "learning_rate": 6.1068702290076346e-06, "loss": 0.2304, "step": 80 }, { "epoch": 0.07, "learning_rate": 6.18320610687023e-06, "loss": 0.2372, "step": 81 }, { "epoch": 0.08, "learning_rate": 6.259541984732826e-06, "loss": 0.2266, "step": 82 }, { "epoch": 0.08, "learning_rate": 6.335877862595419e-06, "loss": 0.2169, "step": 83 }, { "epoch": 0.08, "learning_rate": 6.412213740458016e-06, "loss": 0.2239, "step": 84 }, { "epoch": 0.08, "learning_rate": 6.488549618320611e-06, "loss": 0.2286, "step": 85 }, { "epoch": 0.08, "learning_rate": 6.564885496183207e-06, "loss": 0.219, "step": 86 }, { "epoch": 0.08, "learning_rate": 6.641221374045802e-06, "loss": 0.2326, "step": 87 }, { "epoch": 0.08, "learning_rate": 6.717557251908398e-06, "loss": 0.2201, "step": 88 }, { "epoch": 0.08, "learning_rate": 6.793893129770993e-06, "loss": 0.2194, "step": 89 }, { "epoch": 0.08, "learning_rate": 6.870229007633589e-06, "loss": 0.2241, "step": 90 }, { "epoch": 0.08, "learning_rate": 6.946564885496184e-06, "loss": 0.2226, "step": 91 }, { "epoch": 0.08, "learning_rate": 7.022900763358779e-06, "loss": 0.2205, "step": 92 }, { "epoch": 0.09, "learning_rate": 7.0992366412213746e-06, "loss": 0.2242, "step": 93 }, { "epoch": 0.09, "learning_rate": 7.17557251908397e-06, "loss": 0.2205, "step": 94 }, { "epoch": 0.09, "learning_rate": 7.251908396946566e-06, "loss": 0.2155, "step": 95 }, { "epoch": 0.09, "learning_rate": 7.328244274809161e-06, "loss": 0.2187, "step": 96 }, { "epoch": 0.09, "learning_rate": 7.404580152671757e-06, "loss": 0.2177, "step": 97 }, { "epoch": 0.09, "learning_rate": 7.480916030534352e-06, "loss": 0.2154, "step": 98 }, { "epoch": 0.09, "learning_rate": 7.557251908396948e-06, "loss": 0.2244, "step": 99 }, { "epoch": 0.09, "learning_rate": 7.633587786259543e-06, "loss": 0.2163, "step": 100 }, { "epoch": 0.09, "learning_rate": 7.709923664122137e-06, "loss": 0.2243, "step": 101 }, { "epoch": 0.09, "learning_rate": 7.786259541984733e-06, "loss": 0.2221, "step": 102 }, { "epoch": 0.09, "learning_rate": 7.862595419847328e-06, "loss": 0.2172, "step": 103 }, { "epoch": 0.1, "learning_rate": 7.938931297709924e-06, "loss": 0.2144, "step": 104 }, { "epoch": 0.1, "learning_rate": 8.015267175572519e-06, "loss": 0.2168, "step": 105 }, { "epoch": 0.1, "learning_rate": 8.091603053435115e-06, "loss": 0.2185, "step": 106 }, { "epoch": 0.1, "learning_rate": 8.16793893129771e-06, "loss": 0.2126, "step": 107 }, { "epoch": 0.1, "learning_rate": 8.244274809160306e-06, "loss": 0.2097, "step": 108 }, { "epoch": 0.1, "learning_rate": 8.320610687022901e-06, "loss": 0.2197, "step": 109 }, { "epoch": 0.1, "learning_rate": 8.396946564885497e-06, "loss": 0.2145, "step": 110 }, { "epoch": 0.1, "learning_rate": 8.473282442748092e-06, "loss": 0.2112, "step": 111 }, { "epoch": 0.1, "learning_rate": 8.549618320610688e-06, "loss": 0.2157, "step": 112 }, { "epoch": 0.1, "learning_rate": 8.625954198473283e-06, "loss": 0.2043, "step": 113 }, { "epoch": 0.1, "learning_rate": 8.702290076335879e-06, "loss": 0.2039, "step": 114 }, { "epoch": 0.11, "learning_rate": 8.778625954198474e-06, "loss": 0.206, "step": 115 }, { "epoch": 0.11, "learning_rate": 8.85496183206107e-06, "loss": 0.2174, "step": 116 }, { "epoch": 0.11, "learning_rate": 8.931297709923665e-06, "loss": 0.2102, "step": 117 }, { "epoch": 0.11, "learning_rate": 9.007633587786259e-06, "loss": 0.2104, "step": 118 }, { "epoch": 0.11, "learning_rate": 9.083969465648855e-06, "loss": 0.2103, "step": 119 }, { "epoch": 0.11, "learning_rate": 9.16030534351145e-06, "loss": 0.2039, "step": 120 }, { "epoch": 0.11, "learning_rate": 9.236641221374046e-06, "loss": 0.2019, "step": 121 }, { "epoch": 0.11, "learning_rate": 9.312977099236641e-06, "loss": 0.2074, "step": 122 }, { "epoch": 0.11, "learning_rate": 9.389312977099237e-06, "loss": 0.1953, "step": 123 }, { "epoch": 0.11, "learning_rate": 9.465648854961834e-06, "loss": 0.1967, "step": 124 }, { "epoch": 0.11, "learning_rate": 9.54198473282443e-06, "loss": 0.2199, "step": 125 }, { "epoch": 0.12, "learning_rate": 9.618320610687025e-06, "loss": 0.2058, "step": 126 }, { "epoch": 0.12, "learning_rate": 9.694656488549619e-06, "loss": 0.2009, "step": 127 }, { "epoch": 0.12, "learning_rate": 9.770992366412214e-06, "loss": 0.2062, "step": 128 }, { "epoch": 0.12, "learning_rate": 9.84732824427481e-06, "loss": 0.2075, "step": 129 }, { "epoch": 0.12, "learning_rate": 9.923664122137405e-06, "loss": 0.1991, "step": 130 }, { "epoch": 0.12, "learning_rate": 1e-05, "loss": 0.2046, "step": 131 }, { "epoch": 0.12, "learning_rate": 1.0076335877862595e-05, "loss": 0.1946, "step": 132 }, { "epoch": 0.12, "learning_rate": 1.0152671755725192e-05, "loss": 0.2032, "step": 133 }, { "epoch": 0.12, "learning_rate": 1.0229007633587786e-05, "loss": 0.2076, "step": 134 }, { "epoch": 0.12, "learning_rate": 1.0305343511450383e-05, "loss": 0.2012, "step": 135 }, { "epoch": 0.12, "learning_rate": 1.0381679389312977e-05, "loss": 0.192, "step": 136 }, { "epoch": 0.13, "learning_rate": 1.0458015267175574e-05, "loss": 0.2029, "step": 137 }, { "epoch": 0.13, "learning_rate": 1.0534351145038168e-05, "loss": 0.2018, "step": 138 }, { "epoch": 0.13, "learning_rate": 1.0610687022900765e-05, "loss": 0.1953, "step": 139 }, { "epoch": 0.13, "learning_rate": 1.068702290076336e-05, "loss": 0.1908, "step": 140 }, { "epoch": 0.13, "learning_rate": 1.0763358778625954e-05, "loss": 0.199, "step": 141 }, { "epoch": 0.13, "learning_rate": 1.0839694656488552e-05, "loss": 0.1914, "step": 142 }, { "epoch": 0.13, "learning_rate": 1.0916030534351145e-05, "loss": 0.2008, "step": 143 }, { "epoch": 0.13, "learning_rate": 1.0992366412213743e-05, "loss": 0.1985, "step": 144 }, { "epoch": 0.13, "learning_rate": 1.1068702290076336e-05, "loss": 0.2005, "step": 145 }, { "epoch": 0.13, "learning_rate": 1.1145038167938934e-05, "loss": 0.1987, "step": 146 }, { "epoch": 0.13, "learning_rate": 1.1221374045801527e-05, "loss": 0.194, "step": 147 }, { "epoch": 0.14, "learning_rate": 1.1297709923664125e-05, "loss": 0.1977, "step": 148 }, { "epoch": 0.14, "learning_rate": 1.1374045801526718e-05, "loss": 0.2017, "step": 149 }, { "epoch": 0.14, "learning_rate": 1.1450381679389312e-05, "loss": 0.1996, "step": 150 }, { "epoch": 0.14, "learning_rate": 1.152671755725191e-05, "loss": 0.2046, "step": 151 }, { "epoch": 0.14, "learning_rate": 1.1603053435114503e-05, "loss": 0.1993, "step": 152 }, { "epoch": 0.14, "learning_rate": 1.16793893129771e-05, "loss": 0.2005, "step": 153 }, { "epoch": 0.14, "learning_rate": 1.1755725190839696e-05, "loss": 0.2062, "step": 154 }, { "epoch": 0.14, "learning_rate": 1.1832061068702292e-05, "loss": 0.1942, "step": 155 }, { "epoch": 0.14, "learning_rate": 1.1908396946564887e-05, "loss": 0.1977, "step": 156 }, { "epoch": 0.14, "learning_rate": 1.1984732824427483e-05, "loss": 0.2046, "step": 157 }, { "epoch": 0.14, "learning_rate": 1.2061068702290078e-05, "loss": 0.2048, "step": 158 }, { "epoch": 0.15, "learning_rate": 1.2137404580152672e-05, "loss": 0.1957, "step": 159 }, { "epoch": 0.15, "learning_rate": 1.2213740458015269e-05, "loss": 0.1999, "step": 160 }, { "epoch": 0.15, "learning_rate": 1.2290076335877863e-05, "loss": 0.199, "step": 161 }, { "epoch": 0.15, "learning_rate": 1.236641221374046e-05, "loss": 0.1845, "step": 162 }, { "epoch": 0.15, "learning_rate": 1.2442748091603054e-05, "loss": 0.186, "step": 163 }, { "epoch": 0.15, "learning_rate": 1.2519083969465651e-05, "loss": 0.1975, "step": 164 }, { "epoch": 0.15, "learning_rate": 1.2595419847328245e-05, "loss": 0.1904, "step": 165 }, { "epoch": 0.15, "learning_rate": 1.2671755725190839e-05, "loss": 0.2061, "step": 166 }, { "epoch": 0.15, "learning_rate": 1.2748091603053436e-05, "loss": 0.1936, "step": 167 }, { "epoch": 0.15, "learning_rate": 1.2824427480916032e-05, "loss": 0.1926, "step": 168 }, { "epoch": 0.15, "learning_rate": 1.2900763358778627e-05, "loss": 0.2103, "step": 169 }, { "epoch": 0.16, "learning_rate": 1.2977099236641223e-05, "loss": 0.1894, "step": 170 }, { "epoch": 0.16, "learning_rate": 1.3053435114503818e-05, "loss": 0.1914, "step": 171 }, { "epoch": 0.16, "learning_rate": 1.3129770992366414e-05, "loss": 0.1964, "step": 172 }, { "epoch": 0.16, "learning_rate": 1.3206106870229009e-05, "loss": 0.1979, "step": 173 }, { "epoch": 0.16, "learning_rate": 1.3282442748091605e-05, "loss": 0.1988, "step": 174 }, { "epoch": 0.16, "learning_rate": 1.3358778625954198e-05, "loss": 0.2023, "step": 175 }, { "epoch": 0.16, "learning_rate": 1.3435114503816796e-05, "loss": 0.2042, "step": 176 }, { "epoch": 0.16, "learning_rate": 1.351145038167939e-05, "loss": 0.1953, "step": 177 }, { "epoch": 0.16, "learning_rate": 1.3587786259541987e-05, "loss": 0.1917, "step": 178 }, { "epoch": 0.16, "learning_rate": 1.366412213740458e-05, "loss": 0.2005, "step": 179 }, { "epoch": 0.16, "learning_rate": 1.3740458015267178e-05, "loss": 0.1874, "step": 180 }, { "epoch": 0.17, "learning_rate": 1.3816793893129772e-05, "loss": 0.1952, "step": 181 }, { "epoch": 0.17, "learning_rate": 1.3893129770992369e-05, "loss": 0.1967, "step": 182 }, { "epoch": 0.17, "learning_rate": 1.3969465648854963e-05, "loss": 0.1963, "step": 183 }, { "epoch": 0.17, "learning_rate": 1.4045801526717558e-05, "loss": 0.1819, "step": 184 }, { "epoch": 0.17, "learning_rate": 1.4122137404580154e-05, "loss": 0.1911, "step": 185 }, { "epoch": 0.17, "learning_rate": 1.4198473282442749e-05, "loss": 0.2008, "step": 186 }, { "epoch": 0.17, "learning_rate": 1.4274809160305345e-05, "loss": 0.1991, "step": 187 }, { "epoch": 0.17, "learning_rate": 1.435114503816794e-05, "loss": 0.1962, "step": 188 }, { "epoch": 0.17, "learning_rate": 1.4427480916030536e-05, "loss": 0.1796, "step": 189 }, { "epoch": 0.17, "learning_rate": 1.4503816793893131e-05, "loss": 0.1859, "step": 190 }, { "epoch": 0.18, "learning_rate": 1.4580152671755727e-05, "loss": 0.192, "step": 191 }, { "epoch": 0.18, "learning_rate": 1.4656488549618322e-05, "loss": 0.1951, "step": 192 }, { "epoch": 0.18, "learning_rate": 1.4732824427480916e-05, "loss": 0.1926, "step": 193 }, { "epoch": 0.18, "learning_rate": 1.4809160305343513e-05, "loss": 0.1907, "step": 194 }, { "epoch": 0.18, "learning_rate": 1.4885496183206107e-05, "loss": 0.2032, "step": 195 }, { "epoch": 0.18, "learning_rate": 1.4961832061068704e-05, "loss": 0.1918, "step": 196 }, { "epoch": 0.18, "learning_rate": 1.5038167938931298e-05, "loss": 0.185, "step": 197 }, { "epoch": 0.18, "learning_rate": 1.5114503816793895e-05, "loss": 0.1973, "step": 198 }, { "epoch": 0.18, "learning_rate": 1.519083969465649e-05, "loss": 0.1963, "step": 199 }, { "epoch": 0.18, "learning_rate": 1.5267175572519086e-05, "loss": 0.1823, "step": 200 }, { "epoch": 0.18, "learning_rate": 1.5343511450381682e-05, "loss": 0.1971, "step": 201 }, { "epoch": 0.19, "learning_rate": 1.5419847328244274e-05, "loss": 0.1885, "step": 202 }, { "epoch": 0.19, "learning_rate": 1.5496183206106873e-05, "loss": 0.1977, "step": 203 }, { "epoch": 0.19, "learning_rate": 1.5572519083969465e-05, "loss": 0.1946, "step": 204 }, { "epoch": 0.19, "learning_rate": 1.5648854961832064e-05, "loss": 0.185, "step": 205 }, { "epoch": 0.19, "learning_rate": 1.5725190839694656e-05, "loss": 0.187, "step": 206 }, { "epoch": 0.19, "learning_rate": 1.5801526717557255e-05, "loss": 0.1846, "step": 207 }, { "epoch": 0.19, "learning_rate": 1.5877862595419847e-05, "loss": 0.1965, "step": 208 }, { "epoch": 0.19, "learning_rate": 1.5954198473282446e-05, "loss": 0.2008, "step": 209 }, { "epoch": 0.19, "learning_rate": 1.6030534351145038e-05, "loss": 0.2, "step": 210 }, { "epoch": 0.19, "learning_rate": 1.6106870229007634e-05, "loss": 0.2018, "step": 211 }, { "epoch": 0.19, "learning_rate": 1.618320610687023e-05, "loss": 0.1877, "step": 212 }, { "epoch": 0.2, "learning_rate": 1.6259541984732825e-05, "loss": 0.1951, "step": 213 }, { "epoch": 0.2, "learning_rate": 1.633587786259542e-05, "loss": 0.1945, "step": 214 }, { "epoch": 0.2, "learning_rate": 1.6412213740458016e-05, "loss": 0.1957, "step": 215 }, { "epoch": 0.2, "learning_rate": 1.648854961832061e-05, "loss": 0.2022, "step": 216 }, { "epoch": 0.2, "learning_rate": 1.6564885496183207e-05, "loss": 0.1944, "step": 217 }, { "epoch": 0.2, "learning_rate": 1.6641221374045802e-05, "loss": 0.1866, "step": 218 }, { "epoch": 0.2, "learning_rate": 1.6717557251908398e-05, "loss": 0.2014, "step": 219 }, { "epoch": 0.2, "learning_rate": 1.6793893129770993e-05, "loss": 0.1933, "step": 220 }, { "epoch": 0.2, "learning_rate": 1.687022900763359e-05, "loss": 0.1817, "step": 221 }, { "epoch": 0.2, "learning_rate": 1.6946564885496184e-05, "loss": 0.1896, "step": 222 }, { "epoch": 0.2, "learning_rate": 1.702290076335878e-05, "loss": 0.1924, "step": 223 }, { "epoch": 0.21, "learning_rate": 1.7099236641221375e-05, "loss": 0.1973, "step": 224 }, { "epoch": 0.21, "learning_rate": 1.717557251908397e-05, "loss": 0.1825, "step": 225 }, { "epoch": 0.21, "learning_rate": 1.7251908396946566e-05, "loss": 0.1912, "step": 226 }, { "epoch": 0.21, "learning_rate": 1.7328244274809162e-05, "loss": 0.2022, "step": 227 }, { "epoch": 0.21, "learning_rate": 1.7404580152671757e-05, "loss": 0.1919, "step": 228 }, { "epoch": 0.21, "learning_rate": 1.7480916030534353e-05, "loss": 0.1853, "step": 229 }, { "epoch": 0.21, "learning_rate": 1.755725190839695e-05, "loss": 0.1914, "step": 230 }, { "epoch": 0.21, "learning_rate": 1.7633587786259544e-05, "loss": 0.1951, "step": 231 }, { "epoch": 0.21, "learning_rate": 1.770992366412214e-05, "loss": 0.1905, "step": 232 }, { "epoch": 0.21, "learning_rate": 1.7786259541984735e-05, "loss": 0.1791, "step": 233 }, { "epoch": 0.21, "learning_rate": 1.786259541984733e-05, "loss": 0.189, "step": 234 }, { "epoch": 0.22, "learning_rate": 1.7938931297709926e-05, "loss": 0.1929, "step": 235 }, { "epoch": 0.22, "learning_rate": 1.8015267175572518e-05, "loss": 0.2021, "step": 236 }, { "epoch": 0.22, "learning_rate": 1.8091603053435117e-05, "loss": 0.191, "step": 237 }, { "epoch": 0.22, "learning_rate": 1.816793893129771e-05, "loss": 0.1775, "step": 238 }, { "epoch": 0.22, "learning_rate": 1.8244274809160308e-05, "loss": 0.1966, "step": 239 }, { "epoch": 0.22, "learning_rate": 1.83206106870229e-05, "loss": 0.1836, "step": 240 }, { "epoch": 0.22, "learning_rate": 1.83969465648855e-05, "loss": 0.1979, "step": 241 }, { "epoch": 0.22, "learning_rate": 1.847328244274809e-05, "loss": 0.1863, "step": 242 }, { "epoch": 0.22, "learning_rate": 1.854961832061069e-05, "loss": 0.1794, "step": 243 }, { "epoch": 0.22, "learning_rate": 1.8625954198473282e-05, "loss": 0.1881, "step": 244 }, { "epoch": 0.22, "learning_rate": 1.8702290076335878e-05, "loss": 0.1941, "step": 245 }, { "epoch": 0.23, "learning_rate": 1.8778625954198473e-05, "loss": 0.1855, "step": 246 }, { "epoch": 0.23, "learning_rate": 1.885496183206107e-05, "loss": 0.1917, "step": 247 }, { "epoch": 0.23, "learning_rate": 1.8931297709923668e-05, "loss": 0.1845, "step": 248 }, { "epoch": 0.23, "learning_rate": 1.900763358778626e-05, "loss": 0.1907, "step": 249 }, { "epoch": 0.23, "learning_rate": 1.908396946564886e-05, "loss": 0.1973, "step": 250 }, { "epoch": 0.23, "learning_rate": 1.916030534351145e-05, "loss": 0.1931, "step": 251 }, { "epoch": 0.23, "learning_rate": 1.923664122137405e-05, "loss": 0.1992, "step": 252 }, { "epoch": 0.23, "learning_rate": 1.9312977099236642e-05, "loss": 0.185, "step": 253 }, { "epoch": 0.23, "learning_rate": 1.9389312977099238e-05, "loss": 0.1945, "step": 254 }, { "epoch": 0.23, "learning_rate": 1.9465648854961833e-05, "loss": 0.1828, "step": 255 }, { "epoch": 0.23, "learning_rate": 1.954198473282443e-05, "loss": 0.187, "step": 256 }, { "epoch": 0.24, "learning_rate": 1.9618320610687024e-05, "loss": 0.1833, "step": 257 }, { "epoch": 0.24, "learning_rate": 1.969465648854962e-05, "loss": 0.1794, "step": 258 }, { "epoch": 0.24, "learning_rate": 1.9770992366412215e-05, "loss": 0.1943, "step": 259 }, { "epoch": 0.24, "learning_rate": 1.984732824427481e-05, "loss": 0.1849, "step": 260 }, { "epoch": 0.24, "learning_rate": 1.9923664122137406e-05, "loss": 0.1876, "step": 261 }, { "epoch": 0.24, "learning_rate": 2e-05, "loss": 0.1865, "step": 262 }, { "epoch": 0.24, "learning_rate": 1.9999999311485303e-05, "loss": 0.1837, "step": 263 }, { "epoch": 0.24, "learning_rate": 1.9999997245941302e-05, "loss": 0.1861, "step": 264 }, { "epoch": 0.24, "learning_rate": 1.999999380336828e-05, "loss": 0.1814, "step": 265 }, { "epoch": 0.24, "learning_rate": 1.999998898376671e-05, "loss": 0.1914, "step": 266 }, { "epoch": 0.24, "learning_rate": 1.9999982787137265e-05, "loss": 0.207, "step": 267 }, { "epoch": 0.25, "learning_rate": 1.9999975213480787e-05, "loss": 0.1819, "step": 268 }, { "epoch": 0.25, "learning_rate": 1.9999966262798327e-05, "loss": 0.2023, "step": 269 }, { "epoch": 0.25, "learning_rate": 1.9999955935091114e-05, "loss": 0.191, "step": 270 }, { "epoch": 0.25, "learning_rate": 1.999994423036057e-05, "loss": 0.1927, "step": 271 }, { "epoch": 0.25, "learning_rate": 1.9999931148608308e-05, "loss": 0.1912, "step": 272 }, { "epoch": 0.25, "learning_rate": 1.9999916689836126e-05, "loss": 0.1994, "step": 273 }, { "epoch": 0.25, "learning_rate": 1.9999900854046022e-05, "loss": 0.1891, "step": 274 }, { "epoch": 0.25, "learning_rate": 1.999988364124017e-05, "loss": 0.1874, "step": 275 }, { "epoch": 0.25, "learning_rate": 1.9999865051420943e-05, "loss": 0.1817, "step": 276 }, { "epoch": 0.25, "learning_rate": 1.99998450845909e-05, "loss": 0.1921, "step": 277 }, { "epoch": 0.25, "learning_rate": 1.9999823740752793e-05, "loss": 0.1767, "step": 278 }, { "epoch": 0.26, "learning_rate": 1.9999801019909557e-05, "loss": 0.1849, "step": 279 }, { "epoch": 0.26, "learning_rate": 1.9999776922064324e-05, "loss": 0.1888, "step": 280 }, { "epoch": 0.26, "learning_rate": 1.9999751447220413e-05, "loss": 0.1979, "step": 281 }, { "epoch": 0.26, "learning_rate": 1.9999724595381325e-05, "loss": 0.1979, "step": 282 }, { "epoch": 0.26, "learning_rate": 1.9999696366550767e-05, "loss": 0.1933, "step": 283 }, { "epoch": 0.26, "learning_rate": 1.999966676073262e-05, "loss": 0.1864, "step": 284 }, { "epoch": 0.26, "learning_rate": 1.9999635777930962e-05, "loss": 0.1825, "step": 285 }, { "epoch": 0.26, "learning_rate": 1.9999603418150065e-05, "loss": 0.199, "step": 286 }, { "epoch": 0.26, "learning_rate": 1.9999569681394376e-05, "loss": 0.1762, "step": 287 }, { "epoch": 0.26, "learning_rate": 1.999953456766855e-05, "loss": 0.1919, "step": 288 }, { "epoch": 0.26, "learning_rate": 1.9999498076977407e-05, "loss": 0.1791, "step": 289 }, { "epoch": 0.27, "learning_rate": 1.999946020932599e-05, "loss": 0.1895, "step": 290 }, { "epoch": 0.27, "learning_rate": 1.9999420964719503e-05, "loss": 0.1905, "step": 291 }, { "epoch": 0.27, "learning_rate": 1.9999380343163356e-05, "loss": 0.1844, "step": 292 }, { "epoch": 0.27, "learning_rate": 1.9999338344663135e-05, "loss": 0.1924, "step": 293 }, { "epoch": 0.27, "learning_rate": 1.999929496922463e-05, "loss": 0.1878, "step": 294 }, { "epoch": 0.27, "learning_rate": 1.999925021685381e-05, "loss": 0.1936, "step": 295 }, { "epoch": 0.27, "learning_rate": 1.999920408755684e-05, "loss": 0.1947, "step": 296 }, { "epoch": 0.27, "learning_rate": 1.999915658134007e-05, "loss": 0.1867, "step": 297 }, { "epoch": 0.27, "learning_rate": 1.9999107698210043e-05, "loss": 0.1716, "step": 298 }, { "epoch": 0.27, "learning_rate": 1.9999057438173493e-05, "loss": 0.1912, "step": 299 }, { "epoch": 0.27, "learning_rate": 1.9999005801237338e-05, "loss": 0.1817, "step": 300 }, { "epoch": 0.28, "learning_rate": 1.9998952787408687e-05, "loss": 0.1888, "step": 301 }, { "epoch": 0.28, "learning_rate": 1.999889839669484e-05, "loss": 0.1876, "step": 302 }, { "epoch": 0.28, "learning_rate": 1.9998842629103293e-05, "loss": 0.1728, "step": 303 }, { "epoch": 0.28, "learning_rate": 1.999878548464172e-05, "loss": 0.1921, "step": 304 }, { "epoch": 0.28, "learning_rate": 1.999872696331799e-05, "loss": 0.1945, "step": 305 }, { "epoch": 0.28, "learning_rate": 1.9998667065140163e-05, "loss": 0.189, "step": 306 }, { "epoch": 0.28, "learning_rate": 1.9998605790116486e-05, "loss": 0.1843, "step": 307 }, { "epoch": 0.28, "learning_rate": 1.9998543138255397e-05, "loss": 0.1798, "step": 308 }, { "epoch": 0.28, "learning_rate": 1.9998479109565525e-05, "loss": 0.1818, "step": 309 }, { "epoch": 0.28, "learning_rate": 1.9998413704055687e-05, "loss": 0.1762, "step": 310 }, { "epoch": 0.29, "learning_rate": 1.9998346921734887e-05, "loss": 0.183, "step": 311 }, { "epoch": 0.29, "learning_rate": 1.9998278762612322e-05, "loss": 0.1848, "step": 312 }, { "epoch": 0.29, "learning_rate": 1.999820922669738e-05, "loss": 0.1942, "step": 313 }, { "epoch": 0.29, "learning_rate": 1.9998138313999635e-05, "loss": 0.1937, "step": 314 }, { "epoch": 0.29, "learning_rate": 1.9998066024528845e-05, "loss": 0.1927, "step": 315 }, { "epoch": 0.29, "learning_rate": 1.9997992358294976e-05, "loss": 0.1977, "step": 316 }, { "epoch": 0.29, "learning_rate": 1.9997917315308167e-05, "loss": 0.186, "step": 317 }, { "epoch": 0.29, "learning_rate": 1.9997840895578753e-05, "loss": 0.1905, "step": 318 }, { "epoch": 0.29, "learning_rate": 1.9997763099117253e-05, "loss": 0.1738, "step": 319 }, { "epoch": 0.29, "learning_rate": 1.9997683925934385e-05, "loss": 0.1731, "step": 320 }, { "epoch": 0.29, "learning_rate": 1.9997603376041047e-05, "loss": 0.1854, "step": 321 }, { "epoch": 0.3, "learning_rate": 1.999752144944833e-05, "loss": 0.1876, "step": 322 }, { "epoch": 0.3, "learning_rate": 1.9997438146167526e-05, "loss": 0.181, "step": 323 }, { "epoch": 0.3, "learning_rate": 1.9997353466210093e-05, "loss": 0.184, "step": 324 }, { "epoch": 0.3, "learning_rate": 1.9997267409587697e-05, "loss": 0.1917, "step": 325 }, { "epoch": 0.3, "learning_rate": 1.9997179976312193e-05, "loss": 0.1859, "step": 326 }, { "epoch": 0.3, "learning_rate": 1.999709116639561e-05, "loss": 0.1797, "step": 327 }, { "epoch": 0.3, "learning_rate": 1.999700097985019e-05, "loss": 0.1784, "step": 328 }, { "epoch": 0.3, "learning_rate": 1.999690941668834e-05, "loss": 0.1875, "step": 329 }, { "epoch": 0.3, "learning_rate": 1.999681647692268e-05, "loss": 0.1973, "step": 330 }, { "epoch": 0.3, "learning_rate": 1.9996722160565996e-05, "loss": 0.1903, "step": 331 }, { "epoch": 0.3, "learning_rate": 1.9996626467631283e-05, "loss": 0.187, "step": 332 }, { "epoch": 0.31, "learning_rate": 1.999652939813172e-05, "loss": 0.1835, "step": 333 }, { "epoch": 0.31, "learning_rate": 1.999643095208067e-05, "loss": 0.1772, "step": 334 }, { "epoch": 0.31, "learning_rate": 1.999633112949169e-05, "loss": 0.1823, "step": 335 }, { "epoch": 0.31, "learning_rate": 1.9996229930378525e-05, "loss": 0.1963, "step": 336 }, { "epoch": 0.31, "learning_rate": 1.9996127354755112e-05, "loss": 0.1828, "step": 337 }, { "epoch": 0.31, "learning_rate": 1.9996023402635576e-05, "loss": 0.1788, "step": 338 }, { "epoch": 0.31, "learning_rate": 1.999591807403423e-05, "loss": 0.179, "step": 339 }, { "epoch": 0.31, "learning_rate": 1.999581136896558e-05, "loss": 0.1818, "step": 340 }, { "epoch": 0.31, "learning_rate": 1.9995703287444317e-05, "loss": 0.1892, "step": 341 }, { "epoch": 0.31, "learning_rate": 1.9995593829485328e-05, "loss": 0.1767, "step": 342 }, { "epoch": 0.31, "learning_rate": 1.999548299510368e-05, "loss": 0.1915, "step": 343 }, { "epoch": 0.32, "learning_rate": 1.9995370784314643e-05, "loss": 0.1754, "step": 344 }, { "epoch": 0.32, "learning_rate": 1.999525719713366e-05, "loss": 0.1809, "step": 345 }, { "epoch": 0.32, "learning_rate": 1.999514223357638e-05, "loss": 0.1723, "step": 346 }, { "epoch": 0.32, "learning_rate": 1.9995025893658627e-05, "loss": 0.1816, "step": 347 }, { "epoch": 0.32, "learning_rate": 1.9994908177396427e-05, "loss": 0.1875, "step": 348 }, { "epoch": 0.32, "learning_rate": 1.999478908480599e-05, "loss": 0.182, "step": 349 }, { "epoch": 0.32, "learning_rate": 1.9994668615903707e-05, "loss": 0.192, "step": 350 }, { "epoch": 0.32, "learning_rate": 1.9994546770706177e-05, "loss": 0.1951, "step": 351 }, { "epoch": 0.32, "learning_rate": 1.9994423549230175e-05, "loss": 0.1817, "step": 352 }, { "epoch": 0.32, "learning_rate": 1.999429895149267e-05, "loss": 0.1746, "step": 353 }, { "epoch": 0.32, "learning_rate": 1.9994172977510813e-05, "loss": 0.1891, "step": 354 }, { "epoch": 0.33, "learning_rate": 1.9994045627301957e-05, "loss": 0.1937, "step": 355 }, { "epoch": 0.33, "learning_rate": 1.999391690088364e-05, "loss": 0.1817, "step": 356 }, { "epoch": 0.33, "learning_rate": 1.9993786798273584e-05, "loss": 0.1921, "step": 357 }, { "epoch": 0.33, "learning_rate": 1.9993655319489703e-05, "loss": 0.1878, "step": 358 }, { "epoch": 0.33, "learning_rate": 1.999352246455011e-05, "loss": 0.1817, "step": 359 }, { "epoch": 0.33, "learning_rate": 1.999338823347309e-05, "loss": 0.1844, "step": 360 }, { "epoch": 0.33, "learning_rate": 1.9993252626277133e-05, "loss": 0.1824, "step": 361 }, { "epoch": 0.33, "learning_rate": 1.9993115642980912e-05, "loss": 0.1828, "step": 362 }, { "epoch": 0.33, "learning_rate": 1.999297728360329e-05, "loss": 0.1904, "step": 363 }, { "epoch": 0.33, "learning_rate": 1.9992837548163315e-05, "loss": 0.1848, "step": 364 }, { "epoch": 0.33, "learning_rate": 1.9992696436680235e-05, "loss": 0.1834, "step": 365 }, { "epoch": 0.34, "learning_rate": 1.9992553949173476e-05, "loss": 0.1876, "step": 366 }, { "epoch": 0.34, "learning_rate": 1.9992410085662667e-05, "loss": 0.1799, "step": 367 }, { "epoch": 0.34, "learning_rate": 1.9992264846167608e-05, "loss": 0.1749, "step": 368 }, { "epoch": 0.34, "learning_rate": 1.9992118230708306e-05, "loss": 0.1769, "step": 369 }, { "epoch": 0.34, "learning_rate": 1.999197023930495e-05, "loss": 0.1915, "step": 370 }, { "epoch": 0.34, "learning_rate": 1.9991820871977917e-05, "loss": 0.1735, "step": 371 }, { "epoch": 0.34, "learning_rate": 1.9991670128747773e-05, "loss": 0.1899, "step": 372 }, { "epoch": 0.34, "learning_rate": 1.999151800963528e-05, "loss": 0.1765, "step": 373 }, { "epoch": 0.34, "learning_rate": 1.9991364514661384e-05, "loss": 0.1788, "step": 374 }, { "epoch": 0.34, "learning_rate": 1.9991209643847223e-05, "loss": 0.1821, "step": 375 }, { "epoch": 0.34, "learning_rate": 1.9991053397214122e-05, "loss": 0.1914, "step": 376 }, { "epoch": 0.35, "learning_rate": 1.9990895774783593e-05, "loss": 0.1756, "step": 377 }, { "epoch": 0.35, "learning_rate": 1.9990736776577346e-05, "loss": 0.1931, "step": 378 }, { "epoch": 0.35, "learning_rate": 1.9990576402617273e-05, "loss": 0.1939, "step": 379 }, { "epoch": 0.35, "learning_rate": 1.999041465292546e-05, "loss": 0.1844, "step": 380 }, { "epoch": 0.35, "learning_rate": 1.9990251527524178e-05, "loss": 0.1812, "step": 381 }, { "epoch": 0.35, "learning_rate": 1.9990087026435893e-05, "loss": 0.178, "step": 382 }, { "epoch": 0.35, "learning_rate": 1.9989921149683255e-05, "loss": 0.1947, "step": 383 }, { "epoch": 0.35, "learning_rate": 1.9989753897289104e-05, "loss": 0.1806, "step": 384 }, { "epoch": 0.35, "learning_rate": 1.9989585269276475e-05, "loss": 0.18, "step": 385 }, { "epoch": 0.35, "learning_rate": 1.9989415265668587e-05, "loss": 0.1841, "step": 386 }, { "epoch": 0.35, "learning_rate": 1.9989243886488846e-05, "loss": 0.1761, "step": 387 }, { "epoch": 0.36, "learning_rate": 1.998907113176086e-05, "loss": 0.1821, "step": 388 }, { "epoch": 0.36, "learning_rate": 1.998889700150841e-05, "loss": 0.1843, "step": 389 }, { "epoch": 0.36, "learning_rate": 1.9988721495755477e-05, "loss": 0.1924, "step": 390 }, { "epoch": 0.36, "learning_rate": 1.9988544614526232e-05, "loss": 0.1877, "step": 391 }, { "epoch": 0.36, "learning_rate": 1.9988366357845027e-05, "loss": 0.1816, "step": 392 }, { "epoch": 0.36, "learning_rate": 1.998818672573641e-05, "loss": 0.1677, "step": 393 }, { "epoch": 0.36, "learning_rate": 1.9988005718225117e-05, "loss": 0.186, "step": 394 }, { "epoch": 0.36, "learning_rate": 1.9987823335336073e-05, "loss": 0.18, "step": 395 }, { "epoch": 0.36, "learning_rate": 1.9987639577094397e-05, "loss": 0.1945, "step": 396 }, { "epoch": 0.36, "learning_rate": 1.9987454443525386e-05, "loss": 0.1736, "step": 397 }, { "epoch": 0.36, "learning_rate": 1.998726793465454e-05, "loss": 0.1783, "step": 398 }, { "epoch": 0.37, "learning_rate": 1.9987080050507534e-05, "loss": 0.1762, "step": 399 }, { "epoch": 0.37, "learning_rate": 1.998689079111025e-05, "loss": 0.1809, "step": 400 }, { "epoch": 0.37, "learning_rate": 1.998670015648874e-05, "loss": 0.1778, "step": 401 }, { "epoch": 0.37, "learning_rate": 1.998650814666926e-05, "loss": 0.1886, "step": 402 }, { "epoch": 0.37, "learning_rate": 1.998631476167825e-05, "loss": 0.1773, "step": 403 }, { "epoch": 0.37, "learning_rate": 1.998612000154234e-05, "loss": 0.1833, "step": 404 }, { "epoch": 0.37, "learning_rate": 1.9985923866288348e-05, "loss": 0.1857, "step": 405 }, { "epoch": 0.37, "learning_rate": 1.9985726355943284e-05, "loss": 0.1612, "step": 406 }, { "epoch": 0.37, "learning_rate": 1.9985527470534345e-05, "loss": 0.1899, "step": 407 }, { "epoch": 0.37, "learning_rate": 1.9985327210088915e-05, "loss": 0.1778, "step": 408 }, { "epoch": 0.37, "learning_rate": 1.9985125574634573e-05, "loss": 0.1788, "step": 409 }, { "epoch": 0.38, "learning_rate": 1.9984922564199087e-05, "loss": 0.1827, "step": 410 }, { "epoch": 0.38, "learning_rate": 1.9984718178810408e-05, "loss": 0.1866, "step": 411 }, { "epoch": 0.38, "learning_rate": 1.9984512418496683e-05, "loss": 0.183, "step": 412 }, { "epoch": 0.38, "learning_rate": 1.9984305283286244e-05, "loss": 0.1832, "step": 413 }, { "epoch": 0.38, "learning_rate": 1.998409677320762e-05, "loss": 0.1887, "step": 414 }, { "epoch": 0.38, "learning_rate": 1.9983886888289515e-05, "loss": 0.182, "step": 415 }, { "epoch": 0.38, "learning_rate": 1.9983675628560834e-05, "loss": 0.1819, "step": 416 }, { "epoch": 0.38, "learning_rate": 1.998346299405067e-05, "loss": 0.1824, "step": 417 }, { "epoch": 0.38, "learning_rate": 1.9983248984788304e-05, "loss": 0.1668, "step": 418 }, { "epoch": 0.38, "learning_rate": 1.99830336008032e-05, "loss": 0.184, "step": 419 }, { "epoch": 0.38, "learning_rate": 1.9982816842125023e-05, "loss": 0.1885, "step": 420 }, { "epoch": 0.39, "learning_rate": 1.998259870878362e-05, "loss": 0.1818, "step": 421 }, { "epoch": 0.39, "learning_rate": 1.9982379200809024e-05, "loss": 0.1716, "step": 422 }, { "epoch": 0.39, "learning_rate": 1.998215831823147e-05, "loss": 0.1919, "step": 423 }, { "epoch": 0.39, "learning_rate": 1.9981936061081366e-05, "loss": 0.1779, "step": 424 }, { "epoch": 0.39, "learning_rate": 1.9981712429389324e-05, "loss": 0.1687, "step": 425 }, { "epoch": 0.39, "learning_rate": 1.9981487423186134e-05, "loss": 0.1891, "step": 426 }, { "epoch": 0.39, "learning_rate": 1.9981261042502783e-05, "loss": 0.1804, "step": 427 }, { "epoch": 0.39, "learning_rate": 1.9981033287370443e-05, "loss": 0.188, "step": 428 }, { "epoch": 0.39, "learning_rate": 1.9980804157820474e-05, "loss": 0.1926, "step": 429 }, { "epoch": 0.39, "learning_rate": 1.9980573653884437e-05, "loss": 0.1817, "step": 430 }, { "epoch": 0.4, "learning_rate": 1.998034177559406e-05, "loss": 0.1742, "step": 431 }, { "epoch": 0.4, "learning_rate": 1.9980108522981287e-05, "loss": 0.1789, "step": 432 }, { "epoch": 0.4, "learning_rate": 1.9979873896078226e-05, "loss": 0.1892, "step": 433 }, { "epoch": 0.4, "learning_rate": 1.9979637894917192e-05, "loss": 0.1787, "step": 434 }, { "epoch": 0.4, "learning_rate": 1.997940051953068e-05, "loss": 0.1871, "step": 435 }, { "epoch": 0.4, "learning_rate": 1.9979161769951377e-05, "loss": 0.1842, "step": 436 }, { "epoch": 0.4, "learning_rate": 1.9978921646212166e-05, "loss": 0.1825, "step": 437 }, { "epoch": 0.4, "learning_rate": 1.9978680148346106e-05, "loss": 0.1889, "step": 438 }, { "epoch": 0.4, "learning_rate": 1.9978437276386454e-05, "loss": 0.1671, "step": 439 }, { "epoch": 0.4, "learning_rate": 1.9978193030366655e-05, "loss": 0.1703, "step": 440 }, { "epoch": 0.4, "learning_rate": 1.997794741032034e-05, "loss": 0.1824, "step": 441 }, { "epoch": 0.41, "learning_rate": 1.9977700416281332e-05, "loss": 0.1771, "step": 442 }, { "epoch": 0.41, "learning_rate": 1.9977452048283648e-05, "loss": 0.185, "step": 443 }, { "epoch": 0.41, "learning_rate": 1.997720230636148e-05, "loss": 0.1879, "step": 444 }, { "epoch": 0.41, "learning_rate": 1.9976951190549228e-05, "loss": 0.1824, "step": 445 }, { "epoch": 0.41, "learning_rate": 1.9976698700881464e-05, "loss": 0.1687, "step": 446 }, { "epoch": 0.41, "learning_rate": 1.9976444837392958e-05, "loss": 0.1876, "step": 447 }, { "epoch": 0.41, "learning_rate": 1.9976189600118673e-05, "loss": 0.1662, "step": 448 }, { "epoch": 0.41, "learning_rate": 1.9975932989093746e-05, "loss": 0.1777, "step": 449 }, { "epoch": 0.41, "learning_rate": 1.9975675004353523e-05, "loss": 0.181, "step": 450 }, { "epoch": 0.41, "learning_rate": 1.9975415645933525e-05, "loss": 0.1797, "step": 451 }, { "epoch": 0.41, "learning_rate": 1.9975154913869467e-05, "loss": 0.1905, "step": 452 }, { "epoch": 0.42, "learning_rate": 1.9974892808197248e-05, "loss": 0.182, "step": 453 }, { "epoch": 0.42, "learning_rate": 1.997462932895297e-05, "loss": 0.1784, "step": 454 }, { "epoch": 0.42, "learning_rate": 1.9974364476172908e-05, "loss": 0.1828, "step": 455 }, { "epoch": 0.42, "learning_rate": 1.997409824989353e-05, "loss": 0.1851, "step": 456 }, { "epoch": 0.42, "learning_rate": 1.997383065015151e-05, "loss": 0.1901, "step": 457 }, { "epoch": 0.42, "learning_rate": 1.997356167698368e-05, "loss": 0.1749, "step": 458 }, { "epoch": 0.42, "learning_rate": 1.997329133042709e-05, "loss": 0.1821, "step": 459 }, { "epoch": 0.42, "learning_rate": 1.9973019610518965e-05, "loss": 0.1729, "step": 460 }, { "epoch": 0.42, "learning_rate": 1.997274651729672e-05, "loss": 0.164, "step": 461 }, { "epoch": 0.42, "learning_rate": 1.997247205079796e-05, "loss": 0.1835, "step": 462 }, { "epoch": 0.42, "learning_rate": 1.9972196211060482e-05, "loss": 0.1888, "step": 463 }, { "epoch": 0.43, "learning_rate": 1.997191899812227e-05, "loss": 0.1716, "step": 464 }, { "epoch": 0.43, "learning_rate": 1.9971640412021494e-05, "loss": 0.1843, "step": 465 }, { "epoch": 0.43, "learning_rate": 1.9971360452796523e-05, "loss": 0.1806, "step": 466 }, { "epoch": 0.43, "learning_rate": 1.99710791204859e-05, "loss": 0.1702, "step": 467 }, { "epoch": 0.43, "learning_rate": 1.9970796415128372e-05, "loss": 0.173, "step": 468 }, { "epoch": 0.43, "learning_rate": 1.997051233676286e-05, "loss": 0.169, "step": 469 }, { "epoch": 0.43, "learning_rate": 1.997022688542849e-05, "loss": 0.1824, "step": 470 }, { "epoch": 0.43, "learning_rate": 1.9969940061164567e-05, "loss": 0.1888, "step": 471 }, { "epoch": 0.43, "learning_rate": 1.996965186401059e-05, "loss": 0.1688, "step": 472 }, { "epoch": 0.43, "learning_rate": 1.996936229400624e-05, "loss": 0.1739, "step": 473 }, { "epoch": 0.43, "learning_rate": 1.9969071351191392e-05, "loss": 0.173, "step": 474 }, { "epoch": 0.44, "learning_rate": 1.9968779035606118e-05, "loss": 0.1719, "step": 475 }, { "epoch": 0.44, "learning_rate": 1.9968485347290657e-05, "loss": 0.1748, "step": 476 }, { "epoch": 0.44, "learning_rate": 1.9968190286285463e-05, "loss": 0.188, "step": 477 }, { "epoch": 0.44, "learning_rate": 1.996789385263116e-05, "loss": 0.1873, "step": 478 }, { "epoch": 0.44, "learning_rate": 1.996759604636857e-05, "loss": 0.1702, "step": 479 }, { "epoch": 0.44, "learning_rate": 1.99672968675387e-05, "loss": 0.1774, "step": 480 }, { "epoch": 0.44, "learning_rate": 1.9966996316182752e-05, "loss": 0.1785, "step": 481 }, { "epoch": 0.44, "learning_rate": 1.9966694392342107e-05, "loss": 0.1763, "step": 482 }, { "epoch": 0.44, "learning_rate": 1.9966391096058345e-05, "loss": 0.1777, "step": 483 }, { "epoch": 0.44, "learning_rate": 1.996608642737323e-05, "loss": 0.1757, "step": 484 }, { "epoch": 0.44, "learning_rate": 1.996578038632872e-05, "loss": 0.1847, "step": 485 }, { "epoch": 0.45, "learning_rate": 1.9965472972966947e-05, "loss": 0.1735, "step": 486 }, { "epoch": 0.45, "learning_rate": 1.996516418733025e-05, "loss": 0.1901, "step": 487 }, { "epoch": 0.45, "learning_rate": 1.9964854029461148e-05, "loss": 0.176, "step": 488 }, { "epoch": 0.45, "learning_rate": 1.9964542499402352e-05, "loss": 0.1832, "step": 489 }, { "epoch": 0.45, "learning_rate": 1.9964229597196758e-05, "loss": 0.1748, "step": 490 }, { "epoch": 0.45, "learning_rate": 1.9963915322887457e-05, "loss": 0.1777, "step": 491 }, { "epoch": 0.45, "learning_rate": 1.9963599676517728e-05, "loss": 0.1914, "step": 492 }, { "epoch": 0.45, "learning_rate": 1.9963282658131024e-05, "loss": 0.1728, "step": 493 }, { "epoch": 0.45, "learning_rate": 1.9962964267771014e-05, "loss": 0.1804, "step": 494 }, { "epoch": 0.45, "learning_rate": 1.9962644505481532e-05, "loss": 0.1713, "step": 495 }, { "epoch": 0.45, "learning_rate": 1.9962323371306616e-05, "loss": 0.1777, "step": 496 }, { "epoch": 0.46, "learning_rate": 1.996200086529048e-05, "loss": 0.1797, "step": 497 }, { "epoch": 0.46, "learning_rate": 1.9961676987477542e-05, "loss": 0.1773, "step": 498 }, { "epoch": 0.46, "learning_rate": 1.9961351737912394e-05, "loss": 0.1803, "step": 499 }, { "epoch": 0.46, "learning_rate": 1.996102511663983e-05, "loss": 0.171, "step": 500 }, { "epoch": 0.46, "learning_rate": 1.9960697123704824e-05, "loss": 0.1909, "step": 501 }, { "epoch": 0.46, "learning_rate": 1.9960367759152543e-05, "loss": 0.1669, "step": 502 }, { "epoch": 0.46, "learning_rate": 1.9960037023028332e-05, "loss": 0.1818, "step": 503 }, { "epoch": 0.46, "learning_rate": 1.9959704915377748e-05, "loss": 0.1741, "step": 504 }, { "epoch": 0.46, "learning_rate": 1.9959371436246516e-05, "loss": 0.1893, "step": 505 }, { "epoch": 0.46, "learning_rate": 1.995903658568056e-05, "loss": 0.1686, "step": 506 }, { "epoch": 0.46, "learning_rate": 1.995870036372599e-05, "loss": 0.179, "step": 507 }, { "epoch": 0.47, "learning_rate": 1.9958362770429098e-05, "loss": 0.1787, "step": 508 }, { "epoch": 0.47, "learning_rate": 1.995802380583638e-05, "loss": 0.1952, "step": 509 }, { "epoch": 0.47, "learning_rate": 1.9957683469994507e-05, "loss": 0.1679, "step": 510 }, { "epoch": 0.47, "learning_rate": 1.9957341762950346e-05, "loss": 0.172, "step": 511 }, { "epoch": 0.47, "learning_rate": 1.9956998684750952e-05, "loss": 0.1792, "step": 512 }, { "epoch": 0.47, "learning_rate": 1.9956654235443567e-05, "loss": 0.1697, "step": 513 }, { "epoch": 0.47, "learning_rate": 1.9956308415075627e-05, "loss": 0.1754, "step": 514 }, { "epoch": 0.47, "learning_rate": 1.9955961223694744e-05, "loss": 0.1824, "step": 515 }, { "epoch": 0.47, "learning_rate": 1.9955612661348735e-05, "loss": 0.1841, "step": 516 }, { "epoch": 0.47, "learning_rate": 1.9955262728085592e-05, "loss": 0.177, "step": 517 }, { "epoch": 0.47, "learning_rate": 1.9954911423953503e-05, "loss": 0.1742, "step": 518 }, { "epoch": 0.48, "learning_rate": 1.9954558749000846e-05, "loss": 0.1853, "step": 519 }, { "epoch": 0.48, "learning_rate": 1.995420470327619e-05, "loss": 0.1657, "step": 520 }, { "epoch": 0.48, "learning_rate": 1.9953849286828276e-05, "loss": 0.1814, "step": 521 }, { "epoch": 0.48, "learning_rate": 1.9953492499706057e-05, "loss": 0.1903, "step": 522 }, { "epoch": 0.48, "learning_rate": 1.9953134341958656e-05, "loss": 0.184, "step": 523 }, { "epoch": 0.48, "learning_rate": 1.99527748136354e-05, "loss": 0.1865, "step": 524 }, { "epoch": 0.48, "learning_rate": 1.9952413914785793e-05, "loss": 0.1797, "step": 525 }, { "epoch": 0.48, "learning_rate": 1.9952051645459524e-05, "loss": 0.1777, "step": 526 }, { "epoch": 0.48, "learning_rate": 1.995168800570649e-05, "loss": 0.1747, "step": 527 }, { "epoch": 0.48, "learning_rate": 1.9951322995576766e-05, "loss": 0.1762, "step": 528 }, { "epoch": 0.48, "learning_rate": 1.9950956615120604e-05, "loss": 0.1816, "step": 529 }, { "epoch": 0.49, "learning_rate": 1.9950588864388465e-05, "loss": 0.1715, "step": 530 }, { "epoch": 0.49, "learning_rate": 1.995021974343099e-05, "loss": 0.1609, "step": 531 }, { "epoch": 0.49, "learning_rate": 1.9949849252298996e-05, "loss": 0.1767, "step": 532 }, { "epoch": 0.49, "learning_rate": 1.9949477391043514e-05, "loss": 0.189, "step": 533 }, { "epoch": 0.49, "learning_rate": 1.9949104159715746e-05, "loss": 0.17, "step": 534 }, { "epoch": 0.49, "learning_rate": 1.9948729558367084e-05, "loss": 0.175, "step": 535 }, { "epoch": 0.49, "learning_rate": 1.9948353587049115e-05, "loss": 0.1762, "step": 536 }, { "epoch": 0.49, "learning_rate": 1.994797624581361e-05, "loss": 0.1834, "step": 537 }, { "epoch": 0.49, "learning_rate": 1.9947597534712533e-05, "loss": 0.1746, "step": 538 }, { "epoch": 0.49, "learning_rate": 1.994721745379803e-05, "loss": 0.1676, "step": 539 }, { "epoch": 0.49, "learning_rate": 1.9946836003122443e-05, "loss": 0.1731, "step": 540 }, { "epoch": 0.5, "learning_rate": 1.9946453182738296e-05, "loss": 0.1794, "step": 541 }, { "epoch": 0.5, "learning_rate": 1.99460689926983e-05, "loss": 0.175, "step": 542 }, { "epoch": 0.5, "learning_rate": 1.9945683433055367e-05, "loss": 0.1726, "step": 543 }, { "epoch": 0.5, "learning_rate": 1.994529650386259e-05, "loss": 0.1901, "step": 544 }, { "epoch": 0.5, "learning_rate": 1.994490820517324e-05, "loss": 0.1708, "step": 545 }, { "epoch": 0.5, "learning_rate": 1.9944518537040797e-05, "loss": 0.187, "step": 546 }, { "epoch": 0.5, "learning_rate": 1.994412749951892e-05, "loss": 0.1568, "step": 547 }, { "epoch": 0.5, "learning_rate": 1.994373509266145e-05, "loss": 0.19, "step": 548 }, { "epoch": 0.5, "learning_rate": 1.994334131652242e-05, "loss": 0.1773, "step": 549 }, { "epoch": 0.5, "learning_rate": 1.9942946171156062e-05, "loss": 0.1751, "step": 550 }, { "epoch": 0.51, "learning_rate": 1.994254965661679e-05, "loss": 0.1869, "step": 551 }, { "epoch": 0.51, "learning_rate": 1.9942151772959197e-05, "loss": 0.1703, "step": 552 }, { "epoch": 0.51, "learning_rate": 1.9941752520238076e-05, "loss": 0.183, "step": 553 }, { "epoch": 0.51, "learning_rate": 1.994135189850841e-05, "loss": 0.1717, "step": 554 }, { "epoch": 0.51, "learning_rate": 1.9940949907825357e-05, "loss": 0.1764, "step": 555 }, { "epoch": 0.51, "learning_rate": 1.994054654824428e-05, "loss": 0.1806, "step": 556 }, { "epoch": 0.51, "learning_rate": 1.9940141819820717e-05, "loss": 0.182, "step": 557 }, { "epoch": 0.51, "learning_rate": 1.9939735722610408e-05, "loss": 0.1741, "step": 558 }, { "epoch": 0.51, "learning_rate": 1.9939328256669266e-05, "loss": 0.172, "step": 559 }, { "epoch": 0.51, "learning_rate": 1.9938919422053402e-05, "loss": 0.1686, "step": 560 }, { "epoch": 0.51, "learning_rate": 1.9938509218819114e-05, "loss": 0.1805, "step": 561 }, { "epoch": 0.52, "learning_rate": 1.9938097647022895e-05, "loss": 0.1854, "step": 562 }, { "epoch": 0.52, "learning_rate": 1.9937684706721408e-05, "loss": 0.1699, "step": 563 }, { "epoch": 0.52, "learning_rate": 1.9937270397971523e-05, "loss": 0.1849, "step": 564 }, { "epoch": 0.52, "learning_rate": 1.993685472083029e-05, "loss": 0.1705, "step": 565 }, { "epoch": 0.52, "learning_rate": 1.9936437675354953e-05, "loss": 0.1782, "step": 566 }, { "epoch": 0.52, "learning_rate": 1.9936019261602933e-05, "loss": 0.1735, "step": 567 }, { "epoch": 0.52, "learning_rate": 1.993559947963185e-05, "loss": 0.1727, "step": 568 }, { "epoch": 0.52, "learning_rate": 1.993517832949951e-05, "loss": 0.1776, "step": 569 }, { "epoch": 0.52, "learning_rate": 1.993475581126391e-05, "loss": 0.1753, "step": 570 }, { "epoch": 0.52, "learning_rate": 1.9934331924983222e-05, "loss": 0.1714, "step": 571 }, { "epoch": 0.52, "learning_rate": 1.993390667071583e-05, "loss": 0.1685, "step": 572 }, { "epoch": 0.53, "learning_rate": 1.993348004852028e-05, "loss": 0.1763, "step": 573 }, { "epoch": 0.53, "learning_rate": 1.9933052058455325e-05, "loss": 0.179, "step": 574 }, { "epoch": 0.53, "learning_rate": 1.9932622700579903e-05, "loss": 0.1762, "step": 575 }, { "epoch": 0.53, "learning_rate": 1.9932191974953134e-05, "loss": 0.1676, "step": 576 }, { "epoch": 0.53, "learning_rate": 1.9931759881634334e-05, "loss": 0.1755, "step": 577 }, { "epoch": 0.53, "learning_rate": 1.9931326420682997e-05, "loss": 0.1787, "step": 578 }, { "epoch": 0.53, "learning_rate": 1.9930891592158814e-05, "loss": 0.1695, "step": 579 }, { "epoch": 0.53, "learning_rate": 1.9930455396121665e-05, "loss": 0.1762, "step": 580 }, { "epoch": 0.53, "learning_rate": 1.9930017832631616e-05, "loss": 0.1748, "step": 581 }, { "epoch": 0.53, "learning_rate": 1.9929578901748917e-05, "loss": 0.1842, "step": 582 }, { "epoch": 0.53, "learning_rate": 1.9929138603534013e-05, "loss": 0.1744, "step": 583 }, { "epoch": 0.54, "learning_rate": 1.9928696938047532e-05, "loss": 0.1808, "step": 584 }, { "epoch": 0.54, "learning_rate": 1.9928253905350297e-05, "loss": 0.1749, "step": 585 }, { "epoch": 0.54, "learning_rate": 1.9927809505503307e-05, "loss": 0.1825, "step": 586 }, { "epoch": 0.54, "learning_rate": 1.9927363738567767e-05, "loss": 0.1865, "step": 587 }, { "epoch": 0.54, "learning_rate": 1.9926916604605055e-05, "loss": 0.1762, "step": 588 }, { "epoch": 0.54, "learning_rate": 1.992646810367674e-05, "loss": 0.1821, "step": 589 }, { "epoch": 0.54, "learning_rate": 1.9926018235844586e-05, "loss": 0.1701, "step": 590 }, { "epoch": 0.54, "learning_rate": 1.9925567001170543e-05, "loss": 0.1856, "step": 591 }, { "epoch": 0.54, "learning_rate": 1.992511439971674e-05, "loss": 0.1725, "step": 592 }, { "epoch": 0.54, "learning_rate": 1.9924660431545512e-05, "loss": 0.1759, "step": 593 }, { "epoch": 0.54, "learning_rate": 1.992420509671936e-05, "loss": 0.1791, "step": 594 }, { "epoch": 0.55, "learning_rate": 1.992374839530099e-05, "loss": 0.1594, "step": 595 }, { "epoch": 0.55, "learning_rate": 1.9923290327353296e-05, "loss": 0.1756, "step": 596 }, { "epoch": 0.55, "learning_rate": 1.9922830892939352e-05, "loss": 0.17, "step": 597 }, { "epoch": 0.55, "learning_rate": 1.992237009212242e-05, "loss": 0.1782, "step": 598 }, { "epoch": 0.55, "learning_rate": 1.9921907924965954e-05, "loss": 0.1783, "step": 599 }, { "epoch": 0.55, "learning_rate": 1.9921444391533603e-05, "loss": 0.1713, "step": 600 }, { "epoch": 0.55, "learning_rate": 1.992097949188919e-05, "loss": 0.1713, "step": 601 }, { "epoch": 0.55, "learning_rate": 1.9920513226096735e-05, "loss": 0.1778, "step": 602 }, { "epoch": 0.55, "learning_rate": 1.9920045594220442e-05, "loss": 0.1792, "step": 603 }, { "epoch": 0.55, "learning_rate": 1.991957659632471e-05, "loss": 0.1792, "step": 604 }, { "epoch": 0.55, "learning_rate": 1.991910623247412e-05, "loss": 0.1762, "step": 605 }, { "epoch": 0.56, "learning_rate": 1.9918634502733437e-05, "loss": 0.1877, "step": 606 }, { "epoch": 0.56, "learning_rate": 1.9918161407167628e-05, "loss": 0.1646, "step": 607 }, { "epoch": 0.56, "learning_rate": 1.9917686945841833e-05, "loss": 0.1744, "step": 608 }, { "epoch": 0.56, "learning_rate": 1.9917211118821392e-05, "loss": 0.1736, "step": 609 }, { "epoch": 0.56, "learning_rate": 1.9916733926171824e-05, "loss": 0.1713, "step": 610 }, { "epoch": 0.56, "learning_rate": 1.991625536795884e-05, "loss": 0.1758, "step": 611 }, { "epoch": 0.56, "learning_rate": 1.991577544424834e-05, "loss": 0.1656, "step": 612 }, { "epoch": 0.56, "learning_rate": 1.991529415510641e-05, "loss": 0.1758, "step": 613 }, { "epoch": 0.56, "learning_rate": 1.9914811500599326e-05, "loss": 0.167, "step": 614 }, { "epoch": 0.56, "learning_rate": 1.991432748079355e-05, "loss": 0.1871, "step": 615 }, { "epoch": 0.56, "learning_rate": 1.9913842095755737e-05, "loss": 0.1818, "step": 616 }, { "epoch": 0.57, "learning_rate": 1.9913355345552717e-05, "loss": 0.18, "step": 617 }, { "epoch": 0.57, "learning_rate": 1.9912867230251528e-05, "loss": 0.1695, "step": 618 }, { "epoch": 0.57, "learning_rate": 1.9912377749919373e-05, "loss": 0.1721, "step": 619 }, { "epoch": 0.57, "learning_rate": 1.9911886904623666e-05, "loss": 0.1692, "step": 620 }, { "epoch": 0.57, "learning_rate": 1.991139469443199e-05, "loss": 0.1666, "step": 621 }, { "epoch": 0.57, "learning_rate": 1.9910901119412132e-05, "loss": 0.1718, "step": 622 }, { "epoch": 0.57, "learning_rate": 1.991040617963205e-05, "loss": 0.1851, "step": 623 }, { "epoch": 0.57, "learning_rate": 1.99099098751599e-05, "loss": 0.1723, "step": 624 }, { "epoch": 0.57, "learning_rate": 1.990941220606403e-05, "loss": 0.156, "step": 625 }, { "epoch": 0.57, "learning_rate": 1.9908913172412968e-05, "loss": 0.1731, "step": 626 }, { "epoch": 0.57, "learning_rate": 1.990841277427543e-05, "loss": 0.1805, "step": 627 }, { "epoch": 0.58, "learning_rate": 1.9907911011720323e-05, "loss": 0.1813, "step": 628 }, { "epoch": 0.58, "learning_rate": 1.990740788481674e-05, "loss": 0.1672, "step": 629 }, { "epoch": 0.58, "learning_rate": 1.990690339363397e-05, "loss": 0.1553, "step": 630 }, { "epoch": 0.58, "learning_rate": 1.9906397538241474e-05, "loss": 0.1762, "step": 631 }, { "epoch": 0.58, "learning_rate": 1.9905890318708918e-05, "loss": 0.1832, "step": 632 }, { "epoch": 0.58, "learning_rate": 1.9905381735106145e-05, "loss": 0.1728, "step": 633 }, { "epoch": 0.58, "learning_rate": 1.990487178750318e-05, "loss": 0.1737, "step": 634 }, { "epoch": 0.58, "learning_rate": 1.9904360475970255e-05, "loss": 0.1715, "step": 635 }, { "epoch": 0.58, "learning_rate": 1.9903847800577777e-05, "loss": 0.1742, "step": 636 }, { "epoch": 0.58, "learning_rate": 1.990333376139634e-05, "loss": 0.181, "step": 637 }, { "epoch": 0.58, "learning_rate": 1.9902818358496728e-05, "loss": 0.1677, "step": 638 }, { "epoch": 0.59, "learning_rate": 1.990230159194992e-05, "loss": 0.1699, "step": 639 }, { "epoch": 0.59, "learning_rate": 1.9901783461827066e-05, "loss": 0.1797, "step": 640 }, { "epoch": 0.59, "learning_rate": 1.9901263968199524e-05, "loss": 0.1744, "step": 641 }, { "epoch": 0.59, "learning_rate": 1.9900743111138822e-05, "loss": 0.1671, "step": 642 }, { "epoch": 0.59, "learning_rate": 1.990022089071669e-05, "loss": 0.1673, "step": 643 }, { "epoch": 0.59, "learning_rate": 1.9899697307005036e-05, "loss": 0.1689, "step": 644 }, { "epoch": 0.59, "learning_rate": 1.989917236007596e-05, "loss": 0.1761, "step": 645 }, { "epoch": 0.59, "learning_rate": 1.989864605000175e-05, "loss": 0.1847, "step": 646 }, { "epoch": 0.59, "learning_rate": 1.9898118376854873e-05, "loss": 0.1734, "step": 647 }, { "epoch": 0.59, "learning_rate": 1.9897589340708002e-05, "loss": 0.1856, "step": 648 }, { "epoch": 0.59, "learning_rate": 1.989705894163398e-05, "loss": 0.1711, "step": 649 }, { "epoch": 0.6, "learning_rate": 1.989652717970584e-05, "loss": 0.1624, "step": 650 }, { "epoch": 0.6, "learning_rate": 1.989599405499682e-05, "loss": 0.1762, "step": 651 }, { "epoch": 0.6, "learning_rate": 1.9895459567580326e-05, "loss": 0.1852, "step": 652 }, { "epoch": 0.6, "learning_rate": 1.9894923717529956e-05, "loss": 0.1673, "step": 653 }, { "epoch": 0.6, "learning_rate": 1.9894386504919507e-05, "loss": 0.1662, "step": 654 }, { "epoch": 0.6, "learning_rate": 1.989384792982294e-05, "loss": 0.1623, "step": 655 }, { "epoch": 0.6, "learning_rate": 1.9893307992314433e-05, "loss": 0.1656, "step": 656 }, { "epoch": 0.6, "learning_rate": 1.9892766692468327e-05, "loss": 0.1731, "step": 657 }, { "epoch": 0.6, "learning_rate": 1.9892224030359167e-05, "loss": 0.1768, "step": 658 }, { "epoch": 0.6, "learning_rate": 1.9891680006061673e-05, "loss": 0.1784, "step": 659 }, { "epoch": 0.6, "learning_rate": 1.9891134619650764e-05, "loss": 0.1737, "step": 660 }, { "epoch": 0.61, "learning_rate": 1.989058787120154e-05, "loss": 0.1779, "step": 661 }, { "epoch": 0.61, "learning_rate": 1.9890039760789294e-05, "loss": 0.1691, "step": 662 }, { "epoch": 0.61, "learning_rate": 1.988949028848949e-05, "loss": 0.1726, "step": 663 }, { "epoch": 0.61, "learning_rate": 1.9888939454377806e-05, "loss": 0.1829, "step": 664 }, { "epoch": 0.61, "learning_rate": 1.9888387258530084e-05, "loss": 0.1791, "step": 665 }, { "epoch": 0.61, "learning_rate": 1.9887833701022366e-05, "loss": 0.1773, "step": 666 }, { "epoch": 0.61, "learning_rate": 1.988727878193088e-05, "loss": 0.1705, "step": 667 }, { "epoch": 0.61, "learning_rate": 1.988672250133204e-05, "loss": 0.1797, "step": 668 }, { "epoch": 0.61, "learning_rate": 1.9886164859302447e-05, "loss": 0.1703, "step": 669 }, { "epoch": 0.61, "learning_rate": 1.9885605855918887e-05, "loss": 0.1687, "step": 670 }, { "epoch": 0.62, "learning_rate": 1.988504549125834e-05, "loss": 0.1824, "step": 671 }, { "epoch": 0.62, "learning_rate": 1.9884483765397968e-05, "loss": 0.176, "step": 672 }, { "epoch": 0.62, "learning_rate": 1.9883920678415124e-05, "loss": 0.1696, "step": 673 }, { "epoch": 0.62, "learning_rate": 1.9883356230387344e-05, "loss": 0.178, "step": 674 }, { "epoch": 0.62, "learning_rate": 1.9882790421392354e-05, "loss": 0.1766, "step": 675 }, { "epoch": 0.62, "learning_rate": 1.9882223251508074e-05, "loss": 0.1793, "step": 676 }, { "epoch": 0.62, "learning_rate": 1.9881654720812594e-05, "loss": 0.1709, "step": 677 }, { "epoch": 0.62, "learning_rate": 1.9881084829384214e-05, "loss": 0.1782, "step": 678 }, { "epoch": 0.62, "learning_rate": 1.9880513577301403e-05, "loss": 0.1728, "step": 679 }, { "epoch": 0.62, "learning_rate": 1.9879940964642827e-05, "loss": 0.1756, "step": 680 }, { "epoch": 0.62, "learning_rate": 1.987936699148733e-05, "loss": 0.1803, "step": 681 }, { "epoch": 0.63, "learning_rate": 1.987879165791396e-05, "loss": 0.1669, "step": 682 }, { "epoch": 0.63, "learning_rate": 1.9878214964001933e-05, "loss": 0.1792, "step": 683 }, { "epoch": 0.63, "learning_rate": 1.9877636909830665e-05, "loss": 0.1736, "step": 684 }, { "epoch": 0.63, "learning_rate": 1.9877057495479758e-05, "loss": 0.1664, "step": 685 }, { "epoch": 0.63, "learning_rate": 1.9876476721028996e-05, "loss": 0.1744, "step": 686 }, { "epoch": 0.63, "learning_rate": 1.9875894586558357e-05, "loss": 0.1684, "step": 687 }, { "epoch": 0.63, "learning_rate": 1.9875311092148e-05, "loss": 0.185, "step": 688 }, { "epoch": 0.63, "learning_rate": 1.987472623787827e-05, "loss": 0.1745, "step": 689 }, { "epoch": 0.63, "learning_rate": 1.9874140023829706e-05, "loss": 0.1704, "step": 690 }, { "epoch": 0.63, "learning_rate": 1.9873552450083036e-05, "loss": 0.1732, "step": 691 }, { "epoch": 0.63, "learning_rate": 1.9872963516719164e-05, "loss": 0.1732, "step": 692 }, { "epoch": 0.64, "learning_rate": 1.9872373223819196e-05, "loss": 0.1685, "step": 693 }, { "epoch": 0.64, "learning_rate": 1.987178157146441e-05, "loss": 0.1767, "step": 694 }, { "epoch": 0.64, "learning_rate": 1.987118855973628e-05, "loss": 0.1749, "step": 695 }, { "epoch": 0.64, "learning_rate": 1.9870594188716463e-05, "loss": 0.1773, "step": 696 }, { "epoch": 0.64, "learning_rate": 1.986999845848681e-05, "loss": 0.1755, "step": 697 }, { "epoch": 0.64, "learning_rate": 1.9869401369129355e-05, "loss": 0.1751, "step": 698 }, { "epoch": 0.64, "learning_rate": 1.986880292072632e-05, "loss": 0.1844, "step": 699 }, { "epoch": 0.64, "learning_rate": 1.9868203113360102e-05, "loss": 0.172, "step": 700 }, { "epoch": 0.64, "learning_rate": 1.986760194711331e-05, "loss": 0.18, "step": 701 }, { "epoch": 0.64, "learning_rate": 1.986699942206872e-05, "loss": 0.1773, "step": 702 }, { "epoch": 0.64, "learning_rate": 1.9866395538309302e-05, "loss": 0.1681, "step": 703 }, { "epoch": 0.65, "learning_rate": 1.9865790295918212e-05, "loss": 0.1688, "step": 704 }, { "epoch": 0.65, "learning_rate": 1.9865183694978798e-05, "loss": 0.177, "step": 705 }, { "epoch": 0.65, "learning_rate": 1.9864575735574583e-05, "loss": 0.1753, "step": 706 }, { "epoch": 0.65, "learning_rate": 1.986396641778929e-05, "loss": 0.1627, "step": 707 }, { "epoch": 0.65, "learning_rate": 1.9863355741706823e-05, "loss": 0.1638, "step": 708 }, { "epoch": 0.65, "learning_rate": 1.9862743707411275e-05, "loss": 0.1737, "step": 709 }, { "epoch": 0.65, "learning_rate": 1.9862130314986924e-05, "loss": 0.1697, "step": 710 }, { "epoch": 0.65, "learning_rate": 1.9861515564518234e-05, "loss": 0.1696, "step": 711 }, { "epoch": 0.65, "learning_rate": 1.9860899456089863e-05, "loss": 0.1691, "step": 712 }, { "epoch": 0.65, "learning_rate": 1.9860281989786644e-05, "loss": 0.1742, "step": 713 }, { "epoch": 0.65, "learning_rate": 1.985966316569361e-05, "loss": 0.1624, "step": 714 }, { "epoch": 0.66, "learning_rate": 1.9859042983895967e-05, "loss": 0.1756, "step": 715 }, { "epoch": 0.66, "learning_rate": 1.9858421444479126e-05, "loss": 0.1582, "step": 716 }, { "epoch": 0.66, "learning_rate": 1.985779854752867e-05, "loss": 0.1751, "step": 717 }, { "epoch": 0.66, "learning_rate": 1.9857174293130374e-05, "loss": 0.1764, "step": 718 }, { "epoch": 0.66, "learning_rate": 1.98565486813702e-05, "loss": 0.1712, "step": 719 }, { "epoch": 0.66, "learning_rate": 1.9855921712334293e-05, "loss": 0.17, "step": 720 }, { "epoch": 0.66, "learning_rate": 1.9855293386108995e-05, "loss": 0.1793, "step": 721 }, { "epoch": 0.66, "learning_rate": 1.985466370278082e-05, "loss": 0.171, "step": 722 }, { "epoch": 0.66, "learning_rate": 1.9854032662436486e-05, "loss": 0.174, "step": 723 }, { "epoch": 0.66, "learning_rate": 1.9853400265162884e-05, "loss": 0.1697, "step": 724 }, { "epoch": 0.66, "learning_rate": 1.9852766511047098e-05, "loss": 0.1705, "step": 725 }, { "epoch": 0.67, "learning_rate": 1.98521314001764e-05, "loss": 0.1711, "step": 726 }, { "epoch": 0.67, "learning_rate": 1.9851494932638246e-05, "loss": 0.1814, "step": 727 }, { "epoch": 0.67, "learning_rate": 1.985085710852027e-05, "loss": 0.1657, "step": 728 }, { "epoch": 0.67, "learning_rate": 1.9850217927910318e-05, "loss": 0.1759, "step": 729 }, { "epoch": 0.67, "learning_rate": 1.98495773908964e-05, "loss": 0.1677, "step": 730 }, { "epoch": 0.67, "learning_rate": 1.984893549756671e-05, "loss": 0.1591, "step": 731 }, { "epoch": 0.67, "learning_rate": 1.9848292248009657e-05, "loss": 0.1719, "step": 732 }, { "epoch": 0.67, "learning_rate": 1.9847647642313806e-05, "loss": 0.1774, "step": 733 }, { "epoch": 0.67, "learning_rate": 1.9847001680567925e-05, "loss": 0.169, "step": 734 }, { "epoch": 0.67, "learning_rate": 1.9846354362860964e-05, "loss": 0.1573, "step": 735 }, { "epoch": 0.67, "learning_rate": 1.984570568928206e-05, "loss": 0.1767, "step": 736 }, { "epoch": 0.68, "learning_rate": 1.9845055659920538e-05, "loss": 0.1721, "step": 737 }, { "epoch": 0.68, "learning_rate": 1.984440427486591e-05, "loss": 0.1721, "step": 738 }, { "epoch": 0.68, "learning_rate": 1.9843751534207868e-05, "loss": 0.1728, "step": 739 }, { "epoch": 0.68, "learning_rate": 1.9843097438036307e-05, "loss": 0.1599, "step": 740 }, { "epoch": 0.68, "learning_rate": 1.9842441986441286e-05, "loss": 0.1693, "step": 741 }, { "epoch": 0.68, "learning_rate": 1.9841785179513074e-05, "loss": 0.1604, "step": 742 }, { "epoch": 0.68, "learning_rate": 1.9841127017342108e-05, "loss": 0.1764, "step": 743 }, { "epoch": 0.68, "learning_rate": 1.984046750001902e-05, "loss": 0.1719, "step": 744 }, { "epoch": 0.68, "learning_rate": 1.9839806627634628e-05, "loss": 0.1551, "step": 745 }, { "epoch": 0.68, "learning_rate": 1.9839144400279936e-05, "loss": 0.1735, "step": 746 }, { "epoch": 0.68, "learning_rate": 1.9838480818046137e-05, "loss": 0.1683, "step": 747 }, { "epoch": 0.69, "learning_rate": 1.98378158810246e-05, "loss": 0.1855, "step": 748 }, { "epoch": 0.69, "learning_rate": 1.9837149589306897e-05, "loss": 0.1663, "step": 749 }, { "epoch": 0.69, "learning_rate": 1.983648194298478e-05, "loss": 0.1701, "step": 750 }, { "epoch": 0.69, "learning_rate": 1.983581294215018e-05, "loss": 0.172, "step": 751 }, { "epoch": 0.69, "learning_rate": 1.9835142586895218e-05, "loss": 0.1629, "step": 752 }, { "epoch": 0.69, "learning_rate": 1.9834470877312213e-05, "loss": 0.1706, "step": 753 }, { "epoch": 0.69, "learning_rate": 1.9833797813493656e-05, "loss": 0.1677, "step": 754 }, { "epoch": 0.69, "learning_rate": 1.9833123395532227e-05, "loss": 0.1646, "step": 755 }, { "epoch": 0.69, "learning_rate": 1.9832447623520802e-05, "loss": 0.1633, "step": 756 }, { "epoch": 0.69, "learning_rate": 1.9831770497552432e-05, "loss": 0.1721, "step": 757 }, { "epoch": 0.69, "learning_rate": 1.9831092017720365e-05, "loss": 0.1688, "step": 758 }, { "epoch": 0.7, "learning_rate": 1.983041218411802e-05, "loss": 0.1601, "step": 759 }, { "epoch": 0.7, "learning_rate": 1.982973099683902e-05, "loss": 0.1796, "step": 760 }, { "epoch": 0.7, "learning_rate": 1.9829048455977166e-05, "loss": 0.1694, "step": 761 }, { "epoch": 0.7, "learning_rate": 1.982836456162644e-05, "loss": 0.1732, "step": 762 }, { "epoch": 0.7, "learning_rate": 1.982767931388102e-05, "loss": 0.1779, "step": 763 }, { "epoch": 0.7, "learning_rate": 1.9826992712835268e-05, "loss": 0.1705, "step": 764 }, { "epoch": 0.7, "learning_rate": 1.9826304758583732e-05, "loss": 0.1676, "step": 765 }, { "epoch": 0.7, "learning_rate": 1.9825615451221142e-05, "loss": 0.1723, "step": 766 }, { "epoch": 0.7, "learning_rate": 1.9824924790842418e-05, "loss": 0.1765, "step": 767 }, { "epoch": 0.7, "learning_rate": 1.9824232777542666e-05, "loss": 0.1616, "step": 768 }, { "epoch": 0.7, "learning_rate": 1.9823539411417178e-05, "loss": 0.1718, "step": 769 }, { "epoch": 0.71, "learning_rate": 1.982284469256144e-05, "loss": 0.168, "step": 770 }, { "epoch": 0.71, "learning_rate": 1.9822148621071102e-05, "loss": 0.1694, "step": 771 }, { "epoch": 0.71, "learning_rate": 1.9821451197042028e-05, "loss": 0.1713, "step": 772 }, { "epoch": 0.71, "learning_rate": 1.982075242057025e-05, "loss": 0.1742, "step": 773 }, { "epoch": 0.71, "learning_rate": 1.9820052291751993e-05, "loss": 0.1828, "step": 774 }, { "epoch": 0.71, "learning_rate": 1.9819350810683665e-05, "loss": 0.1739, "step": 775 }, { "epoch": 0.71, "learning_rate": 1.9818647977461864e-05, "loss": 0.1728, "step": 776 }, { "epoch": 0.71, "learning_rate": 1.9817943792183373e-05, "loss": 0.168, "step": 777 }, { "epoch": 0.71, "learning_rate": 1.9817238254945158e-05, "loss": 0.169, "step": 778 }, { "epoch": 0.71, "learning_rate": 1.9816531365844373e-05, "loss": 0.1692, "step": 779 }, { "epoch": 0.71, "learning_rate": 1.9815823124978357e-05, "loss": 0.1686, "step": 780 }, { "epoch": 0.72, "learning_rate": 1.9815113532444645e-05, "loss": 0.1782, "step": 781 }, { "epoch": 0.72, "learning_rate": 1.9814402588340944e-05, "loss": 0.1751, "step": 782 }, { "epoch": 0.72, "learning_rate": 1.981369029276515e-05, "loss": 0.1588, "step": 783 }, { "epoch": 0.72, "learning_rate": 1.9812976645815357e-05, "loss": 0.1626, "step": 784 }, { "epoch": 0.72, "learning_rate": 1.981226164758983e-05, "loss": 0.1685, "step": 785 }, { "epoch": 0.72, "learning_rate": 1.9811545298187027e-05, "loss": 0.178, "step": 786 }, { "epoch": 0.72, "learning_rate": 1.9810827597705595e-05, "loss": 0.1698, "step": 787 }, { "epoch": 0.72, "learning_rate": 1.9810108546244355e-05, "loss": 0.1612, "step": 788 }, { "epoch": 0.72, "learning_rate": 1.9809388143902334e-05, "loss": 0.1685, "step": 789 }, { "epoch": 0.72, "learning_rate": 1.9808666390778724e-05, "loss": 0.1798, "step": 790 }, { "epoch": 0.73, "learning_rate": 1.9807943286972918e-05, "loss": 0.1711, "step": 791 }, { "epoch": 0.73, "learning_rate": 1.9807218832584488e-05, "loss": 0.1699, "step": 792 }, { "epoch": 0.73, "learning_rate": 1.980649302771319e-05, "loss": 0.1584, "step": 793 }, { "epoch": 0.73, "learning_rate": 1.9805765872458976e-05, "loss": 0.1599, "step": 794 }, { "epoch": 0.73, "learning_rate": 1.9805037366921973e-05, "loss": 0.1669, "step": 795 }, { "epoch": 0.73, "learning_rate": 1.98043075112025e-05, "loss": 0.1661, "step": 796 }, { "epoch": 0.73, "learning_rate": 1.980357630540106e-05, "loss": 0.1692, "step": 797 }, { "epoch": 0.73, "learning_rate": 1.9802843749618338e-05, "loss": 0.1628, "step": 798 }, { "epoch": 0.73, "learning_rate": 1.980210984395522e-05, "loss": 0.1705, "step": 799 }, { "epoch": 0.73, "learning_rate": 1.9801374588512754e-05, "loss": 0.1611, "step": 800 }, { "epoch": 0.73, "learning_rate": 1.9800637983392193e-05, "loss": 0.1769, "step": 801 }, { "epoch": 0.74, "learning_rate": 1.9799900028694972e-05, "loss": 0.1744, "step": 802 }, { "epoch": 0.74, "learning_rate": 1.979916072452271e-05, "loss": 0.1638, "step": 803 }, { "epoch": 0.74, "learning_rate": 1.9798420070977205e-05, "loss": 0.1686, "step": 804 }, { "epoch": 0.74, "learning_rate": 1.9797678068160448e-05, "loss": 0.1648, "step": 805 }, { "epoch": 0.74, "learning_rate": 1.979693471617462e-05, "loss": 0.1643, "step": 806 }, { "epoch": 0.74, "learning_rate": 1.979619001512208e-05, "loss": 0.1699, "step": 807 }, { "epoch": 0.74, "learning_rate": 1.9795443965105377e-05, "loss": 0.1774, "step": 808 }, { "epoch": 0.74, "learning_rate": 1.9794696566227242e-05, "loss": 0.1868, "step": 809 }, { "epoch": 0.74, "learning_rate": 1.9793947818590596e-05, "loss": 0.1636, "step": 810 }, { "epoch": 0.74, "learning_rate": 1.9793197722298544e-05, "loss": 0.1662, "step": 811 }, { "epoch": 0.74, "learning_rate": 1.9792446277454373e-05, "loss": 0.1645, "step": 812 }, { "epoch": 0.75, "learning_rate": 1.9791693484161562e-05, "loss": 0.1721, "step": 813 }, { "epoch": 0.75, "learning_rate": 1.9790939342523773e-05, "loss": 0.1711, "step": 814 }, { "epoch": 0.75, "learning_rate": 1.979018385264485e-05, "loss": 0.1606, "step": 815 }, { "epoch": 0.75, "learning_rate": 1.9789427014628836e-05, "loss": 0.1674, "step": 816 }, { "epoch": 0.75, "learning_rate": 1.9788668828579936e-05, "loss": 0.1724, "step": 817 }, { "epoch": 0.75, "learning_rate": 1.9787909294602564e-05, "loss": 0.1657, "step": 818 }, { "epoch": 0.75, "learning_rate": 1.9787148412801307e-05, "loss": 0.168, "step": 819 }, { "epoch": 0.75, "learning_rate": 1.9786386183280937e-05, "loss": 0.1653, "step": 820 }, { "epoch": 0.75, "learning_rate": 1.9785622606146426e-05, "loss": 0.1621, "step": 821 }, { "epoch": 0.75, "learning_rate": 1.9784857681502912e-05, "loss": 0.1674, "step": 822 }, { "epoch": 0.75, "learning_rate": 1.9784091409455728e-05, "loss": 0.1666, "step": 823 }, { "epoch": 0.76, "learning_rate": 1.9783323790110396e-05, "loss": 0.1655, "step": 824 }, { "epoch": 0.76, "learning_rate": 1.9782554823572615e-05, "loss": 0.1619, "step": 825 }, { "epoch": 0.76, "learning_rate": 1.9781784509948275e-05, "loss": 0.1675, "step": 826 }, { "epoch": 0.76, "learning_rate": 1.9781012849343452e-05, "loss": 0.1645, "step": 827 }, { "epoch": 0.76, "learning_rate": 1.9780239841864407e-05, "loss": 0.1642, "step": 828 }, { "epoch": 0.76, "learning_rate": 1.9779465487617583e-05, "loss": 0.176, "step": 829 }, { "epoch": 0.76, "learning_rate": 1.977868978670961e-05, "loss": 0.1604, "step": 830 }, { "epoch": 0.76, "learning_rate": 1.9777912739247305e-05, "loss": 0.1806, "step": 831 }, { "epoch": 0.76, "learning_rate": 1.977713434533767e-05, "loss": 0.1751, "step": 832 }, { "epoch": 0.76, "learning_rate": 1.9776354605087896e-05, "loss": 0.162, "step": 833 }, { "epoch": 0.76, "learning_rate": 1.9775573518605348e-05, "loss": 0.1717, "step": 834 }, { "epoch": 0.77, "learning_rate": 1.9774791085997593e-05, "loss": 0.1708, "step": 835 }, { "epoch": 0.77, "learning_rate": 1.9774007307372366e-05, "loss": 0.1608, "step": 836 }, { "epoch": 0.77, "learning_rate": 1.9773222182837597e-05, "loss": 0.172, "step": 837 }, { "epoch": 0.77, "learning_rate": 1.9772435712501404e-05, "loss": 0.1694, "step": 838 }, { "epoch": 0.77, "learning_rate": 1.9771647896472083e-05, "loss": 0.1678, "step": 839 }, { "epoch": 0.77, "learning_rate": 1.9770858734858123e-05, "loss": 0.1665, "step": 840 }, { "epoch": 0.77, "learning_rate": 1.9770068227768188e-05, "loss": 0.1661, "step": 841 }, { "epoch": 0.77, "learning_rate": 1.9769276375311136e-05, "loss": 0.169, "step": 842 }, { "epoch": 0.77, "learning_rate": 1.9768483177596008e-05, "loss": 0.1651, "step": 843 }, { "epoch": 0.77, "learning_rate": 1.9767688634732028e-05, "loss": 0.1627, "step": 844 }, { "epoch": 0.77, "learning_rate": 1.976689274682861e-05, "loss": 0.1686, "step": 845 }, { "epoch": 0.78, "learning_rate": 1.9766095513995344e-05, "loss": 0.17, "step": 846 }, { "epoch": 0.78, "learning_rate": 1.976529693634202e-05, "loss": 0.1689, "step": 847 }, { "epoch": 0.78, "learning_rate": 1.9764497013978594e-05, "loss": 0.161, "step": 848 }, { "epoch": 0.78, "learning_rate": 1.9763695747015227e-05, "loss": 0.1742, "step": 849 }, { "epoch": 0.78, "learning_rate": 1.976289313556225e-05, "loss": 0.178, "step": 850 }, { "epoch": 0.78, "learning_rate": 1.9762089179730188e-05, "loss": 0.1738, "step": 851 }, { "epoch": 0.78, "learning_rate": 1.9761283879629743e-05, "loss": 0.1723, "step": 852 }, { "epoch": 0.78, "learning_rate": 1.9760477235371817e-05, "loss": 0.1646, "step": 853 }, { "epoch": 0.78, "learning_rate": 1.975966924706748e-05, "loss": 0.1714, "step": 854 }, { "epoch": 0.78, "learning_rate": 1.9758859914827997e-05, "loss": 0.1726, "step": 855 }, { "epoch": 0.78, "learning_rate": 1.9758049238764815e-05, "loss": 0.1634, "step": 856 }, { "epoch": 0.79, "learning_rate": 1.9757237218989563e-05, "loss": 0.1646, "step": 857 }, { "epoch": 0.79, "learning_rate": 1.975642385561406e-05, "loss": 0.1614, "step": 858 }, { "epoch": 0.79, "learning_rate": 1.975560914875031e-05, "loss": 0.1645, "step": 859 }, { "epoch": 0.79, "learning_rate": 1.9754793098510505e-05, "loss": 0.1675, "step": 860 }, { "epoch": 0.79, "learning_rate": 1.9753975705007012e-05, "loss": 0.1698, "step": 861 }, { "epoch": 0.79, "learning_rate": 1.9753156968352388e-05, "loss": 0.1626, "step": 862 }, { "epoch": 0.79, "learning_rate": 1.9752336888659374e-05, "loss": 0.1744, "step": 863 }, { "epoch": 0.79, "learning_rate": 1.9751515466040905e-05, "loss": 0.1697, "step": 864 }, { "epoch": 0.79, "learning_rate": 1.9750692700610087e-05, "loss": 0.1704, "step": 865 }, { "epoch": 0.79, "learning_rate": 1.974986859248022e-05, "loss": 0.1549, "step": 866 }, { "epoch": 0.79, "learning_rate": 1.974904314176478e-05, "loss": 0.1641, "step": 867 }, { "epoch": 0.8, "learning_rate": 1.974821634857744e-05, "loss": 0.1713, "step": 868 }, { "epoch": 0.8, "learning_rate": 1.9747388213032056e-05, "loss": 0.1801, "step": 869 }, { "epoch": 0.8, "learning_rate": 1.9746558735242657e-05, "loss": 0.1694, "step": 870 }, { "epoch": 0.8, "learning_rate": 1.9745727915323468e-05, "loss": 0.1608, "step": 871 }, { "epoch": 0.8, "learning_rate": 1.9744895753388892e-05, "loss": 0.1739, "step": 872 }, { "epoch": 0.8, "learning_rate": 1.974406224955352e-05, "loss": 0.178, "step": 873 }, { "epoch": 0.8, "learning_rate": 1.9743227403932135e-05, "loss": 0.1535, "step": 874 }, { "epoch": 0.8, "learning_rate": 1.9742391216639692e-05, "loss": 0.1786, "step": 875 }, { "epoch": 0.8, "learning_rate": 1.9741553687791338e-05, "loss": 0.1637, "step": 876 }, { "epoch": 0.8, "learning_rate": 1.97407148175024e-05, "loss": 0.1746, "step": 877 }, { "epoch": 0.8, "learning_rate": 1.97398746058884e-05, "loss": 0.1568, "step": 878 }, { "epoch": 0.81, "learning_rate": 1.9739033053065027e-05, "loss": 0.1665, "step": 879 }, { "epoch": 0.81, "learning_rate": 1.973819015914818e-05, "loss": 0.166, "step": 880 }, { "epoch": 0.81, "learning_rate": 1.9737345924253913e-05, "loss": 0.1693, "step": 881 }, { "epoch": 0.81, "learning_rate": 1.9736500348498485e-05, "loss": 0.1553, "step": 882 }, { "epoch": 0.81, "learning_rate": 1.973565343199834e-05, "loss": 0.177, "step": 883 }, { "epoch": 0.81, "learning_rate": 1.9734805174870094e-05, "loss": 0.1618, "step": 884 }, { "epoch": 0.81, "learning_rate": 1.973395557723056e-05, "loss": 0.1781, "step": 885 }, { "epoch": 0.81, "learning_rate": 1.9733104639196724e-05, "loss": 0.1623, "step": 886 }, { "epoch": 0.81, "learning_rate": 1.9732252360885763e-05, "loss": 0.1655, "step": 887 }, { "epoch": 0.81, "learning_rate": 1.9731398742415044e-05, "loss": 0.1683, "step": 888 }, { "epoch": 0.81, "learning_rate": 1.973054378390211e-05, "loss": 0.1612, "step": 889 }, { "epoch": 0.82, "learning_rate": 1.972968748546469e-05, "loss": 0.1623, "step": 890 }, { "epoch": 0.82, "learning_rate": 1.97288298472207e-05, "loss": 0.1617, "step": 891 }, { "epoch": 0.82, "learning_rate": 1.972797086928824e-05, "loss": 0.1646, "step": 892 }, { "epoch": 0.82, "learning_rate": 1.972711055178559e-05, "loss": 0.1616, "step": 893 }, { "epoch": 0.82, "learning_rate": 1.9726248894831222e-05, "loss": 0.1697, "step": 894 }, { "epoch": 0.82, "learning_rate": 1.9725385898543784e-05, "loss": 0.1653, "step": 895 }, { "epoch": 0.82, "learning_rate": 1.9724521563042124e-05, "loss": 0.1679, "step": 896 }, { "epoch": 0.82, "learning_rate": 1.972365588844525e-05, "loss": 0.1708, "step": 897 }, { "epoch": 0.82, "learning_rate": 1.972278887487238e-05, "loss": 0.171, "step": 898 }, { "epoch": 0.82, "learning_rate": 1.9721920522442898e-05, "loss": 0.1742, "step": 899 }, { "epoch": 0.82, "learning_rate": 1.9721050831276372e-05, "loss": 0.1742, "step": 900 }, { "epoch": 0.83, "learning_rate": 1.9720179801492577e-05, "loss": 0.1738, "step": 901 }, { "epoch": 0.83, "learning_rate": 1.9719307433211442e-05, "loss": 0.1696, "step": 902 }, { "epoch": 0.83, "learning_rate": 1.9718433726553106e-05, "loss": 0.161, "step": 903 }, { "epoch": 0.83, "learning_rate": 1.9717558681637873e-05, "loss": 0.1733, "step": 904 }, { "epoch": 0.83, "learning_rate": 1.971668229858624e-05, "loss": 0.17, "step": 905 }, { "epoch": 0.83, "learning_rate": 1.9715804577518887e-05, "loss": 0.1697, "step": 906 }, { "epoch": 0.83, "learning_rate": 1.9714925518556686e-05, "loss": 0.1628, "step": 907 }, { "epoch": 0.83, "learning_rate": 1.9714045121820676e-05, "loss": 0.1604, "step": 908 }, { "epoch": 0.83, "learning_rate": 1.97131633874321e-05, "loss": 0.1643, "step": 909 }, { "epoch": 0.83, "learning_rate": 1.971228031551237e-05, "loss": 0.1773, "step": 910 }, { "epoch": 0.84, "learning_rate": 1.9711395906183086e-05, "loss": 0.1705, "step": 911 }, { "epoch": 0.84, "learning_rate": 1.9710510159566033e-05, "loss": 0.1636, "step": 912 }, { "epoch": 0.84, "learning_rate": 1.970962307578319e-05, "loss": 0.1709, "step": 913 }, { "epoch": 0.84, "learning_rate": 1.9708734654956703e-05, "loss": 0.1632, "step": 914 }, { "epoch": 0.84, "learning_rate": 1.9707844897208907e-05, "loss": 0.169, "step": 915 }, { "epoch": 0.84, "learning_rate": 1.9706953802662333e-05, "loss": 0.1644, "step": 916 }, { "epoch": 0.84, "learning_rate": 1.9706061371439686e-05, "loss": 0.1648, "step": 917 }, { "epoch": 0.84, "learning_rate": 1.970516760366385e-05, "loss": 0.1616, "step": 918 }, { "epoch": 0.84, "learning_rate": 1.9704272499457908e-05, "loss": 0.1692, "step": 919 }, { "epoch": 0.84, "learning_rate": 1.970337605894511e-05, "loss": 0.1585, "step": 920 }, { "epoch": 0.84, "learning_rate": 1.9702478282248904e-05, "loss": 0.143, "step": 921 }, { "epoch": 0.85, "learning_rate": 1.9701579169492916e-05, "loss": 0.165, "step": 922 }, { "epoch": 0.85, "learning_rate": 1.970067872080095e-05, "loss": 0.1715, "step": 923 }, { "epoch": 0.85, "learning_rate": 1.9699776936297015e-05, "loss": 0.1667, "step": 924 }, { "epoch": 0.85, "learning_rate": 1.9698873816105272e-05, "loss": 0.1653, "step": 925 }, { "epoch": 0.85, "learning_rate": 1.9697969360350098e-05, "loss": 0.1645, "step": 926 }, { "epoch": 0.85, "learning_rate": 1.9697063569156027e-05, "loss": 0.1594, "step": 927 }, { "epoch": 0.85, "learning_rate": 1.96961564426478e-05, "loss": 0.1636, "step": 928 }, { "epoch": 0.85, "learning_rate": 1.9695247980950323e-05, "loss": 0.1696, "step": 929 }, { "epoch": 0.85, "learning_rate": 1.9694338184188696e-05, "loss": 0.1662, "step": 930 }, { "epoch": 0.85, "learning_rate": 1.96934270524882e-05, "loss": 0.1627, "step": 931 }, { "epoch": 0.85, "learning_rate": 1.9692514585974306e-05, "loss": 0.1633, "step": 932 }, { "epoch": 0.86, "learning_rate": 1.9691600784772657e-05, "loss": 0.17, "step": 933 }, { "epoch": 0.86, "learning_rate": 1.969068564900909e-05, "loss": 0.1686, "step": 934 }, { "epoch": 0.86, "learning_rate": 1.968976917880962e-05, "loss": 0.1748, "step": 935 }, { "epoch": 0.86, "learning_rate": 1.968885137430044e-05, "loss": 0.1668, "step": 936 }, { "epoch": 0.86, "learning_rate": 1.9687932235607953e-05, "loss": 0.1565, "step": 937 }, { "epoch": 0.86, "learning_rate": 1.9687011762858713e-05, "loss": 0.1659, "step": 938 }, { "epoch": 0.86, "learning_rate": 1.9686089956179473e-05, "loss": 0.1566, "step": 939 }, { "epoch": 0.86, "learning_rate": 1.968516681569717e-05, "loss": 0.1671, "step": 940 }, { "epoch": 0.86, "learning_rate": 1.9684242341538926e-05, "loss": 0.1735, "step": 941 }, { "epoch": 0.86, "learning_rate": 1.968331653383204e-05, "loss": 0.1623, "step": 942 }, { "epoch": 0.86, "learning_rate": 1.9682389392704003e-05, "loss": 0.1659, "step": 943 }, { "epoch": 0.87, "learning_rate": 1.968146091828248e-05, "loss": 0.1615, "step": 944 }, { "epoch": 0.87, "learning_rate": 1.9680531110695328e-05, "loss": 0.1778, "step": 945 }, { "epoch": 0.87, "learning_rate": 1.967959997007058e-05, "loss": 0.1587, "step": 946 }, { "epoch": 0.87, "learning_rate": 1.9678667496536465e-05, "loss": 0.1688, "step": 947 }, { "epoch": 0.87, "learning_rate": 1.9677733690221375e-05, "loss": 0.1608, "step": 948 }, { "epoch": 0.87, "learning_rate": 1.967679855125391e-05, "loss": 0.1656, "step": 949 }, { "epoch": 0.87, "learning_rate": 1.9675862079762838e-05, "loss": 0.1666, "step": 950 }, { "epoch": 0.87, "learning_rate": 1.967492427587711e-05, "loss": 0.171, "step": 951 }, { "epoch": 0.87, "learning_rate": 1.9673985139725864e-05, "loss": 0.1598, "step": 952 }, { "epoch": 0.87, "learning_rate": 1.967304467143843e-05, "loss": 0.1555, "step": 953 }, { "epoch": 0.87, "learning_rate": 1.9672102871144302e-05, "loss": 0.1652, "step": 954 }, { "epoch": 0.88, "learning_rate": 1.9671159738973177e-05, "loss": 0.158, "step": 955 }, { "epoch": 0.88, "learning_rate": 1.9670215275054923e-05, "loss": 0.1758, "step": 956 }, { "epoch": 0.88, "learning_rate": 1.96692694795196e-05, "loss": 0.1667, "step": 957 }, { "epoch": 0.88, "learning_rate": 1.966832235249744e-05, "loss": 0.166, "step": 958 }, { "epoch": 0.88, "learning_rate": 1.9667373894118872e-05, "loss": 0.1648, "step": 959 }, { "epoch": 0.88, "learning_rate": 1.9666424104514495e-05, "loss": 0.1644, "step": 960 }, { "epoch": 0.88, "learning_rate": 1.96654729838151e-05, "loss": 0.1624, "step": 961 }, { "epoch": 0.88, "learning_rate": 1.9664520532151664e-05, "loss": 0.1572, "step": 962 }, { "epoch": 0.88, "learning_rate": 1.9663566749655342e-05, "loss": 0.1609, "step": 963 }, { "epoch": 0.88, "learning_rate": 1.9662611636457464e-05, "loss": 0.1632, "step": 964 }, { "epoch": 0.88, "learning_rate": 1.966165519268956e-05, "loss": 0.1706, "step": 965 }, { "epoch": 0.89, "learning_rate": 1.9660697418483327e-05, "loss": 0.1643, "step": 966 }, { "epoch": 0.89, "learning_rate": 1.9659738313970663e-05, "loss": 0.1622, "step": 967 }, { "epoch": 0.89, "learning_rate": 1.9658777879283638e-05, "loss": 0.1601, "step": 968 }, { "epoch": 0.89, "learning_rate": 1.96578161145545e-05, "loss": 0.1587, "step": 969 }, { "epoch": 0.89, "learning_rate": 1.9656853019915692e-05, "loss": 0.1736, "step": 970 }, { "epoch": 0.89, "learning_rate": 1.9655888595499832e-05, "loss": 0.169, "step": 971 }, { "epoch": 0.89, "learning_rate": 1.965492284143973e-05, "loss": 0.1689, "step": 972 }, { "epoch": 0.89, "learning_rate": 1.9653955757868363e-05, "loss": 0.1664, "step": 973 }, { "epoch": 0.89, "learning_rate": 1.9652987344918912e-05, "loss": 0.1652, "step": 974 }, { "epoch": 0.89, "learning_rate": 1.9652017602724722e-05, "loss": 0.1624, "step": 975 }, { "epoch": 0.89, "learning_rate": 1.9651046531419335e-05, "loss": 0.1558, "step": 976 }, { "epoch": 0.9, "learning_rate": 1.9650074131136468e-05, "loss": 0.1635, "step": 977 }, { "epoch": 0.9, "learning_rate": 1.9649100402010024e-05, "loss": 0.173, "step": 978 }, { "epoch": 0.9, "learning_rate": 1.9648125344174085e-05, "loss": 0.1704, "step": 979 }, { "epoch": 0.9, "learning_rate": 1.9647148957762927e-05, "loss": 0.1629, "step": 980 }, { "epoch": 0.9, "learning_rate": 1.9646171242910993e-05, "loss": 0.1663, "step": 981 }, { "epoch": 0.9, "learning_rate": 1.964519219975292e-05, "loss": 0.1644, "step": 982 }, { "epoch": 0.9, "learning_rate": 1.964421182842353e-05, "loss": 0.1558, "step": 983 }, { "epoch": 0.9, "learning_rate": 1.9643230129057817e-05, "loss": 0.1613, "step": 984 }, { "epoch": 0.9, "learning_rate": 1.9642247101790966e-05, "loss": 0.16, "step": 985 }, { "epoch": 0.9, "learning_rate": 1.964126274675834e-05, "loss": 0.1721, "step": 986 }, { "epoch": 0.9, "learning_rate": 1.9640277064095494e-05, "loss": 0.1623, "step": 987 }, { "epoch": 0.91, "learning_rate": 1.9639290053938153e-05, "loss": 0.1712, "step": 988 }, { "epoch": 0.91, "learning_rate": 1.9638301716422236e-05, "loss": 0.176, "step": 989 }, { "epoch": 0.91, "learning_rate": 1.9637312051683835e-05, "loss": 0.1548, "step": 990 }, { "epoch": 0.91, "learning_rate": 1.9636321059859233e-05, "loss": 0.1622, "step": 991 }, { "epoch": 0.91, "learning_rate": 1.9635328741084892e-05, "loss": 0.1693, "step": 992 }, { "epoch": 0.91, "learning_rate": 1.9634335095497458e-05, "loss": 0.1584, "step": 993 }, { "epoch": 0.91, "learning_rate": 1.963334012323376e-05, "loss": 0.175, "step": 994 }, { "epoch": 0.91, "learning_rate": 1.9632343824430804e-05, "loss": 0.1594, "step": 995 }, { "epoch": 0.91, "learning_rate": 1.9631346199225783e-05, "loss": 0.1526, "step": 996 }, { "epoch": 0.91, "learning_rate": 1.9630347247756083e-05, "loss": 0.1709, "step": 997 }, { "epoch": 0.91, "learning_rate": 1.9629346970159253e-05, "loss": 0.1611, "step": 998 }, { "epoch": 0.92, "learning_rate": 1.9628345366573033e-05, "loss": 0.1633, "step": 999 }, { "epoch": 0.92, "learning_rate": 1.9627342437135356e-05, "loss": 0.1697, "step": 1000 }, { "epoch": 0.92, "learning_rate": 1.9626338181984317e-05, "loss": 0.1598, "step": 1001 }, { "epoch": 0.92, "learning_rate": 1.9625332601258214e-05, "loss": 0.1654, "step": 1002 }, { "epoch": 0.92, "learning_rate": 1.9624325695095514e-05, "loss": 0.1676, "step": 1003 }, { "epoch": 0.92, "learning_rate": 1.962331746363487e-05, "loss": 0.1543, "step": 1004 }, { "epoch": 0.92, "learning_rate": 1.9622307907015124e-05, "loss": 0.172, "step": 1005 }, { "epoch": 0.92, "learning_rate": 1.962129702537529e-05, "loss": 0.1588, "step": 1006 }, { "epoch": 0.92, "learning_rate": 1.9620284818854572e-05, "loss": 0.1605, "step": 1007 }, { "epoch": 0.92, "learning_rate": 1.961927128759235e-05, "loss": 0.1729, "step": 1008 }, { "epoch": 0.92, "learning_rate": 1.961825643172819e-05, "loss": 0.1728, "step": 1009 }, { "epoch": 0.93, "learning_rate": 1.961724025140185e-05, "loss": 0.1596, "step": 1010 }, { "epoch": 0.93, "learning_rate": 1.961622274675325e-05, "loss": 0.1598, "step": 1011 }, { "epoch": 0.93, "learning_rate": 1.961520391792251e-05, "loss": 0.1753, "step": 1012 }, { "epoch": 0.93, "learning_rate": 1.9614183765049922e-05, "loss": 0.1702, "step": 1013 }, { "epoch": 0.93, "learning_rate": 1.9613162288275968e-05, "loss": 0.1566, "step": 1014 }, { "epoch": 0.93, "learning_rate": 1.96121394877413e-05, "loss": 0.1719, "step": 1015 }, { "epoch": 0.93, "learning_rate": 1.9611115363586772e-05, "loss": 0.1539, "step": 1016 }, { "epoch": 0.93, "learning_rate": 1.9610089915953403e-05, "loss": 0.1628, "step": 1017 }, { "epoch": 0.93, "learning_rate": 1.9609063144982398e-05, "loss": 0.1738, "step": 1018 }, { "epoch": 0.93, "learning_rate": 1.960803505081515e-05, "loss": 0.169, "step": 1019 }, { "epoch": 0.93, "learning_rate": 1.960700563359323e-05, "loss": 0.1626, "step": 1020 }, { "epoch": 0.94, "learning_rate": 1.960597489345839e-05, "loss": 0.1668, "step": 1021 }, { "epoch": 0.94, "learning_rate": 1.960494283055257e-05, "loss": 0.1558, "step": 1022 }, { "epoch": 0.94, "learning_rate": 1.9603909445017885e-05, "loss": 0.1541, "step": 1023 }, { "epoch": 0.94, "learning_rate": 1.9602874736996632e-05, "loss": 0.1508, "step": 1024 }, { "epoch": 0.94, "learning_rate": 1.96018387066313e-05, "loss": 0.1636, "step": 1025 }, { "epoch": 0.94, "learning_rate": 1.9600801354064552e-05, "loss": 0.1565, "step": 1026 }, { "epoch": 0.94, "learning_rate": 1.959976267943923e-05, "loss": 0.1727, "step": 1027 }, { "epoch": 0.94, "learning_rate": 1.9598722682898365e-05, "loss": 0.163, "step": 1028 }, { "epoch": 0.94, "learning_rate": 1.959768136458517e-05, "loss": 0.1684, "step": 1029 }, { "epoch": 0.94, "learning_rate": 1.9596638724643034e-05, "loss": 0.1601, "step": 1030 }, { "epoch": 0.95, "learning_rate": 1.9595594763215537e-05, "loss": 0.1546, "step": 1031 }, { "epoch": 0.95, "learning_rate": 1.9594549480446426e-05, "loss": 0.1653, "step": 1032 }, { "epoch": 0.95, "learning_rate": 1.9593502876479648e-05, "loss": 0.156, "step": 1033 }, { "epoch": 0.95, "learning_rate": 1.959245495145932e-05, "loss": 0.1635, "step": 1034 }, { "epoch": 0.95, "learning_rate": 1.9591405705529745e-05, "loss": 0.1703, "step": 1035 }, { "epoch": 0.95, "learning_rate": 1.959035513883541e-05, "loss": 0.1639, "step": 1036 }, { "epoch": 0.95, "learning_rate": 1.9589303251520976e-05, "loss": 0.1647, "step": 1037 }, { "epoch": 0.95, "learning_rate": 1.958825004373129e-05, "loss": 0.1647, "step": 1038 }, { "epoch": 0.95, "learning_rate": 1.958719551561139e-05, "loss": 0.1732, "step": 1039 }, { "epoch": 0.95, "learning_rate": 1.958613966730648e-05, "loss": 0.162, "step": 1040 }, { "epoch": 0.95, "learning_rate": 1.958508249896196e-05, "loss": 0.1615, "step": 1041 }, { "epoch": 0.96, "learning_rate": 1.95840240107234e-05, "loss": 0.1617, "step": 1042 }, { "epoch": 0.96, "learning_rate": 1.9582964202736553e-05, "loss": 0.1566, "step": 1043 }, { "epoch": 0.96, "learning_rate": 1.9581903075147372e-05, "loss": 0.1636, "step": 1044 }, { "epoch": 0.96, "learning_rate": 1.958084062810196e-05, "loss": 0.1644, "step": 1045 }, { "epoch": 0.96, "learning_rate": 1.9579776861746634e-05, "loss": 0.1682, "step": 1046 }, { "epoch": 0.96, "learning_rate": 1.9578711776227867e-05, "loss": 0.1652, "step": 1047 }, { "epoch": 0.96, "learning_rate": 1.957764537169233e-05, "loss": 0.1589, "step": 1048 }, { "epoch": 0.96, "learning_rate": 1.957657764828687e-05, "loss": 0.1591, "step": 1049 }, { "epoch": 0.96, "learning_rate": 1.9575508606158513e-05, "loss": 0.1655, "step": 1050 }, { "epoch": 0.96, "learning_rate": 1.9574438245454472e-05, "loss": 0.1574, "step": 1051 }, { "epoch": 0.96, "learning_rate": 1.9573366566322137e-05, "loss": 0.1504, "step": 1052 }, { "epoch": 0.97, "learning_rate": 1.957229356890908e-05, "loss": 0.1559, "step": 1053 }, { "epoch": 0.97, "learning_rate": 1.9571219253363057e-05, "loss": 0.1689, "step": 1054 }, { "epoch": 0.97, "learning_rate": 1.957014361983201e-05, "loss": 0.1553, "step": 1055 }, { "epoch": 0.97, "learning_rate": 1.956906666846405e-05, "loss": 0.1563, "step": 1056 }, { "epoch": 0.97, "learning_rate": 1.9567988399407473e-05, "loss": 0.1784, "step": 1057 }, { "epoch": 0.97, "learning_rate": 1.9566908812810772e-05, "loss": 0.1661, "step": 1058 }, { "epoch": 0.97, "learning_rate": 1.95658279088226e-05, "loss": 0.1589, "step": 1059 }, { "epoch": 0.97, "learning_rate": 1.9564745687591804e-05, "loss": 0.1565, "step": 1060 }, { "epoch": 0.97, "learning_rate": 1.9563662149267405e-05, "loss": 0.165, "step": 1061 }, { "epoch": 0.97, "learning_rate": 1.9562577293998616e-05, "loss": 0.1654, "step": 1062 }, { "epoch": 0.97, "learning_rate": 1.956149112193482e-05, "loss": 0.1641, "step": 1063 }, { "epoch": 0.98, "learning_rate": 1.9560403633225587e-05, "loss": 0.16, "step": 1064 }, { "epoch": 0.98, "learning_rate": 1.955931482802067e-05, "loss": 0.1584, "step": 1065 }, { "epoch": 0.98, "learning_rate": 1.9558224706469996e-05, "loss": 0.1613, "step": 1066 }, { "epoch": 0.98, "learning_rate": 1.955713326872368e-05, "loss": 0.1517, "step": 1067 }, { "epoch": 0.98, "learning_rate": 1.9556040514932016e-05, "loss": 0.1624, "step": 1068 }, { "epoch": 0.98, "learning_rate": 1.9554946445245485e-05, "loss": 0.1631, "step": 1069 }, { "epoch": 0.98, "learning_rate": 1.955385105981473e-05, "loss": 0.1581, "step": 1070 }, { "epoch": 0.98, "learning_rate": 1.9552754358790605e-05, "loss": 0.1659, "step": 1071 }, { "epoch": 0.98, "learning_rate": 1.9551656342324118e-05, "loss": 0.1634, "step": 1072 }, { "epoch": 0.98, "learning_rate": 1.9550557010566473e-05, "loss": 0.1677, "step": 1073 }, { "epoch": 0.98, "learning_rate": 1.9549456363669052e-05, "loss": 0.1667, "step": 1074 }, { "epoch": 0.99, "learning_rate": 1.9548354401783416e-05, "loss": 0.163, "step": 1075 }, { "epoch": 0.99, "learning_rate": 1.9547251125061305e-05, "loss": 0.1609, "step": 1076 }, { "epoch": 0.99, "learning_rate": 1.954614653365465e-05, "loss": 0.1624, "step": 1077 }, { "epoch": 0.99, "learning_rate": 1.9545040627715554e-05, "loss": 0.1599, "step": 1078 }, { "epoch": 0.99, "learning_rate": 1.95439334073963e-05, "loss": 0.1599, "step": 1079 }, { "epoch": 0.99, "learning_rate": 1.954282487284936e-05, "loss": 0.1613, "step": 1080 }, { "epoch": 0.99, "learning_rate": 1.9541715024227382e-05, "loss": 0.1513, "step": 1081 }, { "epoch": 0.99, "learning_rate": 1.954060386168319e-05, "loss": 0.1611, "step": 1082 }, { "epoch": 0.99, "learning_rate": 1.9539491385369803e-05, "loss": 0.165, "step": 1083 }, { "epoch": 0.99, "learning_rate": 1.9538377595440404e-05, "loss": 0.1667, "step": 1084 }, { "epoch": 0.99, "learning_rate": 1.953726249204837e-05, "loss": 0.1525, "step": 1085 }, { "epoch": 1.0, "learning_rate": 1.9536146075347257e-05, "loss": 0.1584, "step": 1086 }, { "epoch": 1.0, "learning_rate": 1.9535028345490792e-05, "loss": 0.1547, "step": 1087 }, { "epoch": 1.0, "learning_rate": 1.953390930263289e-05, "loss": 0.1518, "step": 1088 }, { "epoch": 1.0, "learning_rate": 1.9532788946927653e-05, "loss": 0.1612, "step": 1089 }, { "epoch": 1.0, "learning_rate": 1.953166727852935e-05, "loss": 0.161, "step": 1090 }, { "epoch": 1.0, "learning_rate": 1.953054429759244e-05, "loss": 0.1694, "step": 1091 }, { "epoch": 1.0, "learning_rate": 1.9529420004271568e-05, "loss": 0.1501, "step": 1092 }, { "epoch": 1.0, "learning_rate": 1.9528294398721543e-05, "loss": 0.1617, "step": 1093 }, { "epoch": 1.0, "learning_rate": 1.9527167481097368e-05, "loss": 0.1612, "step": 1094 }, { "epoch": 1.0, "learning_rate": 1.9526039251554222e-05, "loss": 0.1574, "step": 1095 }, { "epoch": 1.0, "learning_rate": 1.9524909710247465e-05, "loss": 0.1603, "step": 1096 }, { "epoch": 1.01, "learning_rate": 1.9523778857332642e-05, "loss": 0.1572, "step": 1097 }, { "epoch": 1.01, "learning_rate": 1.952264669296547e-05, "loss": 0.1576, "step": 1098 }, { "epoch": 1.01, "learning_rate": 1.952151321730186e-05, "loss": 0.1598, "step": 1099 }, { "epoch": 1.01, "learning_rate": 1.9520378430497882e-05, "loss": 0.1504, "step": 1100 }, { "epoch": 1.01, "learning_rate": 1.9519242332709804e-05, "loss": 0.1665, "step": 1101 }, { "epoch": 1.01, "learning_rate": 1.9518104924094077e-05, "loss": 0.1589, "step": 1102 }, { "epoch": 1.01, "learning_rate": 1.951696620480732e-05, "loss": 0.1594, "step": 1103 }, { "epoch": 1.01, "learning_rate": 1.9515826175006333e-05, "loss": 0.1592, "step": 1104 }, { "epoch": 1.01, "learning_rate": 1.951468483484811e-05, "loss": 0.1532, "step": 1105 }, { "epoch": 1.01, "learning_rate": 1.9513542184489814e-05, "loss": 0.1683, "step": 1106 }, { "epoch": 1.01, "learning_rate": 1.951239822408879e-05, "loss": 0.153, "step": 1107 }, { "epoch": 1.02, "learning_rate": 1.9511252953802568e-05, "loss": 0.1671, "step": 1108 }, { "epoch": 1.02, "learning_rate": 1.951010637378885e-05, "loss": 0.1677, "step": 1109 }, { "epoch": 1.02, "learning_rate": 1.950895848420553e-05, "loss": 0.1571, "step": 1110 }, { "epoch": 1.02, "learning_rate": 1.9507809285210672e-05, "loss": 0.1739, "step": 1111 }, { "epoch": 1.02, "learning_rate": 1.9506658776962522e-05, "loss": 0.1626, "step": 1112 }, { "epoch": 1.02, "learning_rate": 1.9505506959619507e-05, "loss": 0.1673, "step": 1113 }, { "epoch": 1.02, "learning_rate": 1.9504353833340244e-05, "loss": 0.162, "step": 1114 }, { "epoch": 1.02, "learning_rate": 1.9503199398283515e-05, "loss": 0.1641, "step": 1115 }, { "epoch": 1.02, "learning_rate": 1.9502043654608293e-05, "loss": 0.166, "step": 1116 }, { "epoch": 1.02, "learning_rate": 1.9500886602473725e-05, "loss": 0.1587, "step": 1117 }, { "epoch": 1.02, "learning_rate": 1.949972824203914e-05, "loss": 0.1583, "step": 1118 }, { "epoch": 1.03, "learning_rate": 1.949856857346405e-05, "loss": 0.1592, "step": 1119 }, { "epoch": 1.03, "learning_rate": 1.949740759690814e-05, "loss": 0.1549, "step": 1120 }, { "epoch": 1.03, "learning_rate": 1.9496245312531286e-05, "loss": 0.1629, "step": 1121 }, { "epoch": 1.03, "learning_rate": 1.9495081720493534e-05, "loss": 0.1613, "step": 1122 }, { "epoch": 1.03, "learning_rate": 1.9493916820955114e-05, "loss": 0.1494, "step": 1123 }, { "epoch": 1.03, "learning_rate": 1.949275061407644e-05, "loss": 0.149, "step": 1124 }, { "epoch": 1.03, "learning_rate": 1.94915831000181e-05, "loss": 0.1528, "step": 1125 }, { "epoch": 1.03, "learning_rate": 1.949041427894086e-05, "loss": 0.1608, "step": 1126 }, { "epoch": 1.03, "learning_rate": 1.9489244151005673e-05, "loss": 0.1591, "step": 1127 }, { "epoch": 1.03, "learning_rate": 1.9488072716373674e-05, "loss": 0.1562, "step": 1128 }, { "epoch": 1.03, "learning_rate": 1.9486899975206165e-05, "loss": 0.1546, "step": 1129 }, { "epoch": 1.04, "learning_rate": 1.9485725927664643e-05, "loss": 0.1621, "step": 1130 }, { "epoch": 1.04, "learning_rate": 1.948455057391077e-05, "loss": 0.1612, "step": 1131 }, { "epoch": 1.04, "learning_rate": 1.9483373914106407e-05, "loss": 0.1644, "step": 1132 }, { "epoch": 1.04, "learning_rate": 1.948219594841357e-05, "loss": 0.158, "step": 1133 }, { "epoch": 1.04, "learning_rate": 1.9481016676994477e-05, "loss": 0.1629, "step": 1134 }, { "epoch": 1.04, "learning_rate": 1.9479836100011512e-05, "loss": 0.1623, "step": 1135 }, { "epoch": 1.04, "learning_rate": 1.947865421762725e-05, "loss": 0.164, "step": 1136 }, { "epoch": 1.04, "learning_rate": 1.947747103000444e-05, "loss": 0.1576, "step": 1137 }, { "epoch": 1.04, "learning_rate": 1.9476286537306002e-05, "loss": 0.1519, "step": 1138 }, { "epoch": 1.04, "learning_rate": 1.947510073969505e-05, "loss": 0.1624, "step": 1139 }, { "epoch": 1.04, "learning_rate": 1.9473913637334873e-05, "loss": 0.1598, "step": 1140 }, { "epoch": 1.05, "learning_rate": 1.9472725230388936e-05, "loss": 0.1597, "step": 1141 }, { "epoch": 1.05, "learning_rate": 1.947153551902089e-05, "loss": 0.1709, "step": 1142 }, { "epoch": 1.05, "learning_rate": 1.9470344503394553e-05, "loss": 0.1694, "step": 1143 }, { "epoch": 1.05, "learning_rate": 1.9469152183673936e-05, "loss": 0.1632, "step": 1144 }, { "epoch": 1.05, "learning_rate": 1.946795856002323e-05, "loss": 0.1627, "step": 1145 }, { "epoch": 1.05, "learning_rate": 1.946676363260679e-05, "loss": 0.1571, "step": 1146 }, { "epoch": 1.05, "learning_rate": 1.946556740158917e-05, "loss": 0.1547, "step": 1147 }, { "epoch": 1.05, "learning_rate": 1.9464369867135097e-05, "loss": 0.1586, "step": 1148 }, { "epoch": 1.05, "learning_rate": 1.9463171029409463e-05, "loss": 0.1573, "step": 1149 }, { "epoch": 1.05, "learning_rate": 1.946197088857736e-05, "loss": 0.161, "step": 1150 }, { "epoch": 1.05, "learning_rate": 1.9460769444804052e-05, "loss": 0.1603, "step": 1151 }, { "epoch": 1.06, "learning_rate": 1.9459566698254974e-05, "loss": 0.1569, "step": 1152 }, { "epoch": 1.06, "learning_rate": 1.9458362649095753e-05, "loss": 0.1556, "step": 1153 }, { "epoch": 1.06, "learning_rate": 1.9457157297492194e-05, "loss": 0.1585, "step": 1154 }, { "epoch": 1.06, "learning_rate": 1.9455950643610268e-05, "loss": 0.1596, "step": 1155 }, { "epoch": 1.06, "learning_rate": 1.945474268761614e-05, "loss": 0.1606, "step": 1156 }, { "epoch": 1.06, "learning_rate": 1.9453533429676148e-05, "loss": 0.1632, "step": 1157 }, { "epoch": 1.06, "learning_rate": 1.9452322869956813e-05, "loss": 0.1544, "step": 1158 }, { "epoch": 1.06, "learning_rate": 1.945111100862483e-05, "loss": 0.1536, "step": 1159 }, { "epoch": 1.06, "learning_rate": 1.9449897845847077e-05, "loss": 0.1544, "step": 1160 }, { "epoch": 1.06, "learning_rate": 1.944868338179061e-05, "loss": 0.1638, "step": 1161 }, { "epoch": 1.07, "learning_rate": 1.944746761662266e-05, "loss": 0.1531, "step": 1162 }, { "epoch": 1.07, "learning_rate": 1.944625055051065e-05, "loss": 0.1611, "step": 1163 }, { "epoch": 1.07, "learning_rate": 1.9445032183622166e-05, "loss": 0.158, "step": 1164 }, { "epoch": 1.07, "learning_rate": 1.9443812516124986e-05, "loss": 0.1623, "step": 1165 }, { "epoch": 1.07, "learning_rate": 1.9442591548187056e-05, "loss": 0.1684, "step": 1166 }, { "epoch": 1.07, "learning_rate": 1.944136927997651e-05, "loss": 0.1654, "step": 1167 }, { "epoch": 1.07, "learning_rate": 1.9440145711661663e-05, "loss": 0.157, "step": 1168 }, { "epoch": 1.07, "learning_rate": 1.9438920843410996e-05, "loss": 0.1539, "step": 1169 }, { "epoch": 1.07, "learning_rate": 1.943769467539318e-05, "loss": 0.1628, "step": 1170 }, { "epoch": 1.07, "learning_rate": 1.9436467207777062e-05, "loss": 0.1671, "step": 1171 }, { "epoch": 1.07, "learning_rate": 1.943523844073167e-05, "loss": 0.1627, "step": 1172 }, { "epoch": 1.08, "learning_rate": 1.9434008374426204e-05, "loss": 0.1611, "step": 1173 }, { "epoch": 1.08, "learning_rate": 1.9432777009030054e-05, "loss": 0.1535, "step": 1174 }, { "epoch": 1.08, "learning_rate": 1.9431544344712776e-05, "loss": 0.1615, "step": 1175 }, { "epoch": 1.08, "learning_rate": 1.9430310381644115e-05, "loss": 0.1523, "step": 1176 }, { "epoch": 1.08, "learning_rate": 1.9429075119993992e-05, "loss": 0.1575, "step": 1177 }, { "epoch": 1.08, "learning_rate": 1.9427838559932507e-05, "loss": 0.1573, "step": 1178 }, { "epoch": 1.08, "learning_rate": 1.942660070162993e-05, "loss": 0.1609, "step": 1179 }, { "epoch": 1.08, "learning_rate": 1.942536154525673e-05, "loss": 0.1565, "step": 1180 }, { "epoch": 1.08, "learning_rate": 1.9424121090983532e-05, "loss": 0.1551, "step": 1181 }, { "epoch": 1.08, "learning_rate": 1.9422879338981158e-05, "loss": 0.1598, "step": 1182 }, { "epoch": 1.08, "learning_rate": 1.9421636289420593e-05, "loss": 0.1658, "step": 1183 }, { "epoch": 1.09, "learning_rate": 1.9420391942473018e-05, "loss": 0.1608, "step": 1184 }, { "epoch": 1.09, "learning_rate": 1.9419146298309775e-05, "loss": 0.1557, "step": 1185 }, { "epoch": 1.09, "learning_rate": 1.9417899357102395e-05, "loss": 0.149, "step": 1186 }, { "epoch": 1.09, "learning_rate": 1.941665111902259e-05, "loss": 0.1514, "step": 1187 }, { "epoch": 1.09, "learning_rate": 1.941540158424224e-05, "loss": 0.1647, "step": 1188 }, { "epoch": 1.09, "learning_rate": 1.9414150752933416e-05, "loss": 0.1591, "step": 1189 }, { "epoch": 1.09, "learning_rate": 1.9412898625268354e-05, "loss": 0.1433, "step": 1190 }, { "epoch": 1.09, "learning_rate": 1.9411645201419477e-05, "loss": 0.157, "step": 1191 }, { "epoch": 1.09, "learning_rate": 1.941039048155939e-05, "loss": 0.1573, "step": 1192 }, { "epoch": 1.09, "learning_rate": 1.940913446586087e-05, "loss": 0.1501, "step": 1193 }, { "epoch": 1.09, "learning_rate": 1.940787715449687e-05, "loss": 0.1644, "step": 1194 }, { "epoch": 1.1, "learning_rate": 1.940661854764053e-05, "loss": 0.1671, "step": 1195 }, { "epoch": 1.1, "learning_rate": 1.940535864546516e-05, "loss": 0.1635, "step": 1196 }, { "epoch": 1.1, "learning_rate": 1.9404097448144256e-05, "loss": 0.1676, "step": 1197 }, { "epoch": 1.1, "learning_rate": 1.940283495585149e-05, "loss": 0.1551, "step": 1198 }, { "epoch": 1.1, "learning_rate": 1.9401571168760705e-05, "loss": 0.1612, "step": 1199 }, { "epoch": 1.1, "learning_rate": 1.940030608704593e-05, "loss": 0.1621, "step": 1200 }, { "epoch": 1.1, "learning_rate": 1.9399039710881375e-05, "loss": 0.1589, "step": 1201 }, { "epoch": 1.1, "learning_rate": 1.9397772040441415e-05, "loss": 0.1511, "step": 1202 }, { "epoch": 1.1, "learning_rate": 1.939650307590062e-05, "loss": 0.1594, "step": 1203 }, { "epoch": 1.1, "learning_rate": 1.9395232817433728e-05, "loss": 0.1584, "step": 1204 }, { "epoch": 1.1, "learning_rate": 1.9393961265215656e-05, "loss": 0.1715, "step": 1205 }, { "epoch": 1.11, "learning_rate": 1.93926884194215e-05, "loss": 0.1488, "step": 1206 }, { "epoch": 1.11, "learning_rate": 1.939141428022654e-05, "loss": 0.1662, "step": 1207 }, { "epoch": 1.11, "learning_rate": 1.939013884780622e-05, "loss": 0.1551, "step": 1208 }, { "epoch": 1.11, "learning_rate": 1.938886212233618e-05, "loss": 0.1584, "step": 1209 }, { "epoch": 1.11, "learning_rate": 1.9387584103992217e-05, "loss": 0.1585, "step": 1210 }, { "epoch": 1.11, "learning_rate": 1.938630479295033e-05, "loss": 0.1579, "step": 1211 }, { "epoch": 1.11, "learning_rate": 1.9385024189386678e-05, "loss": 0.1586, "step": 1212 }, { "epoch": 1.11, "learning_rate": 1.9383742293477604e-05, "loss": 0.1667, "step": 1213 }, { "epoch": 1.11, "learning_rate": 1.9382459105399634e-05, "loss": 0.1617, "step": 1214 }, { "epoch": 1.11, "learning_rate": 1.9381174625329455e-05, "loss": 0.1611, "step": 1215 }, { "epoch": 1.11, "learning_rate": 1.9379888853443954e-05, "loss": 0.1618, "step": 1216 }, { "epoch": 1.12, "learning_rate": 1.9378601789920186e-05, "loss": 0.1571, "step": 1217 }, { "epoch": 1.12, "learning_rate": 1.9377313434935376e-05, "loss": 0.1588, "step": 1218 }, { "epoch": 1.12, "learning_rate": 1.937602378866694e-05, "loss": 0.1605, "step": 1219 }, { "epoch": 1.12, "learning_rate": 1.9374732851292466e-05, "loss": 0.168, "step": 1220 }, { "epoch": 1.12, "learning_rate": 1.9373440622989717e-05, "loss": 0.1564, "step": 1221 }, { "epoch": 1.12, "learning_rate": 1.9372147103936637e-05, "loss": 0.1607, "step": 1222 }, { "epoch": 1.12, "learning_rate": 1.937085229431135e-05, "loss": 0.167, "step": 1223 }, { "epoch": 1.12, "learning_rate": 1.936955619429215e-05, "loss": 0.1507, "step": 1224 }, { "epoch": 1.12, "learning_rate": 1.936825880405752e-05, "loss": 0.1661, "step": 1225 }, { "epoch": 1.12, "learning_rate": 1.9366960123786113e-05, "loss": 0.1516, "step": 1226 }, { "epoch": 1.12, "learning_rate": 1.9365660153656754e-05, "loss": 0.1546, "step": 1227 }, { "epoch": 1.13, "learning_rate": 1.9364358893848464e-05, "loss": 0.1591, "step": 1228 }, { "epoch": 1.13, "learning_rate": 1.9363056344540423e-05, "loss": 0.1558, "step": 1229 }, { "epoch": 1.13, "learning_rate": 1.9361752505911997e-05, "loss": 0.1653, "step": 1230 }, { "epoch": 1.13, "learning_rate": 1.936044737814273e-05, "loss": 0.1619, "step": 1231 }, { "epoch": 1.13, "learning_rate": 1.9359140961412338e-05, "loss": 0.152, "step": 1232 }, { "epoch": 1.13, "learning_rate": 1.9357833255900725e-05, "loss": 0.1656, "step": 1233 }, { "epoch": 1.13, "learning_rate": 1.9356524261787962e-05, "loss": 0.1533, "step": 1234 }, { "epoch": 1.13, "learning_rate": 1.9355213979254302e-05, "loss": 0.1591, "step": 1235 }, { "epoch": 1.13, "learning_rate": 1.9353902408480172e-05, "loss": 0.1529, "step": 1236 }, { "epoch": 1.13, "learning_rate": 1.9352589549646183e-05, "loss": 0.1551, "step": 1237 }, { "epoch": 1.13, "learning_rate": 1.9351275402933116e-05, "loss": 0.1632, "step": 1238 }, { "epoch": 1.14, "learning_rate": 1.9349959968521935e-05, "loss": 0.1622, "step": 1239 }, { "epoch": 1.14, "learning_rate": 1.9348643246593782e-05, "loss": 0.1594, "step": 1240 }, { "epoch": 1.14, "learning_rate": 1.934732523732997e-05, "loss": 0.1603, "step": 1241 }, { "epoch": 1.14, "learning_rate": 1.9346005940911992e-05, "loss": 0.1595, "step": 1242 }, { "epoch": 1.14, "learning_rate": 1.934468535752152e-05, "loss": 0.1621, "step": 1243 }, { "epoch": 1.14, "learning_rate": 1.9343363487340403e-05, "loss": 0.1513, "step": 1244 }, { "epoch": 1.14, "learning_rate": 1.9342040330550665e-05, "loss": 0.1428, "step": 1245 }, { "epoch": 1.14, "learning_rate": 1.934071588733451e-05, "loss": 0.1666, "step": 1246 }, { "epoch": 1.14, "learning_rate": 1.933939015787432e-05, "loss": 0.1515, "step": 1247 }, { "epoch": 1.14, "learning_rate": 1.9338063142352644e-05, "loss": 0.1567, "step": 1248 }, { "epoch": 1.14, "learning_rate": 1.9336734840952223e-05, "loss": 0.1646, "step": 1249 }, { "epoch": 1.15, "learning_rate": 1.933540525385597e-05, "loss": 0.1591, "step": 1250 }, { "epoch": 1.15, "learning_rate": 1.9334074381246965e-05, "loss": 0.1552, "step": 1251 }, { "epoch": 1.15, "learning_rate": 1.9332742223308477e-05, "loss": 0.1655, "step": 1252 }, { "epoch": 1.15, "learning_rate": 1.9331408780223953e-05, "loss": 0.1539, "step": 1253 }, { "epoch": 1.15, "learning_rate": 1.9330074052177e-05, "loss": 0.1494, "step": 1254 }, { "epoch": 1.15, "learning_rate": 1.9328738039351427e-05, "loss": 0.1663, "step": 1255 }, { "epoch": 1.15, "learning_rate": 1.9327400741931196e-05, "loss": 0.1554, "step": 1256 }, { "epoch": 1.15, "learning_rate": 1.9326062160100466e-05, "loss": 0.1547, "step": 1257 }, { "epoch": 1.15, "learning_rate": 1.932472229404356e-05, "loss": 0.1645, "step": 1258 }, { "epoch": 1.15, "learning_rate": 1.932338114394498e-05, "loss": 0.1582, "step": 1259 }, { "epoch": 1.15, "learning_rate": 1.9322038709989406e-05, "loss": 0.1591, "step": 1260 }, { "epoch": 1.16, "learning_rate": 1.9320694992361697e-05, "loss": 0.1599, "step": 1261 }, { "epoch": 1.16, "learning_rate": 1.9319349991246887e-05, "loss": 0.158, "step": 1262 }, { "epoch": 1.16, "learning_rate": 1.9318003706830186e-05, "loss": 0.1573, "step": 1263 }, { "epoch": 1.16, "learning_rate": 1.931665613929698e-05, "loss": 0.1499, "step": 1264 }, { "epoch": 1.16, "learning_rate": 1.9315307288832835e-05, "loss": 0.1639, "step": 1265 }, { "epoch": 1.16, "learning_rate": 1.931395715562349e-05, "loss": 0.158, "step": 1266 }, { "epoch": 1.16, "learning_rate": 1.9312605739854862e-05, "loss": 0.1657, "step": 1267 }, { "epoch": 1.16, "learning_rate": 1.9311253041713053e-05, "loss": 0.1569, "step": 1268 }, { "epoch": 1.16, "learning_rate": 1.9309899061384322e-05, "loss": 0.1571, "step": 1269 }, { "epoch": 1.16, "learning_rate": 1.9308543799055118e-05, "loss": 0.1582, "step": 1270 }, { "epoch": 1.16, "learning_rate": 1.930718725491207e-05, "loss": 0.1586, "step": 1271 }, { "epoch": 1.17, "learning_rate": 1.930582942914198e-05, "loss": 0.1652, "step": 1272 }, { "epoch": 1.17, "learning_rate": 1.930447032193181e-05, "loss": 0.1589, "step": 1273 }, { "epoch": 1.17, "learning_rate": 1.9303109933468732e-05, "loss": 0.1575, "step": 1274 }, { "epoch": 1.17, "learning_rate": 1.9301748263940064e-05, "loss": 0.1622, "step": 1275 }, { "epoch": 1.17, "learning_rate": 1.9300385313533313e-05, "loss": 0.145, "step": 1276 }, { "epoch": 1.17, "learning_rate": 1.9299021082436164e-05, "loss": 0.1515, "step": 1277 }, { "epoch": 1.17, "learning_rate": 1.9297655570836476e-05, "loss": 0.1626, "step": 1278 }, { "epoch": 1.17, "learning_rate": 1.929628877892228e-05, "loss": 0.1623, "step": 1279 }, { "epoch": 1.17, "learning_rate": 1.9294920706881787e-05, "loss": 0.1567, "step": 1280 }, { "epoch": 1.17, "learning_rate": 1.9293551354903393e-05, "loss": 0.1518, "step": 1281 }, { "epoch": 1.18, "learning_rate": 1.9292180723175656e-05, "loss": 0.1615, "step": 1282 }, { "epoch": 1.18, "learning_rate": 1.929080881188731e-05, "loss": 0.149, "step": 1283 }, { "epoch": 1.18, "learning_rate": 1.9289435621227283e-05, "loss": 0.1633, "step": 1284 }, { "epoch": 1.18, "learning_rate": 1.9288061151384658e-05, "loss": 0.157, "step": 1285 }, { "epoch": 1.18, "learning_rate": 1.9286685402548707e-05, "loss": 0.1593, "step": 1286 }, { "epoch": 1.18, "learning_rate": 1.9285308374908876e-05, "loss": 0.1569, "step": 1287 }, { "epoch": 1.18, "learning_rate": 1.9283930068654787e-05, "loss": 0.1673, "step": 1288 }, { "epoch": 1.18, "learning_rate": 1.9282550483976236e-05, "loss": 0.1619, "step": 1289 }, { "epoch": 1.18, "learning_rate": 1.9281169621063188e-05, "loss": 0.1589, "step": 1290 }, { "epoch": 1.18, "learning_rate": 1.9279787480105802e-05, "loss": 0.1499, "step": 1291 }, { "epoch": 1.18, "learning_rate": 1.9278404061294397e-05, "loss": 0.1514, "step": 1292 }, { "epoch": 1.19, "learning_rate": 1.9277019364819476e-05, "loss": 0.166, "step": 1293 }, { "epoch": 1.19, "learning_rate": 1.9275633390871718e-05, "loss": 0.1704, "step": 1294 }, { "epoch": 1.19, "learning_rate": 1.9274246139641972e-05, "loss": 0.1671, "step": 1295 }, { "epoch": 1.19, "learning_rate": 1.9272857611321268e-05, "loss": 0.1584, "step": 1296 }, { "epoch": 1.19, "learning_rate": 1.927146780610081e-05, "loss": 0.1536, "step": 1297 }, { "epoch": 1.19, "learning_rate": 1.927007672417198e-05, "loss": 0.1691, "step": 1298 }, { "epoch": 1.19, "learning_rate": 1.9268684365726326e-05, "loss": 0.1532, "step": 1299 }, { "epoch": 1.19, "learning_rate": 1.9267290730955593e-05, "loss": 0.1573, "step": 1300 }, { "epoch": 1.19, "learning_rate": 1.9265895820051683e-05, "loss": 0.1546, "step": 1301 }, { "epoch": 1.19, "learning_rate": 1.9264499633206675e-05, "loss": 0.1621, "step": 1302 }, { "epoch": 1.19, "learning_rate": 1.9263102170612833e-05, "loss": 0.1611, "step": 1303 }, { "epoch": 1.2, "learning_rate": 1.9261703432462592e-05, "loss": 0.1554, "step": 1304 }, { "epoch": 1.2, "learning_rate": 1.9260303418948554e-05, "loss": 0.1575, "step": 1305 }, { "epoch": 1.2, "learning_rate": 1.9258902130263518e-05, "loss": 0.1556, "step": 1306 }, { "epoch": 1.2, "learning_rate": 1.925749956660044e-05, "loss": 0.162, "step": 1307 }, { "epoch": 1.2, "learning_rate": 1.9256095728152452e-05, "loss": 0.1545, "step": 1308 }, { "epoch": 1.2, "learning_rate": 1.925469061511287e-05, "loss": 0.1531, "step": 1309 }, { "epoch": 1.2, "learning_rate": 1.925328422767519e-05, "loss": 0.1493, "step": 1310 }, { "epoch": 1.2, "learning_rate": 1.9251876566033062e-05, "loss": 0.1676, "step": 1311 }, { "epoch": 1.2, "learning_rate": 1.9250467630380335e-05, "loss": 0.1539, "step": 1312 }, { "epoch": 1.2, "learning_rate": 1.9249057420911018e-05, "loss": 0.1495, "step": 1313 }, { "epoch": 1.2, "learning_rate": 1.9247645937819306e-05, "loss": 0.1636, "step": 1314 }, { "epoch": 1.21, "learning_rate": 1.924623318129956e-05, "loss": 0.1556, "step": 1315 }, { "epoch": 1.21, "learning_rate": 1.9244819151546325e-05, "loss": 0.1583, "step": 1316 }, { "epoch": 1.21, "learning_rate": 1.9243403848754312e-05, "loss": 0.1569, "step": 1317 }, { "epoch": 1.21, "learning_rate": 1.9241987273118416e-05, "loss": 0.1644, "step": 1318 }, { "epoch": 1.21, "learning_rate": 1.9240569424833703e-05, "loss": 0.1594, "step": 1319 }, { "epoch": 1.21, "learning_rate": 1.9239150304095416e-05, "loss": 0.1565, "step": 1320 }, { "epoch": 1.21, "learning_rate": 1.9237729911098972e-05, "loss": 0.1523, "step": 1321 }, { "epoch": 1.21, "learning_rate": 1.9236308246039957e-05, "loss": 0.1511, "step": 1322 }, { "epoch": 1.21, "learning_rate": 1.9234885309114144e-05, "loss": 0.151, "step": 1323 }, { "epoch": 1.21, "learning_rate": 1.923346110051748e-05, "loss": 0.1581, "step": 1324 }, { "epoch": 1.21, "learning_rate": 1.923203562044607e-05, "loss": 0.1505, "step": 1325 }, { "epoch": 1.22, "learning_rate": 1.9230608869096218e-05, "loss": 0.1581, "step": 1326 }, { "epoch": 1.22, "learning_rate": 1.922918084666439e-05, "loss": 0.157, "step": 1327 }, { "epoch": 1.22, "learning_rate": 1.9227751553347226e-05, "loss": 0.1563, "step": 1328 }, { "epoch": 1.22, "learning_rate": 1.9226320989341547e-05, "loss": 0.1516, "step": 1329 }, { "epoch": 1.22, "learning_rate": 1.922488915484434e-05, "loss": 0.16, "step": 1330 }, { "epoch": 1.22, "learning_rate": 1.922345605005278e-05, "loss": 0.1592, "step": 1331 }, { "epoch": 1.22, "learning_rate": 1.922202167516421e-05, "loss": 0.146, "step": 1332 }, { "epoch": 1.22, "learning_rate": 1.9220586030376135e-05, "loss": 0.1489, "step": 1333 }, { "epoch": 1.22, "learning_rate": 1.9219149115886265e-05, "loss": 0.1468, "step": 1334 }, { "epoch": 1.22, "learning_rate": 1.9217710931892453e-05, "loss": 0.1512, "step": 1335 }, { "epoch": 1.22, "learning_rate": 1.9216271478592753e-05, "loss": 0.1631, "step": 1336 }, { "epoch": 1.23, "learning_rate": 1.9214830756185374e-05, "loss": 0.155, "step": 1337 }, { "epoch": 1.23, "learning_rate": 1.9213388764868707e-05, "loss": 0.1578, "step": 1338 }, { "epoch": 1.23, "learning_rate": 1.9211945504841324e-05, "loss": 0.1564, "step": 1339 }, { "epoch": 1.23, "learning_rate": 1.921050097630196e-05, "loss": 0.1569, "step": 1340 }, { "epoch": 1.23, "learning_rate": 1.920905517944954e-05, "loss": 0.1487, "step": 1341 }, { "epoch": 1.23, "learning_rate": 1.9207608114483144e-05, "loss": 0.1596, "step": 1342 }, { "epoch": 1.23, "learning_rate": 1.9206159781602044e-05, "loss": 0.1552, "step": 1343 }, { "epoch": 1.23, "learning_rate": 1.9204710181005676e-05, "loss": 0.1594, "step": 1344 }, { "epoch": 1.23, "learning_rate": 1.920325931289366e-05, "loss": 0.1626, "step": 1345 }, { "epoch": 1.23, "learning_rate": 1.9201807177465775e-05, "loss": 0.1501, "step": 1346 }, { "epoch": 1.23, "learning_rate": 1.9200353774921995e-05, "loss": 0.1511, "step": 1347 }, { "epoch": 1.24, "learning_rate": 1.9198899105462447e-05, "loss": 0.1528, "step": 1348 }, { "epoch": 1.24, "learning_rate": 1.9197443169287454e-05, "loss": 0.1565, "step": 1349 }, { "epoch": 1.24, "learning_rate": 1.9195985966597495e-05, "loss": 0.1499, "step": 1350 }, { "epoch": 1.24, "learning_rate": 1.9194527497593233e-05, "loss": 0.1599, "step": 1351 }, { "epoch": 1.24, "learning_rate": 1.919306776247551e-05, "loss": 0.1614, "step": 1352 }, { "epoch": 1.24, "learning_rate": 1.9191606761445324e-05, "loss": 0.1603, "step": 1353 }, { "epoch": 1.24, "learning_rate": 1.9190144494703865e-05, "loss": 0.1552, "step": 1354 }, { "epoch": 1.24, "learning_rate": 1.9188680962452496e-05, "loss": 0.1686, "step": 1355 }, { "epoch": 1.24, "learning_rate": 1.9187216164892743e-05, "loss": 0.1581, "step": 1356 }, { "epoch": 1.24, "learning_rate": 1.9185750102226318e-05, "loss": 0.1565, "step": 1357 }, { "epoch": 1.24, "learning_rate": 1.9184282774655096e-05, "loss": 0.1525, "step": 1358 }, { "epoch": 1.25, "learning_rate": 1.9182814182381136e-05, "loss": 0.1581, "step": 1359 }, { "epoch": 1.25, "learning_rate": 1.9181344325606667e-05, "loss": 0.1593, "step": 1360 }, { "epoch": 1.25, "learning_rate": 1.9179873204534095e-05, "loss": 0.1545, "step": 1361 }, { "epoch": 1.25, "learning_rate": 1.9178400819365994e-05, "loss": 0.1559, "step": 1362 }, { "epoch": 1.25, "learning_rate": 1.917692717030512e-05, "loss": 0.151, "step": 1363 }, { "epoch": 1.25, "learning_rate": 1.9175452257554393e-05, "loss": 0.1572, "step": 1364 }, { "epoch": 1.25, "learning_rate": 1.9173976081316915e-05, "loss": 0.1561, "step": 1365 }, { "epoch": 1.25, "learning_rate": 1.9172498641795964e-05, "loss": 0.1657, "step": 1366 }, { "epoch": 1.25, "learning_rate": 1.9171019939194982e-05, "loss": 0.1572, "step": 1367 }, { "epoch": 1.25, "learning_rate": 1.9169539973717594e-05, "loss": 0.1679, "step": 1368 }, { "epoch": 1.25, "learning_rate": 1.916805874556759e-05, "loss": 0.1598, "step": 1369 }, { "epoch": 1.26, "learning_rate": 1.9166576254948953e-05, "loss": 0.1545, "step": 1370 }, { "epoch": 1.26, "learning_rate": 1.9165092502065815e-05, "loss": 0.1469, "step": 1371 }, { "epoch": 1.26, "learning_rate": 1.916360748712249e-05, "loss": 0.1578, "step": 1372 }, { "epoch": 1.26, "learning_rate": 1.916212121032348e-05, "loss": 0.156, "step": 1373 }, { "epoch": 1.26, "learning_rate": 1.9160633671873447e-05, "loss": 0.1604, "step": 1374 }, { "epoch": 1.26, "learning_rate": 1.915914487197722e-05, "loss": 0.1598, "step": 1375 }, { "epoch": 1.26, "learning_rate": 1.915765481083982e-05, "loss": 0.1513, "step": 1376 }, { "epoch": 1.26, "learning_rate": 1.9156163488666433e-05, "loss": 0.1619, "step": 1377 }, { "epoch": 1.26, "learning_rate": 1.9154670905662412e-05, "loss": 0.1566, "step": 1378 }, { "epoch": 1.26, "learning_rate": 1.9153177062033297e-05, "loss": 0.1682, "step": 1379 }, { "epoch": 1.26, "learning_rate": 1.9151681957984793e-05, "loss": 0.1547, "step": 1380 }, { "epoch": 1.27, "learning_rate": 1.9150185593722778e-05, "loss": 0.1567, "step": 1381 }, { "epoch": 1.27, "learning_rate": 1.9148687969453304e-05, "loss": 0.1564, "step": 1382 }, { "epoch": 1.27, "learning_rate": 1.9147189085382603e-05, "loss": 0.1508, "step": 1383 }, { "epoch": 1.27, "learning_rate": 1.9145688941717074e-05, "loss": 0.1541, "step": 1384 }, { "epoch": 1.27, "learning_rate": 1.914418753866329e-05, "loss": 0.1602, "step": 1385 }, { "epoch": 1.27, "learning_rate": 1.9142684876428e-05, "loss": 0.1461, "step": 1386 }, { "epoch": 1.27, "learning_rate": 1.9141180955218127e-05, "loss": 0.1626, "step": 1387 }, { "epoch": 1.27, "learning_rate": 1.9139675775240763e-05, "loss": 0.1586, "step": 1388 }, { "epoch": 1.27, "learning_rate": 1.9138169336703173e-05, "loss": 0.162, "step": 1389 }, { "epoch": 1.27, "learning_rate": 1.91366616398128e-05, "loss": 0.1479, "step": 1390 }, { "epoch": 1.27, "learning_rate": 1.9135152684777262e-05, "loss": 0.154, "step": 1391 }, { "epoch": 1.28, "learning_rate": 1.9133642471804342e-05, "loss": 0.1574, "step": 1392 }, { "epoch": 1.28, "learning_rate": 1.9132131001102e-05, "loss": 0.1466, "step": 1393 }, { "epoch": 1.28, "learning_rate": 1.9130618272878374e-05, "loss": 0.1608, "step": 1394 }, { "epoch": 1.28, "learning_rate": 1.9129104287341772e-05, "loss": 0.1595, "step": 1395 }, { "epoch": 1.28, "learning_rate": 1.912758904470067e-05, "loss": 0.1613, "step": 1396 }, { "epoch": 1.28, "learning_rate": 1.912607254516372e-05, "loss": 0.1538, "step": 1397 }, { "epoch": 1.28, "learning_rate": 1.912455478893976e-05, "loss": 0.1602, "step": 1398 }, { "epoch": 1.28, "learning_rate": 1.912303577623777e-05, "loss": 0.1523, "step": 1399 }, { "epoch": 1.28, "learning_rate": 1.912151550726694e-05, "loss": 0.1571, "step": 1400 }, { "epoch": 1.28, "learning_rate": 1.9119993982236608e-05, "loss": 0.1512, "step": 1401 }, { "epoch": 1.29, "learning_rate": 1.9118471201356293e-05, "loss": 0.1523, "step": 1402 }, { "epoch": 1.29, "learning_rate": 1.9116947164835686e-05, "loss": 0.1725, "step": 1403 }, { "epoch": 1.29, "learning_rate": 1.911542187288465e-05, "loss": 0.1614, "step": 1404 }, { "epoch": 1.29, "learning_rate": 1.911389532571323e-05, "loss": 0.1636, "step": 1405 }, { "epoch": 1.29, "learning_rate": 1.9112367523531627e-05, "loss": 0.16, "step": 1406 }, { "epoch": 1.29, "learning_rate": 1.9110838466550227e-05, "loss": 0.1698, "step": 1407 }, { "epoch": 1.29, "learning_rate": 1.9109308154979588e-05, "loss": 0.1586, "step": 1408 }, { "epoch": 1.29, "learning_rate": 1.9107776589030436e-05, "loss": 0.1474, "step": 1409 }, { "epoch": 1.29, "learning_rate": 1.910624376891367e-05, "loss": 0.1707, "step": 1410 }, { "epoch": 1.29, "learning_rate": 1.910470969484037e-05, "loss": 0.1636, "step": 1411 }, { "epoch": 1.29, "learning_rate": 1.9103174367021777e-05, "loss": 0.1518, "step": 1412 }, { "epoch": 1.3, "learning_rate": 1.9101637785669308e-05, "loss": 0.1559, "step": 1413 }, { "epoch": 1.3, "learning_rate": 1.9100099950994564e-05, "loss": 0.1678, "step": 1414 }, { "epoch": 1.3, "learning_rate": 1.9098560863209302e-05, "loss": 0.1674, "step": 1415 }, { "epoch": 1.3, "learning_rate": 1.9097020522525462e-05, "loss": 0.1569, "step": 1416 }, { "epoch": 1.3, "learning_rate": 1.909547892915515e-05, "loss": 0.1529, "step": 1417 }, { "epoch": 1.3, "learning_rate": 1.9093936083310653e-05, "loss": 0.1573, "step": 1418 }, { "epoch": 1.3, "learning_rate": 1.909239198520442e-05, "loss": 0.1581, "step": 1419 }, { "epoch": 1.3, "learning_rate": 1.9090846635049084e-05, "loss": 0.1586, "step": 1420 }, { "epoch": 1.3, "learning_rate": 1.908930003305744e-05, "loss": 0.1644, "step": 1421 }, { "epoch": 1.3, "learning_rate": 1.908775217944246e-05, "loss": 0.1614, "step": 1422 }, { "epoch": 1.3, "learning_rate": 1.9086203074417288e-05, "loss": 0.1616, "step": 1423 }, { "epoch": 1.31, "learning_rate": 1.9084652718195237e-05, "loss": 0.1603, "step": 1424 }, { "epoch": 1.31, "learning_rate": 1.9083101110989807e-05, "loss": 0.1545, "step": 1425 }, { "epoch": 1.31, "learning_rate": 1.9081548253014645e-05, "loss": 0.1567, "step": 1426 }, { "epoch": 1.31, "learning_rate": 1.907999414448359e-05, "loss": 0.1609, "step": 1427 }, { "epoch": 1.31, "learning_rate": 1.9078438785610652e-05, "loss": 0.1631, "step": 1428 }, { "epoch": 1.31, "learning_rate": 1.9076882176610003e-05, "loss": 0.152, "step": 1429 }, { "epoch": 1.31, "learning_rate": 1.9075324317695996e-05, "loss": 0.1575, "step": 1430 }, { "epoch": 1.31, "learning_rate": 1.9073765209083145e-05, "loss": 0.1603, "step": 1431 }, { "epoch": 1.31, "learning_rate": 1.9072204850986153e-05, "loss": 0.1587, "step": 1432 }, { "epoch": 1.31, "learning_rate": 1.907064324361988e-05, "loss": 0.1632, "step": 1433 }, { "epoch": 1.31, "learning_rate": 1.906908038719937e-05, "loss": 0.1558, "step": 1434 }, { "epoch": 1.32, "learning_rate": 1.9067516281939826e-05, "loss": 0.1577, "step": 1435 }, { "epoch": 1.32, "learning_rate": 1.9065950928056636e-05, "loss": 0.1494, "step": 1436 }, { "epoch": 1.32, "learning_rate": 1.9064384325765347e-05, "loss": 0.16, "step": 1437 }, { "epoch": 1.32, "learning_rate": 1.9062816475281693e-05, "loss": 0.1493, "step": 1438 }, { "epoch": 1.32, "learning_rate": 1.9061247376821565e-05, "loss": 0.1595, "step": 1439 }, { "epoch": 1.32, "learning_rate": 1.9059677030601037e-05, "loss": 0.1616, "step": 1440 }, { "epoch": 1.32, "learning_rate": 1.9058105436836345e-05, "loss": 0.1605, "step": 1441 }, { "epoch": 1.32, "learning_rate": 1.9056532595743903e-05, "loss": 0.1588, "step": 1442 }, { "epoch": 1.32, "learning_rate": 1.9054958507540304e-05, "loss": 0.1615, "step": 1443 }, { "epoch": 1.32, "learning_rate": 1.905338317244229e-05, "loss": 0.1528, "step": 1444 }, { "epoch": 1.32, "learning_rate": 1.9051806590666807e-05, "loss": 0.1534, "step": 1445 }, { "epoch": 1.33, "learning_rate": 1.905022876243094e-05, "loss": 0.1603, "step": 1446 }, { "epoch": 1.33, "learning_rate": 1.904864968795197e-05, "loss": 0.1608, "step": 1447 }, { "epoch": 1.33, "learning_rate": 1.9047069367447336e-05, "loss": 0.1633, "step": 1448 }, { "epoch": 1.33, "learning_rate": 1.904548780113465e-05, "loss": 0.1566, "step": 1449 }, { "epoch": 1.33, "learning_rate": 1.9043904989231703e-05, "loss": 0.1669, "step": 1450 }, { "epoch": 1.33, "learning_rate": 1.904232093195645e-05, "loss": 0.1635, "step": 1451 }, { "epoch": 1.33, "learning_rate": 1.9040735629527027e-05, "loss": 0.1666, "step": 1452 }, { "epoch": 1.33, "learning_rate": 1.9039149082161724e-05, "loss": 0.1547, "step": 1453 }, { "epoch": 1.33, "learning_rate": 1.9037561290079022e-05, "loss": 0.1604, "step": 1454 }, { "epoch": 1.33, "learning_rate": 1.903597225349756e-05, "loss": 0.1547, "step": 1455 }, { "epoch": 1.33, "learning_rate": 1.903438197263616e-05, "loss": 0.1589, "step": 1456 }, { "epoch": 1.34, "learning_rate": 1.9032790447713797e-05, "loss": 0.1547, "step": 1457 }, { "epoch": 1.34, "learning_rate": 1.9031197678949633e-05, "loss": 0.1542, "step": 1458 }, { "epoch": 1.34, "learning_rate": 1.9029603666563003e-05, "loss": 0.1532, "step": 1459 }, { "epoch": 1.34, "learning_rate": 1.9028008410773398e-05, "loss": 0.1527, "step": 1460 }, { "epoch": 1.34, "learning_rate": 1.9026411911800496e-05, "loss": 0.1468, "step": 1461 }, { "epoch": 1.34, "learning_rate": 1.902481416986414e-05, "loss": 0.1527, "step": 1462 }, { "epoch": 1.34, "learning_rate": 1.9023215185184342e-05, "loss": 0.1657, "step": 1463 }, { "epoch": 1.34, "learning_rate": 1.9021614957981283e-05, "loss": 0.1503, "step": 1464 }, { "epoch": 1.34, "learning_rate": 1.902001348847532e-05, "loss": 0.1568, "step": 1465 }, { "epoch": 1.34, "learning_rate": 1.901841077688699e-05, "loss": 0.1444, "step": 1466 }, { "epoch": 1.34, "learning_rate": 1.9016806823436978e-05, "loss": 0.1571, "step": 1467 }, { "epoch": 1.35, "learning_rate": 1.901520162834616e-05, "loss": 0.1626, "step": 1468 }, { "epoch": 1.35, "learning_rate": 1.9013595191835575e-05, "loss": 0.1537, "step": 1469 }, { "epoch": 1.35, "learning_rate": 1.9011987514126432e-05, "loss": 0.1566, "step": 1470 }, { "epoch": 1.35, "learning_rate": 1.9010378595440116e-05, "loss": 0.163, "step": 1471 }, { "epoch": 1.35, "learning_rate": 1.900876843599818e-05, "loss": 0.1558, "step": 1472 }, { "epoch": 1.35, "learning_rate": 1.9007157036022345e-05, "loss": 0.1531, "step": 1473 }, { "epoch": 1.35, "learning_rate": 1.9005544395734504e-05, "loss": 0.1635, "step": 1474 }, { "epoch": 1.35, "learning_rate": 1.9003930515356727e-05, "loss": 0.1485, "step": 1475 }, { "epoch": 1.35, "learning_rate": 1.900231539511125e-05, "loss": 0.1589, "step": 1476 }, { "epoch": 1.35, "learning_rate": 1.9000699035220473e-05, "loss": 0.1522, "step": 1477 }, { "epoch": 1.35, "learning_rate": 1.8999081435906984e-05, "loss": 0.1673, "step": 1478 }, { "epoch": 1.36, "learning_rate": 1.8997462597393522e-05, "loss": 0.1618, "step": 1479 }, { "epoch": 1.36, "learning_rate": 1.8995842519903013e-05, "loss": 0.1566, "step": 1480 }, { "epoch": 1.36, "learning_rate": 1.8994221203658538e-05, "loss": 0.1559, "step": 1481 }, { "epoch": 1.36, "learning_rate": 1.899259864888336e-05, "loss": 0.153, "step": 1482 }, { "epoch": 1.36, "learning_rate": 1.8990974855800918e-05, "loss": 0.151, "step": 1483 }, { "epoch": 1.36, "learning_rate": 1.8989349824634803e-05, "loss": 0.1518, "step": 1484 }, { "epoch": 1.36, "learning_rate": 1.8987723555608793e-05, "loss": 0.1592, "step": 1485 }, { "epoch": 1.36, "learning_rate": 1.8986096048946826e-05, "loss": 0.1549, "step": 1486 }, { "epoch": 1.36, "learning_rate": 1.8984467304873013e-05, "loss": 0.1615, "step": 1487 }, { "epoch": 1.36, "learning_rate": 1.898283732361164e-05, "loss": 0.1593, "step": 1488 }, { "epoch": 1.36, "learning_rate": 1.898120610538716e-05, "loss": 0.1532, "step": 1489 }, { "epoch": 1.37, "learning_rate": 1.8979573650424196e-05, "loss": 0.1586, "step": 1490 }, { "epoch": 1.37, "learning_rate": 1.897793995894754e-05, "loss": 0.1557, "step": 1491 }, { "epoch": 1.37, "learning_rate": 1.897630503118216e-05, "loss": 0.158, "step": 1492 }, { "epoch": 1.37, "learning_rate": 1.8974668867353192e-05, "loss": 0.1595, "step": 1493 }, { "epoch": 1.37, "learning_rate": 1.8973031467685933e-05, "loss": 0.1467, "step": 1494 }, { "epoch": 1.37, "learning_rate": 1.897139283240586e-05, "loss": 0.1588, "step": 1495 }, { "epoch": 1.37, "learning_rate": 1.8969752961738625e-05, "loss": 0.1494, "step": 1496 }, { "epoch": 1.37, "learning_rate": 1.896811185591003e-05, "loss": 0.1515, "step": 1497 }, { "epoch": 1.37, "learning_rate": 1.8966469515146076e-05, "loss": 0.1585, "step": 1498 }, { "epoch": 1.37, "learning_rate": 1.8964825939672907e-05, "loss": 0.1522, "step": 1499 }, { "epoch": 1.37, "learning_rate": 1.896318112971685e-05, "loss": 0.1628, "step": 1500 }, { "epoch": 1.38, "learning_rate": 1.8961535085504402e-05, "loss": 0.1505, "step": 1501 }, { "epoch": 1.38, "learning_rate": 1.8959887807262225e-05, "loss": 0.1579, "step": 1502 }, { "epoch": 1.38, "learning_rate": 1.895823929521716e-05, "loss": 0.1432, "step": 1503 }, { "epoch": 1.38, "learning_rate": 1.8956589549596207e-05, "loss": 0.154, "step": 1504 }, { "epoch": 1.38, "learning_rate": 1.895493857062654e-05, "loss": 0.157, "step": 1505 }, { "epoch": 1.38, "learning_rate": 1.895328635853551e-05, "loss": 0.1531, "step": 1506 }, { "epoch": 1.38, "learning_rate": 1.8951632913550625e-05, "loss": 0.1575, "step": 1507 }, { "epoch": 1.38, "learning_rate": 1.8949978235899573e-05, "loss": 0.1595, "step": 1508 }, { "epoch": 1.38, "learning_rate": 1.8948322325810203e-05, "loss": 0.1565, "step": 1509 }, { "epoch": 1.38, "learning_rate": 1.8946665183510546e-05, "loss": 0.1542, "step": 1510 }, { "epoch": 1.38, "learning_rate": 1.8945006809228788e-05, "loss": 0.1553, "step": 1511 }, { "epoch": 1.39, "learning_rate": 1.8943347203193298e-05, "loss": 0.153, "step": 1512 }, { "epoch": 1.39, "learning_rate": 1.8941686365632607e-05, "loss": 0.1583, "step": 1513 }, { "epoch": 1.39, "learning_rate": 1.8940024296775415e-05, "loss": 0.1574, "step": 1514 }, { "epoch": 1.39, "learning_rate": 1.8938360996850597e-05, "loss": 0.1539, "step": 1515 }, { "epoch": 1.39, "learning_rate": 1.893669646608719e-05, "loss": 0.1521, "step": 1516 }, { "epoch": 1.39, "learning_rate": 1.8935030704714407e-05, "loss": 0.1557, "step": 1517 }, { "epoch": 1.39, "learning_rate": 1.893336371296163e-05, "loss": 0.1477, "step": 1518 }, { "epoch": 1.39, "learning_rate": 1.893169549105841e-05, "loss": 0.1567, "step": 1519 }, { "epoch": 1.39, "learning_rate": 1.893002603923446e-05, "loss": 0.1563, "step": 1520 }, { "epoch": 1.39, "learning_rate": 1.8928355357719673e-05, "loss": 0.1523, "step": 1521 }, { "epoch": 1.4, "learning_rate": 1.8926683446744102e-05, "loss": 0.1582, "step": 1522 }, { "epoch": 1.4, "learning_rate": 1.892501030653798e-05, "loss": 0.1486, "step": 1523 }, { "epoch": 1.4, "learning_rate": 1.8923335937331702e-05, "loss": 0.1562, "step": 1524 }, { "epoch": 1.4, "learning_rate": 1.8921660339355828e-05, "loss": 0.1637, "step": 1525 }, { "epoch": 1.4, "learning_rate": 1.8919983512841104e-05, "loss": 0.1586, "step": 1526 }, { "epoch": 1.4, "learning_rate": 1.891830545801842e-05, "loss": 0.1597, "step": 1527 }, { "epoch": 1.4, "learning_rate": 1.8916626175118863e-05, "loss": 0.1514, "step": 1528 }, { "epoch": 1.4, "learning_rate": 1.8914945664373664e-05, "loss": 0.1681, "step": 1529 }, { "epoch": 1.4, "learning_rate": 1.891326392601424e-05, "loss": 0.1517, "step": 1530 }, { "epoch": 1.4, "learning_rate": 1.8911580960272168e-05, "loss": 0.1464, "step": 1531 }, { "epoch": 1.4, "learning_rate": 1.89098967673792e-05, "loss": 0.1544, "step": 1532 }, { "epoch": 1.41, "learning_rate": 1.8908211347567257e-05, "loss": 0.1531, "step": 1533 }, { "epoch": 1.41, "learning_rate": 1.890652470106842e-05, "loss": 0.1527, "step": 1534 }, { "epoch": 1.41, "learning_rate": 1.890483682811495e-05, "loss": 0.1553, "step": 1535 }, { "epoch": 1.41, "learning_rate": 1.8903147728939265e-05, "loss": 0.1703, "step": 1536 }, { "epoch": 1.41, "learning_rate": 1.890145740377397e-05, "loss": 0.152, "step": 1537 }, { "epoch": 1.41, "learning_rate": 1.8899765852851815e-05, "loss": 0.1545, "step": 1538 }, { "epoch": 1.41, "learning_rate": 1.8898073076405744e-05, "loss": 0.1589, "step": 1539 }, { "epoch": 1.41, "learning_rate": 1.8896379074668848e-05, "loss": 0.1475, "step": 1540 }, { "epoch": 1.41, "learning_rate": 1.88946838478744e-05, "loss": 0.146, "step": 1541 }, { "epoch": 1.41, "learning_rate": 1.889298739625584e-05, "loss": 0.1576, "step": 1542 }, { "epoch": 1.41, "learning_rate": 1.8891289720046774e-05, "loss": 0.1577, "step": 1543 }, { "epoch": 1.42, "learning_rate": 1.8889590819480973e-05, "loss": 0.1511, "step": 1544 }, { "epoch": 1.42, "learning_rate": 1.8887890694792377e-05, "loss": 0.1549, "step": 1545 }, { "epoch": 1.42, "learning_rate": 1.8886189346215107e-05, "loss": 0.157, "step": 1546 }, { "epoch": 1.42, "learning_rate": 1.888448677398344e-05, "loss": 0.1584, "step": 1547 }, { "epoch": 1.42, "learning_rate": 1.8882782978331825e-05, "loss": 0.1619, "step": 1548 }, { "epoch": 1.42, "learning_rate": 1.8881077959494882e-05, "loss": 0.1581, "step": 1549 }, { "epoch": 1.42, "learning_rate": 1.8879371717707396e-05, "loss": 0.1565, "step": 1550 }, { "epoch": 1.42, "learning_rate": 1.8877664253204317e-05, "loss": 0.1603, "step": 1551 }, { "epoch": 1.42, "learning_rate": 1.887595556622077e-05, "loss": 0.1654, "step": 1552 }, { "epoch": 1.42, "learning_rate": 1.8874245656992052e-05, "loss": 0.1533, "step": 1553 }, { "epoch": 1.42, "learning_rate": 1.8872534525753617e-05, "loss": 0.1565, "step": 1554 }, { "epoch": 1.43, "learning_rate": 1.8870822172741093e-05, "loss": 0.1488, "step": 1555 }, { "epoch": 1.43, "learning_rate": 1.886910859819028e-05, "loss": 0.1547, "step": 1556 }, { "epoch": 1.43, "learning_rate": 1.886739380233714e-05, "loss": 0.1512, "step": 1557 }, { "epoch": 1.43, "learning_rate": 1.88656777854178e-05, "loss": 0.1511, "step": 1558 }, { "epoch": 1.43, "learning_rate": 1.886396054766857e-05, "loss": 0.1578, "step": 1559 }, { "epoch": 1.43, "learning_rate": 1.886224208932591e-05, "loss": 0.1564, "step": 1560 }, { "epoch": 1.43, "learning_rate": 1.8860522410626462e-05, "loss": 0.1477, "step": 1561 }, { "epoch": 1.43, "learning_rate": 1.885880151180703e-05, "loss": 0.1509, "step": 1562 }, { "epoch": 1.43, "learning_rate": 1.8857079393104588e-05, "loss": 0.1509, "step": 1563 }, { "epoch": 1.43, "learning_rate": 1.8855356054756274e-05, "loss": 0.1474, "step": 1564 }, { "epoch": 1.43, "learning_rate": 1.88536314969994e-05, "loss": 0.165, "step": 1565 }, { "epoch": 1.44, "learning_rate": 1.8851905720071438e-05, "loss": 0.1652, "step": 1566 }, { "epoch": 1.44, "learning_rate": 1.8850178724210037e-05, "loss": 0.1499, "step": 1567 }, { "epoch": 1.44, "learning_rate": 1.8848450509653006e-05, "loss": 0.1545, "step": 1568 }, { "epoch": 1.44, "learning_rate": 1.884672107663833e-05, "loss": 0.1592, "step": 1569 }, { "epoch": 1.44, "learning_rate": 1.8844990425404147e-05, "loss": 0.151, "step": 1570 }, { "epoch": 1.44, "learning_rate": 1.8843258556188787e-05, "loss": 0.1531, "step": 1571 }, { "epoch": 1.44, "learning_rate": 1.8841525469230724e-05, "loss": 0.1531, "step": 1572 }, { "epoch": 1.44, "learning_rate": 1.883979116476861e-05, "loss": 0.151, "step": 1573 }, { "epoch": 1.44, "learning_rate": 1.883805564304127e-05, "loss": 0.1482, "step": 1574 }, { "epoch": 1.44, "learning_rate": 1.883631890428768e-05, "loss": 0.1527, "step": 1575 }, { "epoch": 1.44, "learning_rate": 1.8834580948746997e-05, "loss": 0.1508, "step": 1576 }, { "epoch": 1.45, "learning_rate": 1.8832841776658552e-05, "loss": 0.1563, "step": 1577 }, { "epoch": 1.45, "learning_rate": 1.8831101388261827e-05, "loss": 0.1502, "step": 1578 }, { "epoch": 1.45, "learning_rate": 1.8829359783796474e-05, "loss": 0.1537, "step": 1579 }, { "epoch": 1.45, "learning_rate": 1.8827616963502325e-05, "loss": 0.1585, "step": 1580 }, { "epoch": 1.45, "learning_rate": 1.882587292761937e-05, "loss": 0.1656, "step": 1581 }, { "epoch": 1.45, "learning_rate": 1.8824127676387764e-05, "loss": 0.1632, "step": 1582 }, { "epoch": 1.45, "learning_rate": 1.8822381210047836e-05, "loss": 0.1519, "step": 1583 }, { "epoch": 1.45, "learning_rate": 1.882063352884008e-05, "loss": 0.1603, "step": 1584 }, { "epoch": 1.45, "learning_rate": 1.8818884633005153e-05, "loss": 0.1584, "step": 1585 }, { "epoch": 1.45, "learning_rate": 1.881713452278389e-05, "loss": 0.1482, "step": 1586 }, { "epoch": 1.45, "learning_rate": 1.8815383198417285e-05, "loss": 0.1592, "step": 1587 }, { "epoch": 1.46, "learning_rate": 1.881363066014649e-05, "loss": 0.1583, "step": 1588 }, { "epoch": 1.46, "learning_rate": 1.8811876908212846e-05, "loss": 0.1529, "step": 1589 }, { "epoch": 1.46, "learning_rate": 1.8810121942857848e-05, "loss": 0.1561, "step": 1590 }, { "epoch": 1.46, "learning_rate": 1.8808365764323157e-05, "loss": 0.1488, "step": 1591 }, { "epoch": 1.46, "learning_rate": 1.8806608372850604e-05, "loss": 0.1531, "step": 1592 }, { "epoch": 1.46, "learning_rate": 1.880484976868219e-05, "loss": 0.1519, "step": 1593 }, { "epoch": 1.46, "learning_rate": 1.8803089952060075e-05, "loss": 0.1493, "step": 1594 }, { "epoch": 1.46, "learning_rate": 1.8801328923226596e-05, "loss": 0.1514, "step": 1595 }, { "epoch": 1.46, "learning_rate": 1.879956668242425e-05, "loss": 0.1662, "step": 1596 }, { "epoch": 1.46, "learning_rate": 1.8797803229895702e-05, "loss": 0.1402, "step": 1597 }, { "epoch": 1.46, "learning_rate": 1.8796038565883788e-05, "loss": 0.1523, "step": 1598 }, { "epoch": 1.47, "learning_rate": 1.8794272690631502e-05, "loss": 0.1574, "step": 1599 }, { "epoch": 1.47, "learning_rate": 1.8792505604382013e-05, "loss": 0.1613, "step": 1600 }, { "epoch": 1.47, "learning_rate": 1.8790737307378655e-05, "loss": 0.1577, "step": 1601 }, { "epoch": 1.47, "learning_rate": 1.8788967799864927e-05, "loss": 0.164, "step": 1602 }, { "epoch": 1.47, "learning_rate": 1.8787197082084493e-05, "loss": 0.1505, "step": 1603 }, { "epoch": 1.47, "learning_rate": 1.878542515428119e-05, "loss": 0.1597, "step": 1604 }, { "epoch": 1.47, "learning_rate": 1.8783652016699014e-05, "loss": 0.1577, "step": 1605 }, { "epoch": 1.47, "learning_rate": 1.878187766958213e-05, "loss": 0.1535, "step": 1606 }, { "epoch": 1.47, "learning_rate": 1.878010211317488e-05, "loss": 0.1629, "step": 1607 }, { "epoch": 1.47, "learning_rate": 1.8778325347721754e-05, "loss": 0.1506, "step": 1608 }, { "epoch": 1.47, "learning_rate": 1.877654737346742e-05, "loss": 0.1555, "step": 1609 }, { "epoch": 1.48, "learning_rate": 1.8774768190656713e-05, "loss": 0.151, "step": 1610 }, { "epoch": 1.48, "learning_rate": 1.877298779953463e-05, "loss": 0.1542, "step": 1611 }, { "epoch": 1.48, "learning_rate": 1.8771206200346332e-05, "loss": 0.1609, "step": 1612 }, { "epoch": 1.48, "learning_rate": 1.8769423393337157e-05, "loss": 0.1576, "step": 1613 }, { "epoch": 1.48, "learning_rate": 1.87676393787526e-05, "loss": 0.1445, "step": 1614 }, { "epoch": 1.48, "learning_rate": 1.876585415683833e-05, "loss": 0.1536, "step": 1615 }, { "epoch": 1.48, "learning_rate": 1.8764067727840163e-05, "loss": 0.1592, "step": 1616 }, { "epoch": 1.48, "learning_rate": 1.8762280092004112e-05, "loss": 0.1484, "step": 1617 }, { "epoch": 1.48, "learning_rate": 1.876049124957633e-05, "loss": 0.1532, "step": 1618 }, { "epoch": 1.48, "learning_rate": 1.875870120080315e-05, "loss": 0.1464, "step": 1619 }, { "epoch": 1.48, "learning_rate": 1.8756909945931065e-05, "loss": 0.1538, "step": 1620 }, { "epoch": 1.49, "learning_rate": 1.8755117485206736e-05, "loss": 0.1583, "step": 1621 }, { "epoch": 1.49, "learning_rate": 1.875332381887699e-05, "loss": 0.1479, "step": 1622 }, { "epoch": 1.49, "learning_rate": 1.8751528947188825e-05, "loss": 0.1518, "step": 1623 }, { "epoch": 1.49, "learning_rate": 1.874973287038939e-05, "loss": 0.1566, "step": 1624 }, { "epoch": 1.49, "learning_rate": 1.874793558872602e-05, "loss": 0.147, "step": 1625 }, { "epoch": 1.49, "learning_rate": 1.8746137102446203e-05, "loss": 0.1647, "step": 1626 }, { "epoch": 1.49, "learning_rate": 1.8744337411797594e-05, "loss": 0.1548, "step": 1627 }, { "epoch": 1.49, "learning_rate": 1.8742536517028017e-05, "loss": 0.1526, "step": 1628 }, { "epoch": 1.49, "learning_rate": 1.874073441838546e-05, "loss": 0.1564, "step": 1629 }, { "epoch": 1.49, "learning_rate": 1.8738931116118076e-05, "loss": 0.1585, "step": 1630 }, { "epoch": 1.49, "learning_rate": 1.8737126610474187e-05, "loss": 0.1611, "step": 1631 }, { "epoch": 1.5, "learning_rate": 1.8735320901702278e-05, "loss": 0.1569, "step": 1632 }, { "epoch": 1.5, "learning_rate": 1.8733513990051e-05, "loss": 0.1583, "step": 1633 }, { "epoch": 1.5, "learning_rate": 1.873170587576917e-05, "loss": 0.1478, "step": 1634 }, { "epoch": 1.5, "learning_rate": 1.872989655910577e-05, "loss": 0.1484, "step": 1635 }, { "epoch": 1.5, "learning_rate": 1.8728086040309952e-05, "loss": 0.1512, "step": 1636 }, { "epoch": 1.5, "learning_rate": 1.872627431963103e-05, "loss": 0.1658, "step": 1637 }, { "epoch": 1.5, "learning_rate": 1.8724461397318476e-05, "loss": 0.1559, "step": 1638 }, { "epoch": 1.5, "learning_rate": 1.872264727362194e-05, "loss": 0.1428, "step": 1639 }, { "epoch": 1.5, "learning_rate": 1.872083194879123e-05, "loss": 0.1687, "step": 1640 }, { "epoch": 1.5, "learning_rate": 1.8719015423076324e-05, "loss": 0.1657, "step": 1641 }, { "epoch": 1.51, "learning_rate": 1.8717197696727364e-05, "loss": 0.1615, "step": 1642 }, { "epoch": 1.51, "learning_rate": 1.871537876999465e-05, "loss": 0.1469, "step": 1643 }, { "epoch": 1.51, "learning_rate": 1.8713558643128662e-05, "loss": 0.1598, "step": 1644 }, { "epoch": 1.51, "learning_rate": 1.8711737316380033e-05, "loss": 0.1539, "step": 1645 }, { "epoch": 1.51, "learning_rate": 1.870991478999956e-05, "loss": 0.1568, "step": 1646 }, { "epoch": 1.51, "learning_rate": 1.870809106423822e-05, "loss": 0.1535, "step": 1647 }, { "epoch": 1.51, "learning_rate": 1.8706266139347136e-05, "loss": 0.1557, "step": 1648 }, { "epoch": 1.51, "learning_rate": 1.8704440015577613e-05, "loss": 0.1548, "step": 1649 }, { "epoch": 1.51, "learning_rate": 1.8702612693181107e-05, "loss": 0.1464, "step": 1650 }, { "epoch": 1.51, "learning_rate": 1.870078417240925e-05, "loss": 0.148, "step": 1651 }, { "epoch": 1.51, "learning_rate": 1.869895445351383e-05, "loss": 0.1554, "step": 1652 }, { "epoch": 1.52, "learning_rate": 1.8697123536746815e-05, "loss": 0.1469, "step": 1653 }, { "epoch": 1.52, "learning_rate": 1.8695291422360317e-05, "loss": 0.1531, "step": 1654 }, { "epoch": 1.52, "learning_rate": 1.8693458110606628e-05, "loss": 0.1556, "step": 1655 }, { "epoch": 1.52, "learning_rate": 1.86916236017382e-05, "loss": 0.1556, "step": 1656 }, { "epoch": 1.52, "learning_rate": 1.8689787896007653e-05, "loss": 0.1506, "step": 1657 }, { "epoch": 1.52, "learning_rate": 1.8687950993667762e-05, "loss": 0.1449, "step": 1658 }, { "epoch": 1.52, "learning_rate": 1.868611289497148e-05, "loss": 0.1456, "step": 1659 }, { "epoch": 1.52, "learning_rate": 1.868427360017192e-05, "loss": 0.1496, "step": 1660 }, { "epoch": 1.52, "learning_rate": 1.868243310952235e-05, "loss": 0.1562, "step": 1661 }, { "epoch": 1.52, "learning_rate": 1.868059142327622e-05, "loss": 0.1596, "step": 1662 }, { "epoch": 1.52, "learning_rate": 1.8678748541687127e-05, "loss": 0.1542, "step": 1663 }, { "epoch": 1.53, "learning_rate": 1.867690446500885e-05, "loss": 0.1523, "step": 1664 }, { "epoch": 1.53, "learning_rate": 1.8675059193495317e-05, "loss": 0.144, "step": 1665 }, { "epoch": 1.53, "learning_rate": 1.867321272740063e-05, "loss": 0.143, "step": 1666 }, { "epoch": 1.53, "learning_rate": 1.8671365066979056e-05, "loss": 0.152, "step": 1667 }, { "epoch": 1.53, "learning_rate": 1.8669516212485013e-05, "loss": 0.1477, "step": 1668 }, { "epoch": 1.53, "learning_rate": 1.8667666164173106e-05, "loss": 0.1481, "step": 1669 }, { "epoch": 1.53, "learning_rate": 1.8665814922298088e-05, "loss": 0.1559, "step": 1670 }, { "epoch": 1.53, "learning_rate": 1.8663962487114874e-05, "loss": 0.147, "step": 1671 }, { "epoch": 1.53, "learning_rate": 1.8662108858878558e-05, "loss": 0.1476, "step": 1672 }, { "epoch": 1.53, "learning_rate": 1.866025403784439e-05, "loss": 0.1563, "step": 1673 }, { "epoch": 1.53, "learning_rate": 1.8658398024267777e-05, "loss": 0.1612, "step": 1674 }, { "epoch": 1.54, "learning_rate": 1.8656540818404304e-05, "loss": 0.168, "step": 1675 }, { "epoch": 1.54, "learning_rate": 1.865468242050971e-05, "loss": 0.1602, "step": 1676 }, { "epoch": 1.54, "learning_rate": 1.8652822830839902e-05, "loss": 0.1634, "step": 1677 }, { "epoch": 1.54, "learning_rate": 1.8650962049650955e-05, "loss": 0.1518, "step": 1678 }, { "epoch": 1.54, "learning_rate": 1.8649100077199102e-05, "loss": 0.162, "step": 1679 }, { "epoch": 1.54, "learning_rate": 1.864723691374074e-05, "loss": 0.1557, "step": 1680 }, { "epoch": 1.54, "learning_rate": 1.8645372559532434e-05, "loss": 0.1586, "step": 1681 }, { "epoch": 1.54, "learning_rate": 1.864350701483091e-05, "loss": 0.1665, "step": 1682 }, { "epoch": 1.54, "learning_rate": 1.8641640279893063e-05, "loss": 0.1537, "step": 1683 }, { "epoch": 1.54, "learning_rate": 1.863977235497594e-05, "loss": 0.1453, "step": 1684 }, { "epoch": 1.54, "learning_rate": 1.8637903240336767e-05, "loss": 0.1649, "step": 1685 }, { "epoch": 1.55, "learning_rate": 1.8636032936232923e-05, "loss": 0.1557, "step": 1686 }, { "epoch": 1.55, "learning_rate": 1.8634161442921956e-05, "loss": 0.1415, "step": 1687 }, { "epoch": 1.55, "learning_rate": 1.8632288760661575e-05, "loss": 0.158, "step": 1688 }, { "epoch": 1.55, "learning_rate": 1.863041488970965e-05, "loss": 0.1479, "step": 1689 }, { "epoch": 1.55, "learning_rate": 1.862853983032423e-05, "loss": 0.1511, "step": 1690 }, { "epoch": 1.55, "learning_rate": 1.8626663582763504e-05, "loss": 0.1572, "step": 1691 }, { "epoch": 1.55, "learning_rate": 1.8624786147285846e-05, "loss": 0.1537, "step": 1692 }, { "epoch": 1.55, "learning_rate": 1.8622907524149778e-05, "loss": 0.1604, "step": 1693 }, { "epoch": 1.55, "learning_rate": 1.8621027713613993e-05, "loss": 0.1629, "step": 1694 }, { "epoch": 1.55, "learning_rate": 1.8619146715937345e-05, "loss": 0.1405, "step": 1695 }, { "epoch": 1.55, "learning_rate": 1.861726453137886e-05, "loss": 0.1529, "step": 1696 }, { "epoch": 1.56, "learning_rate": 1.8615381160197718e-05, "loss": 0.1515, "step": 1697 }, { "epoch": 1.56, "learning_rate": 1.861349660265326e-05, "loss": 0.1456, "step": 1698 }, { "epoch": 1.56, "learning_rate": 1.8611610859004993e-05, "loss": 0.1594, "step": 1699 }, { "epoch": 1.56, "learning_rate": 1.8609723929512596e-05, "loss": 0.1597, "step": 1700 }, { "epoch": 1.56, "learning_rate": 1.8607835814435905e-05, "loss": 0.1434, "step": 1701 }, { "epoch": 1.56, "learning_rate": 1.8605946514034917e-05, "loss": 0.159, "step": 1702 }, { "epoch": 1.56, "learning_rate": 1.860405602856979e-05, "loss": 0.1442, "step": 1703 }, { "epoch": 1.56, "learning_rate": 1.860216435830086e-05, "loss": 0.1577, "step": 1704 }, { "epoch": 1.56, "learning_rate": 1.8600271503488608e-05, "loss": 0.1548, "step": 1705 }, { "epoch": 1.56, "learning_rate": 1.8598377464393682e-05, "loss": 0.1566, "step": 1706 }, { "epoch": 1.56, "learning_rate": 1.8596482241276903e-05, "loss": 0.1545, "step": 1707 }, { "epoch": 1.57, "learning_rate": 1.859458583439925e-05, "loss": 0.1558, "step": 1708 }, { "epoch": 1.57, "learning_rate": 1.859268824402186e-05, "loss": 0.1444, "step": 1709 }, { "epoch": 1.57, "learning_rate": 1.8590789470406034e-05, "loss": 0.1578, "step": 1710 }, { "epoch": 1.57, "learning_rate": 1.8588889513813246e-05, "loss": 0.1458, "step": 1711 }, { "epoch": 1.57, "learning_rate": 1.8586988374505122e-05, "loss": 0.16, "step": 1712 }, { "epoch": 1.57, "learning_rate": 1.8585086052743452e-05, "loss": 0.1496, "step": 1713 }, { "epoch": 1.57, "learning_rate": 1.8583182548790196e-05, "loss": 0.1553, "step": 1714 }, { "epoch": 1.57, "learning_rate": 1.858127786290747e-05, "loss": 0.1522, "step": 1715 }, { "epoch": 1.57, "learning_rate": 1.8579371995357555e-05, "loss": 0.1508, "step": 1716 }, { "epoch": 1.57, "learning_rate": 1.8577464946402893e-05, "loss": 0.1568, "step": 1717 }, { "epoch": 1.57, "learning_rate": 1.8575556716306092e-05, "loss": 0.1519, "step": 1718 }, { "epoch": 1.58, "learning_rate": 1.8573647305329922e-05, "loss": 0.1571, "step": 1719 }, { "epoch": 1.58, "learning_rate": 1.8571736713737313e-05, "loss": 0.1496, "step": 1720 }, { "epoch": 1.58, "learning_rate": 1.856982494179136e-05, "loss": 0.1531, "step": 1721 }, { "epoch": 1.58, "learning_rate": 1.8567911989755318e-05, "loss": 0.1475, "step": 1722 }, { "epoch": 1.58, "learning_rate": 1.85659978578926e-05, "loss": 0.1462, "step": 1723 }, { "epoch": 1.58, "learning_rate": 1.8564082546466804e-05, "loss": 0.1494, "step": 1724 }, { "epoch": 1.58, "learning_rate": 1.856216605574166e-05, "loss": 0.1607, "step": 1725 }, { "epoch": 1.58, "learning_rate": 1.856024838598108e-05, "loss": 0.1501, "step": 1726 }, { "epoch": 1.58, "learning_rate": 1.8558329537449128e-05, "loss": 0.1516, "step": 1727 }, { "epoch": 1.58, "learning_rate": 1.8556409510410045e-05, "loss": 0.155, "step": 1728 }, { "epoch": 1.58, "learning_rate": 1.8554488305128214e-05, "loss": 0.1523, "step": 1729 }, { "epoch": 1.59, "learning_rate": 1.8552565921868196e-05, "loss": 0.1512, "step": 1730 }, { "epoch": 1.59, "learning_rate": 1.8550642360894708e-05, "loss": 0.1574, "step": 1731 }, { "epoch": 1.59, "learning_rate": 1.8548717622472628e-05, "loss": 0.1495, "step": 1732 }, { "epoch": 1.59, "learning_rate": 1.8546791706867e-05, "loss": 0.1563, "step": 1733 }, { "epoch": 1.59, "learning_rate": 1.8544864614343026e-05, "loss": 0.1459, "step": 1734 }, { "epoch": 1.59, "learning_rate": 1.8542936345166075e-05, "loss": 0.1644, "step": 1735 }, { "epoch": 1.59, "learning_rate": 1.8541006899601676e-05, "loss": 0.152, "step": 1736 }, { "epoch": 1.59, "learning_rate": 1.853907627791552e-05, "loss": 0.1535, "step": 1737 }, { "epoch": 1.59, "learning_rate": 1.8537144480373455e-05, "loss": 0.1656, "step": 1738 }, { "epoch": 1.59, "learning_rate": 1.8535211507241495e-05, "loss": 0.1579, "step": 1739 }, { "epoch": 1.59, "learning_rate": 1.853327735878582e-05, "loss": 0.1527, "step": 1740 }, { "epoch": 1.6, "learning_rate": 1.8531342035272768e-05, "loss": 0.1647, "step": 1741 }, { "epoch": 1.6, "learning_rate": 1.8529405536968835e-05, "loss": 0.1541, "step": 1742 }, { "epoch": 1.6, "learning_rate": 1.8527467864140687e-05, "loss": 0.1511, "step": 1743 }, { "epoch": 1.6, "learning_rate": 1.8525529017055144e-05, "loss": 0.1507, "step": 1744 }, { "epoch": 1.6, "learning_rate": 1.8523588995979192e-05, "loss": 0.156, "step": 1745 }, { "epoch": 1.6, "learning_rate": 1.8521647801179975e-05, "loss": 0.1519, "step": 1746 }, { "epoch": 1.6, "learning_rate": 1.8519705432924806e-05, "loss": 0.1453, "step": 1747 }, { "epoch": 1.6, "learning_rate": 1.851776189148115e-05, "loss": 0.1511, "step": 1748 }, { "epoch": 1.6, "learning_rate": 1.8515817177116646e-05, "loss": 0.1568, "step": 1749 }, { "epoch": 1.6, "learning_rate": 1.8513871290099074e-05, "loss": 0.1585, "step": 1750 }, { "epoch": 1.6, "learning_rate": 1.85119242306964e-05, "loss": 0.1494, "step": 1751 }, { "epoch": 1.61, "learning_rate": 1.8509975999176736e-05, "loss": 0.1524, "step": 1752 }, { "epoch": 1.61, "learning_rate": 1.850802659580836e-05, "loss": 0.1591, "step": 1753 }, { "epoch": 1.61, "learning_rate": 1.8506076020859708e-05, "loss": 0.1485, "step": 1754 }, { "epoch": 1.61, "learning_rate": 1.850412427459938e-05, "loss": 0.1485, "step": 1755 }, { "epoch": 1.61, "learning_rate": 1.8502171357296144e-05, "loss": 0.1414, "step": 1756 }, { "epoch": 1.61, "learning_rate": 1.850021726921891e-05, "loss": 0.1499, "step": 1757 }, { "epoch": 1.61, "learning_rate": 1.8498262010636777e-05, "loss": 0.1478, "step": 1758 }, { "epoch": 1.61, "learning_rate": 1.8496305581818978e-05, "loss": 0.1486, "step": 1759 }, { "epoch": 1.61, "learning_rate": 1.849434798303492e-05, "loss": 0.1479, "step": 1760 }, { "epoch": 1.61, "learning_rate": 1.8492389214554177e-05, "loss": 0.1481, "step": 1761 }, { "epoch": 1.62, "learning_rate": 1.849042927664647e-05, "loss": 0.1473, "step": 1762 }, { "epoch": 1.62, "learning_rate": 1.848846816958169e-05, "loss": 0.1533, "step": 1763 }, { "epoch": 1.62, "learning_rate": 1.8486505893629894e-05, "loss": 0.1482, "step": 1764 }, { "epoch": 1.62, "learning_rate": 1.8484542449061284e-05, "loss": 0.1522, "step": 1765 }, { "epoch": 1.62, "learning_rate": 1.8482577836146238e-05, "loss": 0.1358, "step": 1766 }, { "epoch": 1.62, "learning_rate": 1.8480612055155282e-05, "loss": 0.1601, "step": 1767 }, { "epoch": 1.62, "learning_rate": 1.8478645106359118e-05, "loss": 0.1539, "step": 1768 }, { "epoch": 1.62, "learning_rate": 1.84766769900286e-05, "loss": 0.15, "step": 1769 }, { "epoch": 1.62, "learning_rate": 1.8474707706434735e-05, "loss": 0.1555, "step": 1770 }, { "epoch": 1.62, "learning_rate": 1.8472737255848705e-05, "loss": 0.1542, "step": 1771 }, { "epoch": 1.62, "learning_rate": 1.847076563854185e-05, "loss": 0.1489, "step": 1772 }, { "epoch": 1.63, "learning_rate": 1.846879285478566e-05, "loss": 0.1495, "step": 1773 }, { "epoch": 1.63, "learning_rate": 1.84668189048518e-05, "loss": 0.1567, "step": 1774 }, { "epoch": 1.63, "learning_rate": 1.8464843789012086e-05, "loss": 0.1589, "step": 1775 }, { "epoch": 1.63, "learning_rate": 1.8462867507538495e-05, "loss": 0.1527, "step": 1776 }, { "epoch": 1.63, "learning_rate": 1.846089006070317e-05, "loss": 0.1473, "step": 1777 }, { "epoch": 1.63, "learning_rate": 1.845891144877841e-05, "loss": 0.1544, "step": 1778 }, { "epoch": 1.63, "learning_rate": 1.8456931672036676e-05, "loss": 0.1513, "step": 1779 }, { "epoch": 1.63, "learning_rate": 1.8454950730750586e-05, "loss": 0.1461, "step": 1780 }, { "epoch": 1.63, "learning_rate": 1.8452968625192928e-05, "loss": 0.141, "step": 1781 }, { "epoch": 1.63, "learning_rate": 1.8450985355636638e-05, "loss": 0.1458, "step": 1782 }, { "epoch": 1.63, "learning_rate": 1.8449000922354823e-05, "loss": 0.1518, "step": 1783 }, { "epoch": 1.64, "learning_rate": 1.844701532562074e-05, "loss": 0.1493, "step": 1784 }, { "epoch": 1.64, "learning_rate": 1.8445028565707814e-05, "loss": 0.1504, "step": 1785 }, { "epoch": 1.64, "learning_rate": 1.844304064288963e-05, "loss": 0.1535, "step": 1786 }, { "epoch": 1.64, "learning_rate": 1.8441051557439923e-05, "loss": 0.1511, "step": 1787 }, { "epoch": 1.64, "learning_rate": 1.8439061309632607e-05, "loss": 0.1608, "step": 1788 }, { "epoch": 1.64, "learning_rate": 1.8437069899741736e-05, "loss": 0.1525, "step": 1789 }, { "epoch": 1.64, "learning_rate": 1.843507732804154e-05, "loss": 0.1605, "step": 1790 }, { "epoch": 1.64, "learning_rate": 1.843308359480639e-05, "loss": 0.152, "step": 1791 }, { "epoch": 1.64, "learning_rate": 1.8431088700310846e-05, "loss": 0.1668, "step": 1792 }, { "epoch": 1.64, "learning_rate": 1.84290926448296e-05, "loss": 0.1451, "step": 1793 }, { "epoch": 1.64, "learning_rate": 1.8427095428637514e-05, "loss": 0.1453, "step": 1794 }, { "epoch": 1.65, "learning_rate": 1.8425097052009615e-05, "loss": 0.1559, "step": 1795 }, { "epoch": 1.65, "learning_rate": 1.8423097515221083e-05, "loss": 0.1517, "step": 1796 }, { "epoch": 1.65, "learning_rate": 1.842109681854726e-05, "loss": 0.1518, "step": 1797 }, { "epoch": 1.65, "learning_rate": 1.841909496226365e-05, "loss": 0.1556, "step": 1798 }, { "epoch": 1.65, "learning_rate": 1.841709194664591e-05, "loss": 0.1512, "step": 1799 }, { "epoch": 1.65, "learning_rate": 1.841508777196987e-05, "loss": 0.1558, "step": 1800 }, { "epoch": 1.65, "learning_rate": 1.8413082438511504e-05, "loss": 0.1411, "step": 1801 }, { "epoch": 1.65, "learning_rate": 1.841107594654695e-05, "loss": 0.1526, "step": 1802 }, { "epoch": 1.65, "learning_rate": 1.840906829635251e-05, "loss": 0.1508, "step": 1803 }, { "epoch": 1.65, "learning_rate": 1.8407059488204647e-05, "loss": 0.1407, "step": 1804 }, { "epoch": 1.65, "learning_rate": 1.8405049522379975e-05, "loss": 0.1444, "step": 1805 }, { "epoch": 1.66, "learning_rate": 1.8403038399155278e-05, "loss": 0.1453, "step": 1806 }, { "epoch": 1.66, "learning_rate": 1.8401026118807487e-05, "loss": 0.1483, "step": 1807 }, { "epoch": 1.66, "learning_rate": 1.8399012681613705e-05, "loss": 0.1546, "step": 1808 }, { "epoch": 1.66, "learning_rate": 1.8396998087851182e-05, "loss": 0.1584, "step": 1809 }, { "epoch": 1.66, "learning_rate": 1.839498233779734e-05, "loss": 0.1522, "step": 1810 }, { "epoch": 1.66, "learning_rate": 1.8392965431729745e-05, "loss": 0.1405, "step": 1811 }, { "epoch": 1.66, "learning_rate": 1.8390947369926138e-05, "loss": 0.1487, "step": 1812 }, { "epoch": 1.66, "learning_rate": 1.838892815266441e-05, "loss": 0.1502, "step": 1813 }, { "epoch": 1.66, "learning_rate": 1.8386907780222615e-05, "loss": 0.1569, "step": 1814 }, { "epoch": 1.66, "learning_rate": 1.838488625287896e-05, "loss": 0.1499, "step": 1815 }, { "epoch": 1.66, "learning_rate": 1.8382863570911823e-05, "loss": 0.1434, "step": 1816 }, { "epoch": 1.67, "learning_rate": 1.8380839734599724e-05, "loss": 0.147, "step": 1817 }, { "epoch": 1.67, "learning_rate": 1.8378814744221358e-05, "loss": 0.1413, "step": 1818 }, { "epoch": 1.67, "learning_rate": 1.8376788600055565e-05, "loss": 0.1532, "step": 1819 }, { "epoch": 1.67, "learning_rate": 1.8374761302381362e-05, "loss": 0.1545, "step": 1820 }, { "epoch": 1.67, "learning_rate": 1.83727328514779e-05, "loss": 0.1466, "step": 1821 }, { "epoch": 1.67, "learning_rate": 1.8370703247624514e-05, "loss": 0.1489, "step": 1822 }, { "epoch": 1.67, "learning_rate": 1.8368672491100686e-05, "loss": 0.1484, "step": 1823 }, { "epoch": 1.67, "learning_rate": 1.8366640582186047e-05, "loss": 0.1579, "step": 1824 }, { "epoch": 1.67, "learning_rate": 1.836460752116041e-05, "loss": 0.1441, "step": 1825 }, { "epoch": 1.67, "learning_rate": 1.836257330830372e-05, "loss": 0.1469, "step": 1826 }, { "epoch": 1.67, "learning_rate": 1.8360537943896104e-05, "loss": 0.1661, "step": 1827 }, { "epoch": 1.68, "learning_rate": 1.8358501428217835e-05, "loss": 0.154, "step": 1828 }, { "epoch": 1.68, "learning_rate": 1.8356463761549347e-05, "loss": 0.1424, "step": 1829 }, { "epoch": 1.68, "learning_rate": 1.835442494417123e-05, "loss": 0.1513, "step": 1830 }, { "epoch": 1.68, "learning_rate": 1.8352384976364236e-05, "loss": 0.1473, "step": 1831 }, { "epoch": 1.68, "learning_rate": 1.835034385840928e-05, "loss": 0.1468, "step": 1832 }, { "epoch": 1.68, "learning_rate": 1.834830159058743e-05, "loss": 0.1512, "step": 1833 }, { "epoch": 1.68, "learning_rate": 1.8346258173179904e-05, "loss": 0.15, "step": 1834 }, { "epoch": 1.68, "learning_rate": 1.8344213606468087e-05, "loss": 0.1497, "step": 1835 }, { "epoch": 1.68, "learning_rate": 1.834216789073353e-05, "loss": 0.1444, "step": 1836 }, { "epoch": 1.68, "learning_rate": 1.834012102625793e-05, "loss": 0.137, "step": 1837 }, { "epoch": 1.68, "learning_rate": 1.8338073013323147e-05, "loss": 0.1496, "step": 1838 }, { "epoch": 1.69, "learning_rate": 1.8336023852211197e-05, "loss": 0.1462, "step": 1839 }, { "epoch": 1.69, "learning_rate": 1.8333973543204256e-05, "loss": 0.1502, "step": 1840 }, { "epoch": 1.69, "learning_rate": 1.8331922086584656e-05, "loss": 0.1492, "step": 1841 }, { "epoch": 1.69, "learning_rate": 1.832986948263489e-05, "loss": 0.1518, "step": 1842 }, { "epoch": 1.69, "learning_rate": 1.8327815731637612e-05, "loss": 0.1524, "step": 1843 }, { "epoch": 1.69, "learning_rate": 1.8325760833875622e-05, "loss": 0.1464, "step": 1844 }, { "epoch": 1.69, "learning_rate": 1.832370478963189e-05, "loss": 0.1489, "step": 1845 }, { "epoch": 1.69, "learning_rate": 1.8321647599189538e-05, "loss": 0.1449, "step": 1846 }, { "epoch": 1.69, "learning_rate": 1.8319589262831847e-05, "loss": 0.1481, "step": 1847 }, { "epoch": 1.69, "learning_rate": 1.8317529780842257e-05, "loss": 0.1555, "step": 1848 }, { "epoch": 1.69, "learning_rate": 1.8315469153504365e-05, "loss": 0.1482, "step": 1849 }, { "epoch": 1.7, "learning_rate": 1.831340738110192e-05, "loss": 0.15, "step": 1850 }, { "epoch": 1.7, "learning_rate": 1.8311344463918846e-05, "loss": 0.1492, "step": 1851 }, { "epoch": 1.7, "learning_rate": 1.83092804022392e-05, "loss": 0.1478, "step": 1852 }, { "epoch": 1.7, "learning_rate": 1.8307215196347215e-05, "loss": 0.1454, "step": 1853 }, { "epoch": 1.7, "learning_rate": 1.8305148846527275e-05, "loss": 0.156, "step": 1854 }, { "epoch": 1.7, "learning_rate": 1.8303081353063928e-05, "loss": 0.1449, "step": 1855 }, { "epoch": 1.7, "learning_rate": 1.8301012716241864e-05, "loss": 0.156, "step": 1856 }, { "epoch": 1.7, "learning_rate": 1.8298942936345948e-05, "loss": 0.1547, "step": 1857 }, { "epoch": 1.7, "learning_rate": 1.8296872013661192e-05, "loss": 0.1466, "step": 1858 }, { "epoch": 1.7, "learning_rate": 1.8294799948472766e-05, "loss": 0.1386, "step": 1859 }, { "epoch": 1.7, "learning_rate": 1.8292726741066008e-05, "loss": 0.1475, "step": 1860 }, { "epoch": 1.71, "learning_rate": 1.829065239172639e-05, "loss": 0.1447, "step": 1861 }, { "epoch": 1.71, "learning_rate": 1.8288576900739573e-05, "loss": 0.1523, "step": 1862 }, { "epoch": 1.71, "learning_rate": 1.8286500268391343e-05, "loss": 0.1403, "step": 1863 }, { "epoch": 1.71, "learning_rate": 1.8284422494967668e-05, "loss": 0.1507, "step": 1864 }, { "epoch": 1.71, "learning_rate": 1.8282343580754663e-05, "loss": 0.1602, "step": 1865 }, { "epoch": 1.71, "learning_rate": 1.8280263526038594e-05, "loss": 0.1484, "step": 1866 }, { "epoch": 1.71, "learning_rate": 1.82781823311059e-05, "loss": 0.1502, "step": 1867 }, { "epoch": 1.71, "learning_rate": 1.8276099996243158e-05, "loss": 0.1453, "step": 1868 }, { "epoch": 1.71, "learning_rate": 1.8274016521737115e-05, "loss": 0.1433, "step": 1869 }, { "epoch": 1.71, "learning_rate": 1.8271931907874675e-05, "loss": 0.1522, "step": 1870 }, { "epoch": 1.71, "learning_rate": 1.8269846154942897e-05, "loss": 0.1443, "step": 1871 }, { "epoch": 1.72, "learning_rate": 1.826775926322899e-05, "loss": 0.1494, "step": 1872 }, { "epoch": 1.72, "learning_rate": 1.826567123302032e-05, "loss": 0.1586, "step": 1873 }, { "epoch": 1.72, "learning_rate": 1.826358206460443e-05, "loss": 0.1595, "step": 1874 }, { "epoch": 1.72, "learning_rate": 1.8261491758268993e-05, "loss": 0.1452, "step": 1875 }, { "epoch": 1.72, "learning_rate": 1.825940031430185e-05, "loss": 0.1461, "step": 1876 }, { "epoch": 1.72, "learning_rate": 1.8257307732991008e-05, "loss": 0.1459, "step": 1877 }, { "epoch": 1.72, "learning_rate": 1.8255214014624613e-05, "loss": 0.1385, "step": 1878 }, { "epoch": 1.72, "learning_rate": 1.8253119159490983e-05, "loss": 0.1553, "step": 1879 }, { "epoch": 1.72, "learning_rate": 1.8251023167878577e-05, "loss": 0.147, "step": 1880 }, { "epoch": 1.72, "learning_rate": 1.8248926040076024e-05, "loss": 0.1393, "step": 1881 }, { "epoch": 1.73, "learning_rate": 1.824682777637211e-05, "loss": 0.1543, "step": 1882 }, { "epoch": 1.73, "learning_rate": 1.8244728377055763e-05, "loss": 0.1533, "step": 1883 }, { "epoch": 1.73, "learning_rate": 1.8242627842416083e-05, "loss": 0.1613, "step": 1884 }, { "epoch": 1.73, "learning_rate": 1.8240526172742314e-05, "loss": 0.1428, "step": 1885 }, { "epoch": 1.73, "learning_rate": 1.823842336832387e-05, "loss": 0.1511, "step": 1886 }, { "epoch": 1.73, "learning_rate": 1.8236319429450304e-05, "loss": 0.1539, "step": 1887 }, { "epoch": 1.73, "learning_rate": 1.8234214356411343e-05, "loss": 0.1507, "step": 1888 }, { "epoch": 1.73, "learning_rate": 1.8232108149496853e-05, "loss": 0.1516, "step": 1889 }, { "epoch": 1.73, "learning_rate": 1.8230000808996875e-05, "loss": 0.1494, "step": 1890 }, { "epoch": 1.73, "learning_rate": 1.8227892335201586e-05, "loss": 0.1445, "step": 1891 }, { "epoch": 1.73, "learning_rate": 1.8225782728401336e-05, "loss": 0.1535, "step": 1892 }, { "epoch": 1.74, "learning_rate": 1.8223671988886623e-05, "loss": 0.1412, "step": 1893 }, { "epoch": 1.74, "learning_rate": 1.8221560116948103e-05, "loss": 0.1514, "step": 1894 }, { "epoch": 1.74, "learning_rate": 1.8219447112876582e-05, "loss": 0.155, "step": 1895 }, { "epoch": 1.74, "learning_rate": 1.8217332976963028e-05, "loss": 0.1491, "step": 1896 }, { "epoch": 1.74, "learning_rate": 1.821521770949857e-05, "loss": 0.1567, "step": 1897 }, { "epoch": 1.74, "learning_rate": 1.821310131077448e-05, "loss": 0.1576, "step": 1898 }, { "epoch": 1.74, "learning_rate": 1.8210983781082192e-05, "loss": 0.1535, "step": 1899 }, { "epoch": 1.74, "learning_rate": 1.82088651207133e-05, "loss": 0.142, "step": 1900 }, { "epoch": 1.74, "learning_rate": 1.820674532995955e-05, "loss": 0.1531, "step": 1901 }, { "epoch": 1.74, "learning_rate": 1.8204624409112842e-05, "loss": 0.1533, "step": 1902 }, { "epoch": 1.74, "learning_rate": 1.820250235846523e-05, "loss": 0.1595, "step": 1903 }, { "epoch": 1.75, "learning_rate": 1.8200379178308928e-05, "loss": 0.1464, "step": 1904 }, { "epoch": 1.75, "learning_rate": 1.819825486893631e-05, "loss": 0.1421, "step": 1905 }, { "epoch": 1.75, "learning_rate": 1.8196129430639895e-05, "loss": 0.1487, "step": 1906 }, { "epoch": 1.75, "learning_rate": 1.819400286371236e-05, "loss": 0.1624, "step": 1907 }, { "epoch": 1.75, "learning_rate": 1.819187516844654e-05, "loss": 0.1456, "step": 1908 }, { "epoch": 1.75, "learning_rate": 1.818974634513543e-05, "loss": 0.1401, "step": 1909 }, { "epoch": 1.75, "learning_rate": 1.818761639407217e-05, "loss": 0.1445, "step": 1910 }, { "epoch": 1.75, "learning_rate": 1.8185485315550062e-05, "loss": 0.1542, "step": 1911 }, { "epoch": 1.75, "learning_rate": 1.8183353109862563e-05, "loss": 0.1454, "step": 1912 }, { "epoch": 1.75, "learning_rate": 1.8181219777303284e-05, "loss": 0.1539, "step": 1913 }, { "epoch": 1.75, "learning_rate": 1.817908531816599e-05, "loss": 0.1437, "step": 1914 }, { "epoch": 1.76, "learning_rate": 1.81769497327446e-05, "loss": 0.1481, "step": 1915 }, { "epoch": 1.76, "learning_rate": 1.81748130213332e-05, "loss": 0.1469, "step": 1916 }, { "epoch": 1.76, "learning_rate": 1.8172675184226008e-05, "loss": 0.1453, "step": 1917 }, { "epoch": 1.76, "learning_rate": 1.8170536221717416e-05, "loss": 0.1419, "step": 1918 }, { "epoch": 1.76, "learning_rate": 1.8168396134101968e-05, "loss": 0.1594, "step": 1919 }, { "epoch": 1.76, "learning_rate": 1.816625492167436e-05, "loss": 0.158, "step": 1920 }, { "epoch": 1.76, "learning_rate": 1.816411258472944e-05, "loss": 0.1524, "step": 1921 }, { "epoch": 1.76, "learning_rate": 1.816196912356222e-05, "loss": 0.1595, "step": 1922 }, { "epoch": 1.76, "learning_rate": 1.815982453846785e-05, "loss": 0.1487, "step": 1923 }, { "epoch": 1.76, "learning_rate": 1.815767882974166e-05, "loss": 0.1449, "step": 1924 }, { "epoch": 1.76, "learning_rate": 1.8155531997679106e-05, "loss": 0.1422, "step": 1925 }, { "epoch": 1.77, "learning_rate": 1.8153384042575827e-05, "loss": 0.1545, "step": 1926 }, { "epoch": 1.77, "learning_rate": 1.8151234964727592e-05, "loss": 0.1463, "step": 1927 }, { "epoch": 1.77, "learning_rate": 1.814908476443034e-05, "loss": 0.1442, "step": 1928 }, { "epoch": 1.77, "learning_rate": 1.814693344198016e-05, "loss": 0.146, "step": 1929 }, { "epoch": 1.77, "learning_rate": 1.8144780997673293e-05, "loss": 0.1614, "step": 1930 }, { "epoch": 1.77, "learning_rate": 1.814262743180614e-05, "loss": 0.1505, "step": 1931 }, { "epoch": 1.77, "learning_rate": 1.8140472744675247e-05, "loss": 0.1532, "step": 1932 }, { "epoch": 1.77, "learning_rate": 1.8138316936577333e-05, "loss": 0.1458, "step": 1933 }, { "epoch": 1.77, "learning_rate": 1.8136160007809246e-05, "loss": 0.1434, "step": 1934 }, { "epoch": 1.77, "learning_rate": 1.8134001958668006e-05, "loss": 0.1518, "step": 1935 }, { "epoch": 1.77, "learning_rate": 1.8131842789450787e-05, "loss": 0.1555, "step": 1936 }, { "epoch": 1.78, "learning_rate": 1.8129682500454907e-05, "loss": 0.1595, "step": 1937 }, { "epoch": 1.78, "learning_rate": 1.812752109197785e-05, "loss": 0.1582, "step": 1938 }, { "epoch": 1.78, "learning_rate": 1.8125358564317242e-05, "loss": 0.1506, "step": 1939 }, { "epoch": 1.78, "learning_rate": 1.8123194917770878e-05, "loss": 0.1422, "step": 1940 }, { "epoch": 1.78, "learning_rate": 1.812103015263669e-05, "loss": 0.1515, "step": 1941 }, { "epoch": 1.78, "learning_rate": 1.8118864269212775e-05, "loss": 0.1453, "step": 1942 }, { "epoch": 1.78, "learning_rate": 1.811669726779738e-05, "loss": 0.1568, "step": 1943 }, { "epoch": 1.78, "learning_rate": 1.8114529148688912e-05, "loss": 0.143, "step": 1944 }, { "epoch": 1.78, "learning_rate": 1.8112359912185923e-05, "loss": 0.1479, "step": 1945 }, { "epoch": 1.78, "learning_rate": 1.8110189558587127e-05, "loss": 0.1502, "step": 1946 }, { "epoch": 1.78, "learning_rate": 1.8108018088191386e-05, "loss": 0.152, "step": 1947 }, { "epoch": 1.79, "learning_rate": 1.8105845501297722e-05, "loss": 0.1576, "step": 1948 }, { "epoch": 1.79, "learning_rate": 1.8103671798205294e-05, "loss": 0.1459, "step": 1949 }, { "epoch": 1.79, "learning_rate": 1.8101496979213443e-05, "loss": 0.1493, "step": 1950 }, { "epoch": 1.79, "learning_rate": 1.8099321044621637e-05, "loss": 0.1539, "step": 1951 }, { "epoch": 1.79, "learning_rate": 1.8097143994729517e-05, "loss": 0.148, "step": 1952 }, { "epoch": 1.79, "learning_rate": 1.809496582983686e-05, "loss": 0.1492, "step": 1953 }, { "epoch": 1.79, "learning_rate": 1.8092786550243613e-05, "loss": 0.1432, "step": 1954 }, { "epoch": 1.79, "learning_rate": 1.8090606156249866e-05, "loss": 0.1374, "step": 1955 }, { "epoch": 1.79, "learning_rate": 1.8088424648155866e-05, "loss": 0.142, "step": 1956 }, { "epoch": 1.79, "learning_rate": 1.8086242026262014e-05, "loss": 0.1453, "step": 1957 }, { "epoch": 1.79, "learning_rate": 1.808405829086886e-05, "loss": 0.145, "step": 1958 }, { "epoch": 1.8, "learning_rate": 1.8081873442277122e-05, "loss": 0.1544, "step": 1959 }, { "epoch": 1.8, "learning_rate": 1.807968748078764e-05, "loss": 0.1429, "step": 1960 }, { "epoch": 1.8, "learning_rate": 1.8077500406701448e-05, "loss": 0.1447, "step": 1961 }, { "epoch": 1.8, "learning_rate": 1.80753122203197e-05, "loss": 0.1418, "step": 1962 }, { "epoch": 1.8, "learning_rate": 1.807312292194372e-05, "loss": 0.1383, "step": 1963 }, { "epoch": 1.8, "learning_rate": 1.807093251187498e-05, "loss": 0.1541, "step": 1964 }, { "epoch": 1.8, "learning_rate": 1.8068740990415105e-05, "loss": 0.1517, "step": 1965 }, { "epoch": 1.8, "learning_rate": 1.8066548357865876e-05, "loss": 0.1612, "step": 1966 }, { "epoch": 1.8, "learning_rate": 1.8064354614529224e-05, "loss": 0.1432, "step": 1967 }, { "epoch": 1.8, "learning_rate": 1.8062159760707235e-05, "loss": 0.154, "step": 1968 }, { "epoch": 1.8, "learning_rate": 1.8059963796702143e-05, "loss": 0.1528, "step": 1969 }, { "epoch": 1.81, "learning_rate": 1.805776672281634e-05, "loss": 0.1451, "step": 1970 }, { "epoch": 1.81, "learning_rate": 1.8055568539352376e-05, "loss": 0.137, "step": 1971 }, { "epoch": 1.81, "learning_rate": 1.805336924661294e-05, "loss": 0.1493, "step": 1972 }, { "epoch": 1.81, "learning_rate": 1.805116884490088e-05, "loss": 0.1553, "step": 1973 }, { "epoch": 1.81, "learning_rate": 1.8048967334519202e-05, "loss": 0.1496, "step": 1974 }, { "epoch": 1.81, "learning_rate": 1.804676471577106e-05, "loss": 0.1452, "step": 1975 }, { "epoch": 1.81, "learning_rate": 1.8044560988959766e-05, "loss": 0.1471, "step": 1976 }, { "epoch": 1.81, "learning_rate": 1.8042356154388768e-05, "loss": 0.1492, "step": 1977 }, { "epoch": 1.81, "learning_rate": 1.8040150212361687e-05, "loss": 0.1497, "step": 1978 }, { "epoch": 1.81, "learning_rate": 1.8037943163182283e-05, "loss": 0.1327, "step": 1979 }, { "epoch": 1.81, "learning_rate": 1.8035735007154477e-05, "loss": 0.1456, "step": 1980 }, { "epoch": 1.82, "learning_rate": 1.8033525744582335e-05, "loss": 0.1449, "step": 1981 }, { "epoch": 1.82, "learning_rate": 1.8031315375770086e-05, "loss": 0.1541, "step": 1982 }, { "epoch": 1.82, "learning_rate": 1.802910390102209e-05, "loss": 0.1501, "step": 1983 }, { "epoch": 1.82, "learning_rate": 1.802689132064289e-05, "loss": 0.1418, "step": 1984 }, { "epoch": 1.82, "learning_rate": 1.8024677634937155e-05, "loss": 0.1456, "step": 1985 }, { "epoch": 1.82, "learning_rate": 1.802246284420972e-05, "loss": 0.1481, "step": 1986 }, { "epoch": 1.82, "learning_rate": 1.8020246948765565e-05, "loss": 0.1474, "step": 1987 }, { "epoch": 1.82, "learning_rate": 1.8018029948909827e-05, "loss": 0.1413, "step": 1988 }, { "epoch": 1.82, "learning_rate": 1.8015811844947794e-05, "loss": 0.1519, "step": 1989 }, { "epoch": 1.82, "learning_rate": 1.8013592637184903e-05, "loss": 0.1538, "step": 1990 }, { "epoch": 1.82, "learning_rate": 1.8011372325926753e-05, "loss": 0.1534, "step": 1991 }, { "epoch": 1.83, "learning_rate": 1.8009150911479077e-05, "loss": 0.1513, "step": 1992 }, { "epoch": 1.83, "learning_rate": 1.8006928394147777e-05, "loss": 0.1397, "step": 1993 }, { "epoch": 1.83, "learning_rate": 1.8004704774238893e-05, "loss": 0.1442, "step": 1994 }, { "epoch": 1.83, "learning_rate": 1.8002480052058636e-05, "loss": 0.1548, "step": 1995 }, { "epoch": 1.83, "learning_rate": 1.8000254227913346e-05, "loss": 0.1431, "step": 1996 }, { "epoch": 1.83, "learning_rate": 1.799802730210953e-05, "loss": 0.1409, "step": 1997 }, { "epoch": 1.83, "learning_rate": 1.7995799274953842e-05, "loss": 0.1469, "step": 1998 }, { "epoch": 1.83, "learning_rate": 1.7993570146753088e-05, "loss": 0.1449, "step": 1999 }, { "epoch": 1.83, "learning_rate": 1.7991339917814227e-05, "loss": 0.1369, "step": 2000 }, { "epoch": 1.83, "learning_rate": 1.7989108588444365e-05, "loss": 0.1457, "step": 2001 }, { "epoch": 1.84, "learning_rate": 1.798687615895076e-05, "loss": 0.1449, "step": 2002 }, { "epoch": 1.84, "learning_rate": 1.7984642629640832e-05, "loss": 0.1463, "step": 2003 }, { "epoch": 1.84, "learning_rate": 1.7982408000822138e-05, "loss": 0.1449, "step": 2004 }, { "epoch": 1.84, "learning_rate": 1.7980172272802398e-05, "loss": 0.1435, "step": 2005 }, { "epoch": 1.84, "learning_rate": 1.7977935445889475e-05, "loss": 0.1357, "step": 2006 }, { "epoch": 1.84, "learning_rate": 1.7975697520391384e-05, "loss": 0.1409, "step": 2007 }, { "epoch": 1.84, "learning_rate": 1.79734584966163e-05, "loss": 0.1449, "step": 2008 }, { "epoch": 1.84, "learning_rate": 1.797121837487254e-05, "loss": 0.148, "step": 2009 }, { "epoch": 1.84, "learning_rate": 1.7968977155468575e-05, "loss": 0.1554, "step": 2010 }, { "epoch": 1.84, "learning_rate": 1.7966734838713032e-05, "loss": 0.154, "step": 2011 }, { "epoch": 1.84, "learning_rate": 1.7964491424914677e-05, "loss": 0.1459, "step": 2012 }, { "epoch": 1.85, "learning_rate": 1.796224691438244e-05, "loss": 0.1433, "step": 2013 }, { "epoch": 1.85, "learning_rate": 1.7960001307425394e-05, "loss": 0.1438, "step": 2014 }, { "epoch": 1.85, "learning_rate": 1.795775460435277e-05, "loss": 0.1363, "step": 2015 }, { "epoch": 1.85, "learning_rate": 1.795550680547394e-05, "loss": 0.1431, "step": 2016 }, { "epoch": 1.85, "learning_rate": 1.7953257911098435e-05, "loss": 0.1484, "step": 2017 }, { "epoch": 1.85, "learning_rate": 1.7951007921535936e-05, "loss": 0.1453, "step": 2018 }, { "epoch": 1.85, "learning_rate": 1.794875683709627e-05, "loss": 0.1422, "step": 2019 }, { "epoch": 1.85, "learning_rate": 1.794650465808942e-05, "loss": 0.1424, "step": 2020 }, { "epoch": 1.85, "learning_rate": 1.7944251384825516e-05, "loss": 0.1556, "step": 2021 }, { "epoch": 1.85, "learning_rate": 1.7941997017614843e-05, "loss": 0.1468, "step": 2022 }, { "epoch": 1.85, "learning_rate": 1.7939741556767837e-05, "loss": 0.1359, "step": 2023 }, { "epoch": 1.86, "learning_rate": 1.7937485002595073e-05, "loss": 0.1368, "step": 2024 }, { "epoch": 1.86, "learning_rate": 1.7935227355407287e-05, "loss": 0.1544, "step": 2025 }, { "epoch": 1.86, "learning_rate": 1.7932968615515368e-05, "loss": 0.1465, "step": 2026 }, { "epoch": 1.86, "learning_rate": 1.793070878323035e-05, "loss": 0.1354, "step": 2027 }, { "epoch": 1.86, "learning_rate": 1.7928447858863417e-05, "loss": 0.1373, "step": 2028 }, { "epoch": 1.86, "learning_rate": 1.7926185842725904e-05, "loss": 0.1439, "step": 2029 }, { "epoch": 1.86, "learning_rate": 1.7923922735129303e-05, "loss": 0.1448, "step": 2030 }, { "epoch": 1.86, "learning_rate": 1.7921658536385244e-05, "loss": 0.148, "step": 2031 }, { "epoch": 1.86, "learning_rate": 1.7919393246805513e-05, "loss": 0.1365, "step": 2032 }, { "epoch": 1.86, "learning_rate": 1.791712686670205e-05, "loss": 0.1503, "step": 2033 }, { "epoch": 1.86, "learning_rate": 1.7914859396386948e-05, "loss": 0.1454, "step": 2034 }, { "epoch": 1.87, "learning_rate": 1.7912590836172438e-05, "loss": 0.1446, "step": 2035 }, { "epoch": 1.87, "learning_rate": 1.7910321186370905e-05, "loss": 0.1576, "step": 2036 }, { "epoch": 1.87, "learning_rate": 1.7908050447294893e-05, "loss": 0.139, "step": 2037 }, { "epoch": 1.87, "learning_rate": 1.7905778619257087e-05, "loss": 0.1469, "step": 2038 }, { "epoch": 1.87, "learning_rate": 1.790350570257032e-05, "loss": 0.1495, "step": 2039 }, { "epoch": 1.87, "learning_rate": 1.7901231697547582e-05, "loss": 0.1405, "step": 2040 }, { "epoch": 1.87, "learning_rate": 1.7898956604502012e-05, "loss": 0.1454, "step": 2041 }, { "epoch": 1.87, "learning_rate": 1.78966804237469e-05, "loss": 0.1392, "step": 2042 }, { "epoch": 1.87, "learning_rate": 1.7894403155595674e-05, "loss": 0.1523, "step": 2043 }, { "epoch": 1.87, "learning_rate": 1.7892124800361927e-05, "loss": 0.1405, "step": 2044 }, { "epoch": 1.87, "learning_rate": 1.7889845358359395e-05, "loss": 0.1364, "step": 2045 }, { "epoch": 1.88, "learning_rate": 1.788756482990196e-05, "loss": 0.131, "step": 2046 }, { "epoch": 1.88, "learning_rate": 1.788528321530366e-05, "loss": 0.1513, "step": 2047 }, { "epoch": 1.88, "learning_rate": 1.788300051487868e-05, "loss": 0.1447, "step": 2048 }, { "epoch": 1.88, "learning_rate": 1.7880716728941355e-05, "loss": 0.1481, "step": 2049 }, { "epoch": 1.88, "learning_rate": 1.787843185780617e-05, "loss": 0.1404, "step": 2050 }, { "epoch": 1.88, "learning_rate": 1.7876145901787755e-05, "loss": 0.1349, "step": 2051 }, { "epoch": 1.88, "learning_rate": 1.7873858861200895e-05, "loss": 0.1498, "step": 2052 }, { "epoch": 1.88, "learning_rate": 1.787157073636052e-05, "loss": 0.1492, "step": 2053 }, { "epoch": 1.88, "learning_rate": 1.7869281527581718e-05, "loss": 0.1446, "step": 2054 }, { "epoch": 1.88, "learning_rate": 1.7866991235179713e-05, "loss": 0.1495, "step": 2055 }, { "epoch": 1.88, "learning_rate": 1.7864699859469887e-05, "loss": 0.1435, "step": 2056 }, { "epoch": 1.89, "learning_rate": 1.786240740076777e-05, "loss": 0.136, "step": 2057 }, { "epoch": 1.89, "learning_rate": 1.7860113859389038e-05, "loss": 0.142, "step": 2058 }, { "epoch": 1.89, "learning_rate": 1.7857819235649525e-05, "loss": 0.1442, "step": 2059 }, { "epoch": 1.89, "learning_rate": 1.78555235298652e-05, "loss": 0.1497, "step": 2060 }, { "epoch": 1.89, "learning_rate": 1.785322674235219e-05, "loss": 0.1506, "step": 2061 }, { "epoch": 1.89, "learning_rate": 1.785092887342677e-05, "loss": 0.1546, "step": 2062 }, { "epoch": 1.89, "learning_rate": 1.7848629923405364e-05, "loss": 0.1418, "step": 2063 }, { "epoch": 1.89, "learning_rate": 1.7846329892604548e-05, "loss": 0.1579, "step": 2064 }, { "epoch": 1.89, "learning_rate": 1.7844028781341033e-05, "loss": 0.1435, "step": 2065 }, { "epoch": 1.89, "learning_rate": 1.78417265899317e-05, "loss": 0.1434, "step": 2066 }, { "epoch": 1.89, "learning_rate": 1.7839423318693558e-05, "loss": 0.1472, "step": 2067 }, { "epoch": 1.9, "learning_rate": 1.783711896794378e-05, "loss": 0.1395, "step": 2068 }, { "epoch": 1.9, "learning_rate": 1.783481353799968e-05, "loss": 0.1439, "step": 2069 }, { "epoch": 1.9, "learning_rate": 1.7832507029178725e-05, "loss": 0.1465, "step": 2070 }, { "epoch": 1.9, "learning_rate": 1.7830199441798526e-05, "loss": 0.1506, "step": 2071 }, { "epoch": 1.9, "learning_rate": 1.782789077617684e-05, "loss": 0.1436, "step": 2072 }, { "epoch": 1.9, "learning_rate": 1.7825581032631584e-05, "loss": 0.1433, "step": 2073 }, { "epoch": 1.9, "learning_rate": 1.7823270211480817e-05, "loss": 0.145, "step": 2074 }, { "epoch": 1.9, "learning_rate": 1.782095831304274e-05, "loss": 0.1423, "step": 2075 }, { "epoch": 1.9, "learning_rate": 1.781864533763571e-05, "loss": 0.1395, "step": 2076 }, { "epoch": 1.9, "learning_rate": 1.781633128557823e-05, "loss": 0.1467, "step": 2077 }, { "epoch": 1.9, "learning_rate": 1.781401615718896e-05, "loss": 0.1409, "step": 2078 }, { "epoch": 1.91, "learning_rate": 1.7811699952786686e-05, "loss": 0.1377, "step": 2079 }, { "epoch": 1.91, "learning_rate": 1.7809382672690368e-05, "loss": 0.1469, "step": 2080 }, { "epoch": 1.91, "learning_rate": 1.7807064317219096e-05, "loss": 0.1437, "step": 2081 }, { "epoch": 1.91, "learning_rate": 1.7804744886692117e-05, "loss": 0.1447, "step": 2082 }, { "epoch": 1.91, "learning_rate": 1.7802424381428823e-05, "loss": 0.1423, "step": 2083 }, { "epoch": 1.91, "learning_rate": 1.7800102801748755e-05, "loss": 0.1398, "step": 2084 }, { "epoch": 1.91, "learning_rate": 1.7797780147971594e-05, "loss": 0.1343, "step": 2085 }, { "epoch": 1.91, "learning_rate": 1.779545642041719e-05, "loss": 0.1495, "step": 2086 }, { "epoch": 1.91, "learning_rate": 1.7793131619405513e-05, "loss": 0.143, "step": 2087 }, { "epoch": 1.91, "learning_rate": 1.7790805745256703e-05, "loss": 0.1337, "step": 2088 }, { "epoch": 1.91, "learning_rate": 1.7788478798291043e-05, "loss": 0.1402, "step": 2089 }, { "epoch": 1.92, "learning_rate": 1.7786150778828952e-05, "loss": 0.1377, "step": 2090 }, { "epoch": 1.92, "learning_rate": 1.778382168719101e-05, "loss": 0.1456, "step": 2091 }, { "epoch": 1.92, "learning_rate": 1.7781491523697938e-05, "loss": 0.153, "step": 2092 }, { "epoch": 1.92, "learning_rate": 1.777916028867061e-05, "loss": 0.1503, "step": 2093 }, { "epoch": 1.92, "learning_rate": 1.7776827982430035e-05, "loss": 0.1518, "step": 2094 }, { "epoch": 1.92, "learning_rate": 1.777449460529739e-05, "loss": 0.1428, "step": 2095 }, { "epoch": 1.92, "learning_rate": 1.7772160157593976e-05, "loss": 0.1411, "step": 2096 }, { "epoch": 1.92, "learning_rate": 1.7769824639641266e-05, "loss": 0.1365, "step": 2097 }, { "epoch": 1.92, "learning_rate": 1.7767488051760858e-05, "loss": 0.1446, "step": 2098 }, { "epoch": 1.92, "learning_rate": 1.776515039427451e-05, "loss": 0.1468, "step": 2099 }, { "epoch": 1.92, "learning_rate": 1.7762811667504123e-05, "loss": 0.1498, "step": 2100 }, { "epoch": 1.93, "learning_rate": 1.7760471871771752e-05, "loss": 0.1487, "step": 2101 }, { "epoch": 1.93, "learning_rate": 1.7758131007399585e-05, "loss": 0.1365, "step": 2102 }, { "epoch": 1.93, "learning_rate": 1.7755789074709973e-05, "loss": 0.1458, "step": 2103 }, { "epoch": 1.93, "learning_rate": 1.775344607402541e-05, "loss": 0.1474, "step": 2104 }, { "epoch": 1.93, "learning_rate": 1.7751102005668523e-05, "loss": 0.1482, "step": 2105 }, { "epoch": 1.93, "learning_rate": 1.7748756869962104e-05, "loss": 0.1371, "step": 2106 }, { "epoch": 1.93, "learning_rate": 1.7746410667229083e-05, "loss": 0.1429, "step": 2107 }, { "epoch": 1.93, "learning_rate": 1.774406339779254e-05, "loss": 0.147, "step": 2108 }, { "epoch": 1.93, "learning_rate": 1.77417150619757e-05, "loss": 0.1404, "step": 2109 }, { "epoch": 1.93, "learning_rate": 1.773936566010194e-05, "loss": 0.1516, "step": 2110 }, { "epoch": 1.93, "learning_rate": 1.7737015192494777e-05, "loss": 0.1431, "step": 2111 }, { "epoch": 1.94, "learning_rate": 1.7734663659477874e-05, "loss": 0.1436, "step": 2112 }, { "epoch": 1.94, "learning_rate": 1.7732311061375048e-05, "loss": 0.141, "step": 2113 }, { "epoch": 1.94, "learning_rate": 1.7729957398510252e-05, "loss": 0.139, "step": 2114 }, { "epoch": 1.94, "learning_rate": 1.7727602671207605e-05, "loss": 0.1404, "step": 2115 }, { "epoch": 1.94, "learning_rate": 1.772524687979135e-05, "loss": 0.1386, "step": 2116 }, { "epoch": 1.94, "learning_rate": 1.772289002458589e-05, "loss": 0.1494, "step": 2117 }, { "epoch": 1.94, "learning_rate": 1.7720532105915763e-05, "loss": 0.1515, "step": 2118 }, { "epoch": 1.94, "learning_rate": 1.7718173124105674e-05, "loss": 0.1507, "step": 2119 }, { "epoch": 1.94, "learning_rate": 1.7715813079480452e-05, "loss": 0.1442, "step": 2120 }, { "epoch": 1.94, "learning_rate": 1.7713451972365087e-05, "loss": 0.1351, "step": 2121 }, { "epoch": 1.95, "learning_rate": 1.7711089803084714e-05, "loss": 0.1566, "step": 2122 }, { "epoch": 1.95, "learning_rate": 1.7708726571964598e-05, "loss": 0.1441, "step": 2123 }, { "epoch": 1.95, "learning_rate": 1.7706362279330175e-05, "loss": 0.1442, "step": 2124 }, { "epoch": 1.95, "learning_rate": 1.7703996925507014e-05, "loss": 0.146, "step": 2125 }, { "epoch": 1.95, "learning_rate": 1.7701630510820826e-05, "loss": 0.1437, "step": 2126 }, { "epoch": 1.95, "learning_rate": 1.769926303559747e-05, "loss": 0.1464, "step": 2127 }, { "epoch": 1.95, "learning_rate": 1.7696894500162964e-05, "loss": 0.1437, "step": 2128 }, { "epoch": 1.95, "learning_rate": 1.7694524904843455e-05, "loss": 0.1367, "step": 2129 }, { "epoch": 1.95, "learning_rate": 1.7692154249965248e-05, "loss": 0.1391, "step": 2130 }, { "epoch": 1.95, "learning_rate": 1.7689782535854786e-05, "loss": 0.1523, "step": 2131 }, { "epoch": 1.95, "learning_rate": 1.7687409762838666e-05, "loss": 0.14, "step": 2132 }, { "epoch": 1.96, "learning_rate": 1.7685035931243618e-05, "loss": 0.1473, "step": 2133 }, { "epoch": 1.96, "learning_rate": 1.7682661041396532e-05, "loss": 0.1374, "step": 2134 }, { "epoch": 1.96, "learning_rate": 1.7680285093624435e-05, "loss": 0.152, "step": 2135 }, { "epoch": 1.96, "learning_rate": 1.7677908088254498e-05, "loss": 0.1385, "step": 2136 }, { "epoch": 1.96, "learning_rate": 1.767553002561405e-05, "loss": 0.1373, "step": 2137 }, { "epoch": 1.96, "learning_rate": 1.767315090603055e-05, "loss": 0.1451, "step": 2138 }, { "epoch": 1.96, "learning_rate": 1.7670770729831616e-05, "loss": 0.1415, "step": 2139 }, { "epoch": 1.96, "learning_rate": 1.7668389497345e-05, "loss": 0.1358, "step": 2140 }, { "epoch": 1.96, "learning_rate": 1.7666007208898602e-05, "loss": 0.1376, "step": 2141 }, { "epoch": 1.96, "learning_rate": 1.7663623864820478e-05, "loss": 0.1424, "step": 2142 }, { "epoch": 1.96, "learning_rate": 1.766123946543882e-05, "loss": 0.135, "step": 2143 }, { "epoch": 1.97, "learning_rate": 1.765885401108196e-05, "loss": 0.1337, "step": 2144 }, { "epoch": 1.97, "learning_rate": 1.7656467502078392e-05, "loss": 0.1603, "step": 2145 }, { "epoch": 1.97, "learning_rate": 1.765407993875674e-05, "loss": 0.1459, "step": 2146 }, { "epoch": 1.97, "learning_rate": 1.765169132144578e-05, "loss": 0.145, "step": 2147 }, { "epoch": 1.97, "learning_rate": 1.7649301650474425e-05, "loss": 0.1504, "step": 2148 }, { "epoch": 1.97, "learning_rate": 1.7646910926171747e-05, "loss": 0.1421, "step": 2149 }, { "epoch": 1.97, "learning_rate": 1.7644519148866957e-05, "loss": 0.1402, "step": 2150 }, { "epoch": 1.97, "learning_rate": 1.76421263188894e-05, "loss": 0.1514, "step": 2151 }, { "epoch": 1.97, "learning_rate": 1.7639732436568588e-05, "loss": 0.1456, "step": 2152 }, { "epoch": 1.97, "learning_rate": 1.7637337502234162e-05, "loss": 0.1493, "step": 2153 }, { "epoch": 1.97, "learning_rate": 1.7634941516215903e-05, "loss": 0.1478, "step": 2154 }, { "epoch": 1.98, "learning_rate": 1.7632544478843756e-05, "loss": 0.1449, "step": 2155 }, { "epoch": 1.98, "learning_rate": 1.7630146390447794e-05, "loss": 0.1534, "step": 2156 }, { "epoch": 1.98, "learning_rate": 1.7627747251358246e-05, "loss": 0.1469, "step": 2157 }, { "epoch": 1.98, "learning_rate": 1.7625347061905475e-05, "loss": 0.1449, "step": 2158 }, { "epoch": 1.98, "learning_rate": 1.7622945822419996e-05, "loss": 0.145, "step": 2159 }, { "epoch": 1.98, "learning_rate": 1.7620543533232468e-05, "loss": 0.1492, "step": 2160 }, { "epoch": 1.98, "learning_rate": 1.761814019467369e-05, "loss": 0.1468, "step": 2161 }, { "epoch": 1.98, "learning_rate": 1.7615735807074616e-05, "loss": 0.1461, "step": 2162 }, { "epoch": 1.98, "learning_rate": 1.7613330370766326e-05, "loss": 0.1537, "step": 2163 }, { "epoch": 1.98, "learning_rate": 1.7610923886080066e-05, "loss": 0.1415, "step": 2164 }, { "epoch": 1.98, "learning_rate": 1.760851635334721e-05, "loss": 0.1418, "step": 2165 }, { "epoch": 1.99, "learning_rate": 1.760610777289929e-05, "loss": 0.1408, "step": 2166 }, { "epoch": 1.99, "learning_rate": 1.7603698145067962e-05, "loss": 0.1519, "step": 2167 }, { "epoch": 1.99, "learning_rate": 1.760128747018505e-05, "loss": 0.1438, "step": 2168 }, { "epoch": 1.99, "learning_rate": 1.7598875748582504e-05, "loss": 0.1376, "step": 2169 }, { "epoch": 1.99, "learning_rate": 1.759646298059243e-05, "loss": 0.1406, "step": 2170 }, { "epoch": 1.99, "learning_rate": 1.7594049166547073e-05, "loss": 0.1412, "step": 2171 }, { "epoch": 1.99, "learning_rate": 1.759163430677882e-05, "loss": 0.1408, "step": 2172 }, { "epoch": 1.99, "learning_rate": 1.75892184016202e-05, "loss": 0.1509, "step": 2173 }, { "epoch": 1.99, "learning_rate": 1.75868014514039e-05, "loss": 0.1376, "step": 2174 }, { "epoch": 1.99, "learning_rate": 1.7584383456462736e-05, "loss": 0.1385, "step": 2175 }, { "epoch": 1.99, "learning_rate": 1.7581964417129673e-05, "loss": 0.1439, "step": 2176 }, { "epoch": 2.0, "learning_rate": 1.7579544333737815e-05, "loss": 0.1295, "step": 2177 }, { "epoch": 2.0, "learning_rate": 1.7577123206620427e-05, "loss": 0.139, "step": 2178 }, { "epoch": 2.0, "learning_rate": 1.7574701036110894e-05, "loss": 0.1497, "step": 2179 }, { "epoch": 2.0, "learning_rate": 1.7572277822542763e-05, "loss": 0.1471, "step": 2180 }, { "epoch": 2.0, "learning_rate": 1.756985356624971e-05, "loss": 0.1373, "step": 2181 }, { "epoch": 2.0, "learning_rate": 1.7567428267565572e-05, "loss": 0.1406, "step": 2182 }, { "epoch": 2.0, "learning_rate": 1.7565001926824313e-05, "loss": 0.1324, "step": 2183 }, { "epoch": 2.0, "learning_rate": 1.7562574544360048e-05, "loss": 0.1406, "step": 2184 }, { "epoch": 2.0, "learning_rate": 1.7560146120507036e-05, "loss": 0.1395, "step": 2185 }, { "epoch": 2.0, "learning_rate": 1.7557716655599676e-05, "loss": 0.153, "step": 2186 }, { "epoch": 2.0, "learning_rate": 1.7555286149972516e-05, "loss": 0.1453, "step": 2187 }, { "epoch": 2.01, "learning_rate": 1.755285460396024e-05, "loss": 0.1305, "step": 2188 }, { "epoch": 2.01, "learning_rate": 1.7550422017897685e-05, "loss": 0.1431, "step": 2189 }, { "epoch": 2.01, "learning_rate": 1.7547988392119818e-05, "loss": 0.1398, "step": 2190 }, { "epoch": 2.01, "learning_rate": 1.7545553726961763e-05, "loss": 0.14, "step": 2191 }, { "epoch": 2.01, "learning_rate": 1.7543118022758773e-05, "loss": 0.1331, "step": 2192 }, { "epoch": 2.01, "learning_rate": 1.7540681279846253e-05, "loss": 0.139, "step": 2193 }, { "epoch": 2.01, "learning_rate": 1.753824349855976e-05, "loss": 0.128, "step": 2194 }, { "epoch": 2.01, "learning_rate": 1.7535804679234975e-05, "loss": 0.1457, "step": 2195 }, { "epoch": 2.01, "learning_rate": 1.7533364822207728e-05, "loss": 0.1437, "step": 2196 }, { "epoch": 2.01, "learning_rate": 1.7530923927814e-05, "loss": 0.1487, "step": 2197 }, { "epoch": 2.01, "learning_rate": 1.7528481996389906e-05, "loss": 0.1342, "step": 2198 }, { "epoch": 2.02, "learning_rate": 1.752603902827171e-05, "loss": 0.1385, "step": 2199 }, { "epoch": 2.02, "learning_rate": 1.7523595023795814e-05, "loss": 0.1387, "step": 2200 }, { "epoch": 2.02, "learning_rate": 1.7521149983298764e-05, "loss": 0.1417, "step": 2201 }, { "epoch": 2.02, "learning_rate": 1.7518703907117253e-05, "loss": 0.1321, "step": 2202 }, { "epoch": 2.02, "learning_rate": 1.751625679558811e-05, "loss": 0.1351, "step": 2203 }, { "epoch": 2.02, "learning_rate": 1.751380864904831e-05, "loss": 0.1394, "step": 2204 }, { "epoch": 2.02, "learning_rate": 1.7511359467834966e-05, "loss": 0.1379, "step": 2205 }, { "epoch": 2.02, "learning_rate": 1.750890925228535e-05, "loss": 0.1336, "step": 2206 }, { "epoch": 2.02, "learning_rate": 1.7506458002736846e-05, "loss": 0.1456, "step": 2207 }, { "epoch": 2.02, "learning_rate": 1.7504005719527012e-05, "loss": 0.1725, "step": 2208 }, { "epoch": 2.02, "learning_rate": 1.7501552402993523e-05, "loss": 0.1707, "step": 2209 }, { "epoch": 2.03, "learning_rate": 1.7499098053474224e-05, "loss": 0.1428, "step": 2210 }, { "epoch": 2.03, "learning_rate": 1.7496642671307075e-05, "loss": 0.1421, "step": 2211 }, { "epoch": 2.03, "learning_rate": 1.749418625683019e-05, "loss": 0.1424, "step": 2212 }, { "epoch": 2.03, "learning_rate": 1.7491728810381822e-05, "loss": 0.158, "step": 2213 }, { "epoch": 2.03, "learning_rate": 1.7489270332300376e-05, "loss": 0.1469, "step": 2214 }, { "epoch": 2.03, "learning_rate": 1.7486810822924388e-05, "loss": 0.1395, "step": 2215 }, { "epoch": 2.03, "learning_rate": 1.748435028259254e-05, "loss": 0.1448, "step": 2216 }, { "epoch": 2.03, "learning_rate": 1.7481888711643655e-05, "loss": 0.1461, "step": 2217 }, { "epoch": 2.03, "learning_rate": 1.74794261104167e-05, "loss": 0.1427, "step": 2218 }, { "epoch": 2.03, "learning_rate": 1.7476962479250782e-05, "loss": 0.1477, "step": 2219 }, { "epoch": 2.03, "learning_rate": 1.7474497818485147e-05, "loss": 0.1501, "step": 2220 }, { "epoch": 2.04, "learning_rate": 1.747203212845919e-05, "loss": 0.1424, "step": 2221 }, { "epoch": 2.04, "learning_rate": 1.746956540951244e-05, "loss": 0.1402, "step": 2222 }, { "epoch": 2.04, "learning_rate": 1.746709766198458e-05, "loss": 0.1268, "step": 2223 }, { "epoch": 2.04, "learning_rate": 1.7464628886215416e-05, "loss": 0.1459, "step": 2224 }, { "epoch": 2.04, "learning_rate": 1.7462159082544907e-05, "loss": 0.1407, "step": 2225 }, { "epoch": 2.04, "learning_rate": 1.745968825131316e-05, "loss": 0.1487, "step": 2226 }, { "epoch": 2.04, "learning_rate": 1.7457216392860406e-05, "loss": 0.1466, "step": 2227 }, { "epoch": 2.04, "learning_rate": 1.7454743507527036e-05, "loss": 0.1454, "step": 2228 }, { "epoch": 2.04, "learning_rate": 1.7452269595653566e-05, "loss": 0.149, "step": 2229 }, { "epoch": 2.04, "learning_rate": 1.7449794657580666e-05, "loss": 0.1418, "step": 2230 }, { "epoch": 2.04, "learning_rate": 1.744731869364914e-05, "loss": 0.1369, "step": 2231 }, { "epoch": 2.05, "learning_rate": 1.7444841704199934e-05, "loss": 0.1392, "step": 2232 }, { "epoch": 2.05, "learning_rate": 1.7442363689574143e-05, "loss": 0.135, "step": 2233 }, { "epoch": 2.05, "learning_rate": 1.743988465011299e-05, "loss": 0.1371, "step": 2234 }, { "epoch": 2.05, "learning_rate": 1.743740458615785e-05, "loss": 0.1517, "step": 2235 }, { "epoch": 2.05, "learning_rate": 1.7434923498050234e-05, "loss": 0.1394, "step": 2236 }, { "epoch": 2.05, "learning_rate": 1.7432441386131792e-05, "loss": 0.139, "step": 2237 }, { "epoch": 2.05, "learning_rate": 1.7429958250744327e-05, "loss": 0.1397, "step": 2238 }, { "epoch": 2.05, "learning_rate": 1.7427474092229763e-05, "loss": 0.1324, "step": 2239 }, { "epoch": 2.05, "learning_rate": 1.742498891093018e-05, "loss": 0.1395, "step": 2240 }, { "epoch": 2.05, "learning_rate": 1.7422502707187798e-05, "loss": 0.1523, "step": 2241 }, { "epoch": 2.05, "learning_rate": 1.7420015481344972e-05, "loss": 0.1415, "step": 2242 }, { "epoch": 2.06, "learning_rate": 1.74175272337442e-05, "loss": 0.151, "step": 2243 }, { "epoch": 2.06, "learning_rate": 1.7415037964728124e-05, "loss": 0.1466, "step": 2244 }, { "epoch": 2.06, "learning_rate": 1.741254767463952e-05, "loss": 0.1449, "step": 2245 }, { "epoch": 2.06, "learning_rate": 1.741005636382131e-05, "loss": 0.1393, "step": 2246 }, { "epoch": 2.06, "learning_rate": 1.7407564032616556e-05, "loss": 0.1356, "step": 2247 }, { "epoch": 2.06, "learning_rate": 1.7405070681368456e-05, "loss": 0.148, "step": 2248 }, { "epoch": 2.06, "learning_rate": 1.7402576310420353e-05, "loss": 0.1348, "step": 2249 }, { "epoch": 2.06, "learning_rate": 1.740008092011573e-05, "loss": 0.1329, "step": 2250 }, { "epoch": 2.06, "learning_rate": 1.7397584510798208e-05, "loss": 0.1292, "step": 2251 }, { "epoch": 2.06, "learning_rate": 1.7395087082811555e-05, "loss": 0.1277, "step": 2252 }, { "epoch": 2.07, "learning_rate": 1.7392588636499666e-05, "loss": 0.1335, "step": 2253 }, { "epoch": 2.07, "learning_rate": 1.7390089172206594e-05, "loss": 0.134, "step": 2254 }, { "epoch": 2.07, "learning_rate": 1.7387588690276514e-05, "loss": 0.1436, "step": 2255 }, { "epoch": 2.07, "learning_rate": 1.7385087191053755e-05, "loss": 0.1268, "step": 2256 }, { "epoch": 2.07, "learning_rate": 1.7382584674882776e-05, "loss": 0.135, "step": 2257 }, { "epoch": 2.07, "learning_rate": 1.7380081142108186e-05, "loss": 0.1386, "step": 2258 }, { "epoch": 2.07, "learning_rate": 1.7377576593074725e-05, "loss": 0.1361, "step": 2259 }, { "epoch": 2.07, "learning_rate": 1.7375071028127277e-05, "loss": 0.1427, "step": 2260 }, { "epoch": 2.07, "learning_rate": 1.737256444761087e-05, "loss": 0.1375, "step": 2261 }, { "epoch": 2.07, "learning_rate": 1.7370056851870665e-05, "loss": 0.1292, "step": 2262 }, { "epoch": 2.07, "learning_rate": 1.7367548241251962e-05, "loss": 0.1342, "step": 2263 }, { "epoch": 2.08, "learning_rate": 1.736503861610021e-05, "loss": 0.1388, "step": 2264 }, { "epoch": 2.08, "learning_rate": 1.7362527976760985e-05, "loss": 0.1414, "step": 2265 }, { "epoch": 2.08, "learning_rate": 1.7360016323580017e-05, "loss": 0.1395, "step": 2266 }, { "epoch": 2.08, "learning_rate": 1.735750365690316e-05, "loss": 0.1306, "step": 2267 }, { "epoch": 2.08, "learning_rate": 1.735498997707642e-05, "loss": 0.1394, "step": 2268 }, { "epoch": 2.08, "learning_rate": 1.735247528444594e-05, "loss": 0.1461, "step": 2269 }, { "epoch": 2.08, "learning_rate": 1.7349959579358003e-05, "loss": 0.1378, "step": 2270 }, { "epoch": 2.08, "learning_rate": 1.734744286215902e-05, "loss": 0.1325, "step": 2271 }, { "epoch": 2.08, "learning_rate": 1.734492513319555e-05, "loss": 0.1331, "step": 2272 }, { "epoch": 2.08, "learning_rate": 1.73424063928143e-05, "loss": 0.1358, "step": 2273 }, { "epoch": 2.08, "learning_rate": 1.7339886641362107e-05, "loss": 0.1375, "step": 2274 }, { "epoch": 2.09, "learning_rate": 1.733736587918594e-05, "loss": 0.1414, "step": 2275 }, { "epoch": 2.09, "learning_rate": 1.7334844106632926e-05, "loss": 0.1527, "step": 2276 }, { "epoch": 2.09, "learning_rate": 1.733232132405031e-05, "loss": 0.1365, "step": 2277 }, { "epoch": 2.09, "learning_rate": 1.73297975317855e-05, "loss": 0.1408, "step": 2278 }, { "epoch": 2.09, "learning_rate": 1.7327272730186012e-05, "loss": 0.1374, "step": 2279 }, { "epoch": 2.09, "learning_rate": 1.7324746919599537e-05, "loss": 0.1451, "step": 2280 }, { "epoch": 2.09, "learning_rate": 1.732222010037387e-05, "loss": 0.139, "step": 2281 }, { "epoch": 2.09, "learning_rate": 1.7319692272856974e-05, "loss": 0.1482, "step": 2282 }, { "epoch": 2.09, "learning_rate": 1.7317163437396934e-05, "loss": 0.1325, "step": 2283 }, { "epoch": 2.09, "learning_rate": 1.7314633594341974e-05, "loss": 0.1403, "step": 2284 }, { "epoch": 2.09, "learning_rate": 1.7312102744040465e-05, "loss": 0.1327, "step": 2285 }, { "epoch": 2.1, "learning_rate": 1.7309570886840913e-05, "loss": 0.1343, "step": 2286 }, { "epoch": 2.1, "learning_rate": 1.730703802309196e-05, "loss": 0.1398, "step": 2287 }, { "epoch": 2.1, "learning_rate": 1.7304504153142392e-05, "loss": 0.13, "step": 2288 }, { "epoch": 2.1, "learning_rate": 1.7301969277341125e-05, "loss": 0.1416, "step": 2289 }, { "epoch": 2.1, "learning_rate": 1.7299433396037223e-05, "loss": 0.1359, "step": 2290 }, { "epoch": 2.1, "learning_rate": 1.7296896509579886e-05, "loss": 0.1378, "step": 2291 }, { "epoch": 2.1, "learning_rate": 1.7294358618318446e-05, "loss": 0.1388, "step": 2292 }, { "epoch": 2.1, "learning_rate": 1.729181972260238e-05, "loss": 0.1397, "step": 2293 }, { "epoch": 2.1, "learning_rate": 1.7289279822781303e-05, "loss": 0.1379, "step": 2294 }, { "epoch": 2.1, "learning_rate": 1.7286738919204964e-05, "loss": 0.1312, "step": 2295 }, { "epoch": 2.1, "learning_rate": 1.7284197012223253e-05, "loss": 0.1414, "step": 2296 }, { "epoch": 2.11, "learning_rate": 1.72816541021862e-05, "loss": 0.145, "step": 2297 }, { "epoch": 2.11, "learning_rate": 1.727911018944397e-05, "loss": 0.136, "step": 2298 }, { "epoch": 2.11, "learning_rate": 1.7276565274346868e-05, "loss": 0.1385, "step": 2299 }, { "epoch": 2.11, "learning_rate": 1.7274019357245334e-05, "loss": 0.1414, "step": 2300 }, { "epoch": 2.11, "learning_rate": 1.7271472438489955e-05, "loss": 0.1252, "step": 2301 }, { "epoch": 2.11, "learning_rate": 1.7268924518431437e-05, "loss": 0.1347, "step": 2302 }, { "epoch": 2.11, "learning_rate": 1.726637559742065e-05, "loss": 0.1449, "step": 2303 }, { "epoch": 2.11, "learning_rate": 1.726382567580858e-05, "loss": 0.132, "step": 2304 }, { "epoch": 2.11, "learning_rate": 1.7261274753946356e-05, "loss": 0.1381, "step": 2305 }, { "epoch": 2.11, "learning_rate": 1.7258722832185254e-05, "loss": 0.1389, "step": 2306 }, { "epoch": 2.11, "learning_rate": 1.725616991087668e-05, "loss": 0.1338, "step": 2307 }, { "epoch": 2.12, "learning_rate": 1.7253615990372177e-05, "loss": 0.129, "step": 2308 }, { "epoch": 2.12, "learning_rate": 1.725106107102343e-05, "loss": 0.1345, "step": 2309 }, { "epoch": 2.12, "learning_rate": 1.7248505153182252e-05, "loss": 0.1402, "step": 2310 }, { "epoch": 2.12, "learning_rate": 1.7245948237200608e-05, "loss": 0.1429, "step": 2311 }, { "epoch": 2.12, "learning_rate": 1.7243390323430587e-05, "loss": 0.1409, "step": 2312 }, { "epoch": 2.12, "learning_rate": 1.7240831412224427e-05, "loss": 0.134, "step": 2313 }, { "epoch": 2.12, "learning_rate": 1.7238271503934494e-05, "loss": 0.1408, "step": 2314 }, { "epoch": 2.12, "learning_rate": 1.7235710598913296e-05, "loss": 0.1329, "step": 2315 }, { "epoch": 2.12, "learning_rate": 1.723314869751348e-05, "loss": 0.135, "step": 2316 }, { "epoch": 2.12, "learning_rate": 1.7230585800087816e-05, "loss": 0.1343, "step": 2317 }, { "epoch": 2.12, "learning_rate": 1.7228021906989235e-05, "loss": 0.1429, "step": 2318 }, { "epoch": 2.13, "learning_rate": 1.722545701857079e-05, "loss": 0.1282, "step": 2319 }, { "epoch": 2.13, "learning_rate": 1.7222891135185672e-05, "loss": 0.137, "step": 2320 }, { "epoch": 2.13, "learning_rate": 1.722032425718721e-05, "loss": 0.1327, "step": 2321 }, { "epoch": 2.13, "learning_rate": 1.7217756384928867e-05, "loss": 0.1289, "step": 2322 }, { "epoch": 2.13, "learning_rate": 1.7215187518764253e-05, "loss": 0.142, "step": 2323 }, { "epoch": 2.13, "learning_rate": 1.721261765904711e-05, "loss": 0.1354, "step": 2324 }, { "epoch": 2.13, "learning_rate": 1.7210046806131307e-05, "loss": 0.1278, "step": 2325 }, { "epoch": 2.13, "learning_rate": 1.7207474960370865e-05, "loss": 0.1387, "step": 2326 }, { "epoch": 2.13, "learning_rate": 1.7204902122119932e-05, "loss": 0.1358, "step": 2327 }, { "epoch": 2.13, "learning_rate": 1.7202328291732797e-05, "loss": 0.1346, "step": 2328 }, { "epoch": 2.13, "learning_rate": 1.719975346956388e-05, "loss": 0.1358, "step": 2329 }, { "epoch": 2.14, "learning_rate": 1.7197177655967745e-05, "loss": 0.1322, "step": 2330 }, { "epoch": 2.14, "learning_rate": 1.719460085129909e-05, "loss": 0.1408, "step": 2331 }, { "epoch": 2.14, "learning_rate": 1.7192023055912743e-05, "loss": 0.1222, "step": 2332 }, { "epoch": 2.14, "learning_rate": 1.718944427016368e-05, "loss": 0.1389, "step": 2333 }, { "epoch": 2.14, "learning_rate": 1.7186864494407007e-05, "loss": 0.136, "step": 2334 }, { "epoch": 2.14, "learning_rate": 1.7184283728997965e-05, "loss": 0.141, "step": 2335 }, { "epoch": 2.14, "learning_rate": 1.7181701974291927e-05, "loss": 0.1321, "step": 2336 }, { "epoch": 2.14, "learning_rate": 1.717911923064442e-05, "loss": 0.1382, "step": 2337 }, { "epoch": 2.14, "learning_rate": 1.7176535498411093e-05, "loss": 0.1297, "step": 2338 }, { "epoch": 2.14, "learning_rate": 1.7173950777947722e-05, "loss": 0.135, "step": 2339 }, { "epoch": 2.14, "learning_rate": 1.7171365069610242e-05, "loss": 0.1275, "step": 2340 }, { "epoch": 2.15, "learning_rate": 1.7168778373754708e-05, "loss": 0.1277, "step": 2341 }, { "epoch": 2.15, "learning_rate": 1.7166190690737318e-05, "loss": 0.1376, "step": 2342 }, { "epoch": 2.15, "learning_rate": 1.7163602020914404e-05, "loss": 0.1381, "step": 2343 }, { "epoch": 2.15, "learning_rate": 1.7161012364642432e-05, "loss": 0.1242, "step": 2344 }, { "epoch": 2.15, "learning_rate": 1.7158421722278004e-05, "loss": 0.1417, "step": 2345 }, { "epoch": 2.15, "learning_rate": 1.7155830094177858e-05, "loss": 0.14, "step": 2346 }, { "epoch": 2.15, "learning_rate": 1.7153237480698877e-05, "loss": 0.1349, "step": 2347 }, { "epoch": 2.15, "learning_rate": 1.715064388219806e-05, "loss": 0.132, "step": 2348 }, { "epoch": 2.15, "learning_rate": 1.7148049299032564e-05, "loss": 0.1336, "step": 2349 }, { "epoch": 2.15, "learning_rate": 1.714545373155966e-05, "loss": 0.1329, "step": 2350 }, { "epoch": 2.15, "learning_rate": 1.7142857180136773e-05, "loss": 0.1496, "step": 2351 }, { "epoch": 2.16, "learning_rate": 1.7140259645121452e-05, "loss": 0.1232, "step": 2352 }, { "epoch": 2.16, "learning_rate": 1.713766112687139e-05, "loss": 0.1356, "step": 2353 }, { "epoch": 2.16, "learning_rate": 1.7135061625744405e-05, "loss": 0.1358, "step": 2354 }, { "epoch": 2.16, "learning_rate": 1.713246114209846e-05, "loss": 0.1311, "step": 2355 }, { "epoch": 2.16, "learning_rate": 1.7129859676291645e-05, "loss": 0.1274, "step": 2356 }, { "epoch": 2.16, "learning_rate": 1.7127257228682196e-05, "loss": 0.1291, "step": 2357 }, { "epoch": 2.16, "learning_rate": 1.712465379962847e-05, "loss": 0.126, "step": 2358 }, { "epoch": 2.16, "learning_rate": 1.7122049389488972e-05, "loss": 0.1227, "step": 2359 }, { "epoch": 2.16, "learning_rate": 1.711944399862234e-05, "loss": 0.1535, "step": 2360 }, { "epoch": 2.16, "learning_rate": 1.7116837627387332e-05, "loss": 0.133, "step": 2361 }, { "epoch": 2.16, "learning_rate": 1.7114230276142866e-05, "loss": 0.1364, "step": 2362 }, { "epoch": 2.17, "learning_rate": 1.711162194524797e-05, "loss": 0.1296, "step": 2363 }, { "epoch": 2.17, "learning_rate": 1.710901263506183e-05, "loss": 0.147, "step": 2364 }, { "epoch": 2.17, "learning_rate": 1.7106402345943748e-05, "loss": 0.1373, "step": 2365 }, { "epoch": 2.17, "learning_rate": 1.7103791078253175e-05, "loss": 0.1353, "step": 2366 }, { "epoch": 2.17, "learning_rate": 1.7101178832349687e-05, "loss": 0.1418, "step": 2367 }, { "epoch": 2.17, "learning_rate": 1.7098565608592993e-05, "loss": 0.1266, "step": 2368 }, { "epoch": 2.17, "learning_rate": 1.709595140734295e-05, "loss": 0.1259, "step": 2369 }, { "epoch": 2.17, "learning_rate": 1.7093336228959538e-05, "loss": 0.1437, "step": 2370 }, { "epoch": 2.17, "learning_rate": 1.709072007380287e-05, "loss": 0.1378, "step": 2371 }, { "epoch": 2.17, "learning_rate": 1.70881029422332e-05, "loss": 0.1343, "step": 2372 }, { "epoch": 2.18, "learning_rate": 1.7085484834610925e-05, "loss": 0.1365, "step": 2373 }, { "epoch": 2.18, "learning_rate": 1.7082865751296553e-05, "loss": 0.1507, "step": 2374 }, { "epoch": 2.18, "learning_rate": 1.7080245692650746e-05, "loss": 0.1476, "step": 2375 }, { "epoch": 2.18, "learning_rate": 1.7077624659034296e-05, "loss": 0.1322, "step": 2376 }, { "epoch": 2.18, "learning_rate": 1.707500265080812e-05, "loss": 0.1335, "step": 2377 }, { "epoch": 2.18, "learning_rate": 1.707237966833328e-05, "loss": 0.127, "step": 2378 }, { "epoch": 2.18, "learning_rate": 1.7069755711970966e-05, "loss": 0.134, "step": 2379 }, { "epoch": 2.18, "learning_rate": 1.706713078208251e-05, "loss": 0.1397, "step": 2380 }, { "epoch": 2.18, "learning_rate": 1.7064504879029366e-05, "loss": 0.1337, "step": 2381 }, { "epoch": 2.18, "learning_rate": 1.7061878003173134e-05, "loss": 0.1388, "step": 2382 }, { "epoch": 2.18, "learning_rate": 1.705925015487554e-05, "loss": 0.1395, "step": 2383 }, { "epoch": 2.19, "learning_rate": 1.7056621334498447e-05, "loss": 0.1361, "step": 2384 }, { "epoch": 2.19, "learning_rate": 1.7053991542403848e-05, "loss": 0.1343, "step": 2385 }, { "epoch": 2.19, "learning_rate": 1.705136077895388e-05, "loss": 0.1411, "step": 2386 }, { "epoch": 2.19, "learning_rate": 1.70487290445108e-05, "loss": 0.1332, "step": 2387 }, { "epoch": 2.19, "learning_rate": 1.7046096339437014e-05, "loss": 0.142, "step": 2388 }, { "epoch": 2.19, "learning_rate": 1.704346266409504e-05, "loss": 0.1338, "step": 2389 }, { "epoch": 2.19, "learning_rate": 1.7040828018847558e-05, "loss": 0.1365, "step": 2390 }, { "epoch": 2.19, "learning_rate": 1.7038192404057352e-05, "loss": 0.1441, "step": 2391 }, { "epoch": 2.19, "learning_rate": 1.7035555820087366e-05, "loss": 0.1505, "step": 2392 }, { "epoch": 2.19, "learning_rate": 1.703291826730066e-05, "loss": 0.1328, "step": 2393 }, { "epoch": 2.19, "learning_rate": 1.703027974606043e-05, "loss": 0.1463, "step": 2394 }, { "epoch": 2.2, "learning_rate": 1.7027640256730013e-05, "loss": 0.1368, "step": 2395 }, { "epoch": 2.2, "learning_rate": 1.702499979967287e-05, "loss": 0.1338, "step": 2396 }, { "epoch": 2.2, "learning_rate": 1.7022358375252607e-05, "loss": 0.1337, "step": 2397 }, { "epoch": 2.2, "learning_rate": 1.701971598383295e-05, "loss": 0.1264, "step": 2398 }, { "epoch": 2.2, "learning_rate": 1.701707262577776e-05, "loss": 0.1354, "step": 2399 }, { "epoch": 2.2, "learning_rate": 1.7014428301451048e-05, "loss": 0.1267, "step": 2400 }, { "epoch": 2.2, "learning_rate": 1.701178301121693e-05, "loss": 0.1319, "step": 2401 }, { "epoch": 2.2, "learning_rate": 1.7009136755439685e-05, "loss": 0.1342, "step": 2402 }, { "epoch": 2.2, "learning_rate": 1.7006489534483702e-05, "loss": 0.1301, "step": 2403 }, { "epoch": 2.2, "learning_rate": 1.700384134871351e-05, "loss": 0.1274, "step": 2404 }, { "epoch": 2.2, "learning_rate": 1.7001192198493776e-05, "loss": 0.15, "step": 2405 }, { "epoch": 2.21, "learning_rate": 1.699854208418929e-05, "loss": 0.1257, "step": 2406 }, { "epoch": 2.21, "learning_rate": 1.699589100616499e-05, "loss": 0.1256, "step": 2407 }, { "epoch": 2.21, "learning_rate": 1.699323896478593e-05, "loss": 0.141, "step": 2408 }, { "epoch": 2.21, "learning_rate": 1.69905859604173e-05, "loss": 0.1251, "step": 2409 }, { "epoch": 2.21, "learning_rate": 1.6987931993424437e-05, "loss": 0.1253, "step": 2410 }, { "epoch": 2.21, "learning_rate": 1.6985277064172797e-05, "loss": 0.1506, "step": 2411 }, { "epoch": 2.21, "learning_rate": 1.6982621173027967e-05, "loss": 0.1396, "step": 2412 }, { "epoch": 2.21, "learning_rate": 1.697996432035568e-05, "loss": 0.1321, "step": 2413 }, { "epoch": 2.21, "learning_rate": 1.6977306506521776e-05, "loss": 0.1296, "step": 2414 }, { "epoch": 2.21, "learning_rate": 1.697464773189226e-05, "loss": 0.1256, "step": 2415 }, { "epoch": 2.21, "learning_rate": 1.6971987996833242e-05, "loss": 0.1339, "step": 2416 }, { "epoch": 2.22, "learning_rate": 1.696932730171099e-05, "loss": 0.131, "step": 2417 }, { "epoch": 2.22, "learning_rate": 1.696666564689187e-05, "loss": 0.1385, "step": 2418 }, { "epoch": 2.22, "learning_rate": 1.6964003032742415e-05, "loss": 0.1317, "step": 2419 }, { "epoch": 2.22, "learning_rate": 1.696133945962927e-05, "loss": 0.1323, "step": 2420 }, { "epoch": 2.22, "learning_rate": 1.6958674927919213e-05, "loss": 0.1261, "step": 2421 }, { "epoch": 2.22, "learning_rate": 1.695600943797916e-05, "loss": 0.1266, "step": 2422 }, { "epoch": 2.22, "learning_rate": 1.6953342990176163e-05, "loss": 0.1388, "step": 2423 }, { "epoch": 2.22, "learning_rate": 1.695067558487739e-05, "loss": 0.1304, "step": 2424 }, { "epoch": 2.22, "learning_rate": 1.6948007222450158e-05, "loss": 0.1328, "step": 2425 }, { "epoch": 2.22, "learning_rate": 1.6945337903261908e-05, "loss": 0.1388, "step": 2426 }, { "epoch": 2.22, "learning_rate": 1.6942667627680206e-05, "loss": 0.1375, "step": 2427 }, { "epoch": 2.23, "learning_rate": 1.693999639607276e-05, "loss": 0.1257, "step": 2428 }, { "epoch": 2.23, "learning_rate": 1.693732420880741e-05, "loss": 0.1353, "step": 2429 }, { "epoch": 2.23, "learning_rate": 1.6934651066252127e-05, "loss": 0.1237, "step": 2430 }, { "epoch": 2.23, "learning_rate": 1.6931976968774998e-05, "loss": 0.1359, "step": 2431 }, { "epoch": 2.23, "learning_rate": 1.6929301916744267e-05, "loss": 0.1458, "step": 2432 }, { "epoch": 2.23, "learning_rate": 1.692662591052829e-05, "loss": 0.1457, "step": 2433 }, { "epoch": 2.23, "learning_rate": 1.6923948950495562e-05, "loss": 0.1326, "step": 2434 }, { "epoch": 2.23, "learning_rate": 1.6921271037014704e-05, "loss": 0.1277, "step": 2435 }, { "epoch": 2.23, "learning_rate": 1.6918592170454483e-05, "loss": 0.1383, "step": 2436 }, { "epoch": 2.23, "learning_rate": 1.6915912351183778e-05, "loss": 0.1245, "step": 2437 }, { "epoch": 2.23, "learning_rate": 1.691323157957161e-05, "loss": 0.1343, "step": 2438 }, { "epoch": 2.24, "learning_rate": 1.691054985598713e-05, "loss": 0.1402, "step": 2439 }, { "epoch": 2.24, "learning_rate": 1.690786718079962e-05, "loss": 0.137, "step": 2440 }, { "epoch": 2.24, "learning_rate": 1.690518355437849e-05, "loss": 0.1338, "step": 2441 }, { "epoch": 2.24, "learning_rate": 1.6902498977093285e-05, "loss": 0.1382, "step": 2442 }, { "epoch": 2.24, "learning_rate": 1.6899813449313678e-05, "loss": 0.1299, "step": 2443 }, { "epoch": 2.24, "learning_rate": 1.6897126971409475e-05, "loss": 0.1449, "step": 2444 }, { "epoch": 2.24, "learning_rate": 1.6894439543750614e-05, "loss": 0.1358, "step": 2445 }, { "epoch": 2.24, "learning_rate": 1.6891751166707156e-05, "loss": 0.1401, "step": 2446 }, { "epoch": 2.24, "learning_rate": 1.6889061840649304e-05, "loss": 0.1319, "step": 2447 }, { "epoch": 2.24, "learning_rate": 1.6886371565947383e-05, "loss": 0.1173, "step": 2448 }, { "epoch": 2.24, "learning_rate": 1.6883680342971854e-05, "loss": 0.1283, "step": 2449 }, { "epoch": 2.25, "learning_rate": 1.6880988172093304e-05, "loss": 0.1413, "step": 2450 }, { "epoch": 2.25, "learning_rate": 1.6878295053682455e-05, "loss": 0.138, "step": 2451 }, { "epoch": 2.25, "learning_rate": 1.6875600988110155e-05, "loss": 0.14, "step": 2452 }, { "epoch": 2.25, "learning_rate": 1.6872905975747382e-05, "loss": 0.1303, "step": 2453 }, { "epoch": 2.25, "learning_rate": 1.6870210016965257e-05, "loss": 0.1345, "step": 2454 }, { "epoch": 2.25, "learning_rate": 1.6867513112135012e-05, "loss": 0.1293, "step": 2455 }, { "epoch": 2.25, "learning_rate": 1.6864815261628022e-05, "loss": 0.1419, "step": 2456 }, { "epoch": 2.25, "learning_rate": 1.686211646581579e-05, "loss": 0.1296, "step": 2457 }, { "epoch": 2.25, "learning_rate": 1.6859416725069947e-05, "loss": 0.1272, "step": 2458 }, { "epoch": 2.25, "learning_rate": 1.685671603976226e-05, "loss": 0.1293, "step": 2459 }, { "epoch": 2.25, "learning_rate": 1.6854014410264613e-05, "loss": 0.1302, "step": 2460 }, { "epoch": 2.26, "learning_rate": 1.6851311836949033e-05, "loss": 0.1345, "step": 2461 }, { "epoch": 2.26, "learning_rate": 1.6848608320187668e-05, "loss": 0.1348, "step": 2462 }, { "epoch": 2.26, "learning_rate": 1.6845903860352807e-05, "loss": 0.1295, "step": 2463 }, { "epoch": 2.26, "learning_rate": 1.6843198457816855e-05, "loss": 0.1417, "step": 2464 }, { "epoch": 2.26, "learning_rate": 1.684049211295236e-05, "loss": 0.1236, "step": 2465 }, { "epoch": 2.26, "learning_rate": 1.6837784826131992e-05, "loss": 0.1454, "step": 2466 }, { "epoch": 2.26, "learning_rate": 1.6835076597728554e-05, "loss": 0.1271, "step": 2467 }, { "epoch": 2.26, "learning_rate": 1.683236742811497e-05, "loss": 0.1335, "step": 2468 }, { "epoch": 2.26, "learning_rate": 1.6829657317664306e-05, "loss": 0.1315, "step": 2469 }, { "epoch": 2.26, "learning_rate": 1.6826946266749753e-05, "loss": 0.1364, "step": 2470 }, { "epoch": 2.26, "learning_rate": 1.6824234275744627e-05, "loss": 0.139, "step": 2471 }, { "epoch": 2.27, "learning_rate": 1.6821521345022377e-05, "loss": 0.1357, "step": 2472 }, { "epoch": 2.27, "learning_rate": 1.681880747495659e-05, "loss": 0.1376, "step": 2473 }, { "epoch": 2.27, "learning_rate": 1.6816092665920963e-05, "loss": 0.1342, "step": 2474 }, { "epoch": 2.27, "learning_rate": 1.6813376918289338e-05, "loss": 0.136, "step": 2475 }, { "epoch": 2.27, "learning_rate": 1.6810660232435685e-05, "loss": 0.1301, "step": 2476 }, { "epoch": 2.27, "learning_rate": 1.6807942608734092e-05, "loss": 0.1372, "step": 2477 }, { "epoch": 2.27, "learning_rate": 1.680522404755879e-05, "loss": 0.1221, "step": 2478 }, { "epoch": 2.27, "learning_rate": 1.680250454928413e-05, "loss": 0.1291, "step": 2479 }, { "epoch": 2.27, "learning_rate": 1.6799784114284593e-05, "loss": 0.1327, "step": 2480 }, { "epoch": 2.27, "learning_rate": 1.6797062742934795e-05, "loss": 0.1239, "step": 2481 }, { "epoch": 2.27, "learning_rate": 1.6794340435609474e-05, "loss": 0.1228, "step": 2482 }, { "epoch": 2.28, "learning_rate": 1.6791617192683506e-05, "loss": 0.1366, "step": 2483 }, { "epoch": 2.28, "learning_rate": 1.678889301453188e-05, "loss": 0.1259, "step": 2484 }, { "epoch": 2.28, "learning_rate": 1.6786167901529724e-05, "loss": 0.1275, "step": 2485 }, { "epoch": 2.28, "learning_rate": 1.6783441854052304e-05, "loss": 0.1339, "step": 2486 }, { "epoch": 2.28, "learning_rate": 1.6780714872474993e-05, "loss": 0.1215, "step": 2487 }, { "epoch": 2.28, "learning_rate": 1.6777986957173315e-05, "loss": 0.1273, "step": 2488 }, { "epoch": 2.28, "learning_rate": 1.677525810852291e-05, "loss": 0.1257, "step": 2489 }, { "epoch": 2.28, "learning_rate": 1.6772528326899538e-05, "loss": 0.1359, "step": 2490 }, { "epoch": 2.28, "learning_rate": 1.676979761267911e-05, "loss": 0.1337, "step": 2491 }, { "epoch": 2.28, "learning_rate": 1.6767065966237644e-05, "loss": 0.1272, "step": 2492 }, { "epoch": 2.29, "learning_rate": 1.6764333387951303e-05, "loss": 0.1313, "step": 2493 }, { "epoch": 2.29, "learning_rate": 1.676159987819637e-05, "loss": 0.1347, "step": 2494 }, { "epoch": 2.29, "learning_rate": 1.6758865437349254e-05, "loss": 0.1307, "step": 2495 }, { "epoch": 2.29, "learning_rate": 1.6756130065786497e-05, "loss": 0.125, "step": 2496 }, { "epoch": 2.29, "learning_rate": 1.675339376388477e-05, "loss": 0.1315, "step": 2497 }, { "epoch": 2.29, "learning_rate": 1.6750656532020867e-05, "loss": 0.139, "step": 2498 }, { "epoch": 2.29, "learning_rate": 1.6747918370571712e-05, "loss": 0.1282, "step": 2499 }, { "epoch": 2.29, "learning_rate": 1.674517927991436e-05, "loss": 0.1302, "step": 2500 }, { "epoch": 2.29, "learning_rate": 1.6742439260425993e-05, "loss": 0.1431, "step": 2501 }, { "epoch": 2.29, "learning_rate": 1.673969831248392e-05, "loss": 0.1318, "step": 2502 }, { "epoch": 2.29, "learning_rate": 1.6736956436465573e-05, "loss": 0.1285, "step": 2503 }, { "epoch": 2.3, "learning_rate": 1.6734213632748518e-05, "loss": 0.1418, "step": 2504 }, { "epoch": 2.3, "learning_rate": 1.6731469901710455e-05, "loss": 0.132, "step": 2505 }, { "epoch": 2.3, "learning_rate": 1.672872524372919e-05, "loss": 0.1372, "step": 2506 }, { "epoch": 2.3, "learning_rate": 1.672597965918268e-05, "loss": 0.1266, "step": 2507 }, { "epoch": 2.3, "learning_rate": 1.6723233148449e-05, "loss": 0.1229, "step": 2508 }, { "epoch": 2.3, "learning_rate": 1.6720485711906346e-05, "loss": 0.1351, "step": 2509 }, { "epoch": 2.3, "learning_rate": 1.6717737349933057e-05, "loss": 0.1413, "step": 2510 }, { "epoch": 2.3, "learning_rate": 1.6714988062907584e-05, "loss": 0.1356, "step": 2511 }, { "epoch": 2.3, "learning_rate": 1.671223785120851e-05, "loss": 0.1345, "step": 2512 }, { "epoch": 2.3, "learning_rate": 1.6709486715214554e-05, "loss": 0.1343, "step": 2513 }, { "epoch": 2.3, "learning_rate": 1.6706734655304556e-05, "loss": 0.1262, "step": 2514 }, { "epoch": 2.31, "learning_rate": 1.6703981671857477e-05, "loss": 0.1394, "step": 2515 }, { "epoch": 2.31, "learning_rate": 1.6701227765252413e-05, "loss": 0.1258, "step": 2516 }, { "epoch": 2.31, "learning_rate": 1.6698472935868583e-05, "loss": 0.1359, "step": 2517 }, { "epoch": 2.31, "learning_rate": 1.6695717184085342e-05, "loss": 0.1382, "step": 2518 }, { "epoch": 2.31, "learning_rate": 1.6692960510282156e-05, "loss": 0.1268, "step": 2519 }, { "epoch": 2.31, "learning_rate": 1.6690202914838636e-05, "loss": 0.1406, "step": 2520 }, { "epoch": 2.31, "learning_rate": 1.6687444398134504e-05, "loss": 0.1308, "step": 2521 }, { "epoch": 2.31, "learning_rate": 1.6684684960549615e-05, "loss": 0.1341, "step": 2522 }, { "epoch": 2.31, "learning_rate": 1.668192460246396e-05, "loss": 0.1316, "step": 2523 }, { "epoch": 2.31, "learning_rate": 1.6679163324257643e-05, "loss": 0.1382, "step": 2524 }, { "epoch": 2.31, "learning_rate": 1.66764011263109e-05, "loss": 0.1427, "step": 2525 }, { "epoch": 2.32, "learning_rate": 1.6673638009004095e-05, "loss": 0.123, "step": 2526 }, { "epoch": 2.32, "learning_rate": 1.6670873972717715e-05, "loss": 0.1315, "step": 2527 }, { "epoch": 2.32, "learning_rate": 1.666810901783238e-05, "loss": 0.1261, "step": 2528 }, { "epoch": 2.32, "learning_rate": 1.666534314472883e-05, "loss": 0.1343, "step": 2529 }, { "epoch": 2.32, "learning_rate": 1.666257635378793e-05, "loss": 0.1255, "step": 2530 }, { "epoch": 2.32, "learning_rate": 1.6659808645390684e-05, "loss": 0.1297, "step": 2531 }, { "epoch": 2.32, "learning_rate": 1.6657040019918206e-05, "loss": 0.1333, "step": 2532 }, { "epoch": 2.32, "learning_rate": 1.6654270477751748e-05, "loss": 0.1334, "step": 2533 }, { "epoch": 2.32, "learning_rate": 1.6651500019272684e-05, "loss": 0.1347, "step": 2534 }, { "epoch": 2.32, "learning_rate": 1.6648728644862512e-05, "loss": 0.1186, "step": 2535 }, { "epoch": 2.32, "learning_rate": 1.664595635490286e-05, "loss": 0.1315, "step": 2536 }, { "epoch": 2.33, "learning_rate": 1.6643183149775476e-05, "loss": 0.1297, "step": 2537 }, { "epoch": 2.33, "learning_rate": 1.6640409029862247e-05, "loss": 0.1356, "step": 2538 }, { "epoch": 2.33, "learning_rate": 1.663763399554517e-05, "loss": 0.1375, "step": 2539 }, { "epoch": 2.33, "learning_rate": 1.663485804720638e-05, "loss": 0.1378, "step": 2540 }, { "epoch": 2.33, "learning_rate": 1.6632081185228132e-05, "loss": 0.1256, "step": 2541 }, { "epoch": 2.33, "learning_rate": 1.662930340999281e-05, "loss": 0.1321, "step": 2542 }, { "epoch": 2.33, "learning_rate": 1.6626524721882914e-05, "loss": 0.1332, "step": 2543 }, { "epoch": 2.33, "learning_rate": 1.662374512128109e-05, "loss": 0.1451, "step": 2544 }, { "epoch": 2.33, "learning_rate": 1.6620964608570083e-05, "loss": 0.1262, "step": 2545 }, { "epoch": 2.33, "learning_rate": 1.661818318413279e-05, "loss": 0.1441, "step": 2546 }, { "epoch": 2.33, "learning_rate": 1.661540084835221e-05, "loss": 0.1288, "step": 2547 }, { "epoch": 2.34, "learning_rate": 1.661261760161149e-05, "loss": 0.1242, "step": 2548 }, { "epoch": 2.34, "learning_rate": 1.6609833444293885e-05, "loss": 0.1317, "step": 2549 }, { "epoch": 2.34, "learning_rate": 1.6607048376782783e-05, "loss": 0.1366, "step": 2550 }, { "epoch": 2.34, "learning_rate": 1.6604262399461696e-05, "loss": 0.1332, "step": 2551 }, { "epoch": 2.34, "learning_rate": 1.6601475512714263e-05, "loss": 0.1314, "step": 2552 }, { "epoch": 2.34, "learning_rate": 1.6598687716924243e-05, "loss": 0.1226, "step": 2553 }, { "epoch": 2.34, "learning_rate": 1.6595899012475523e-05, "loss": 0.1248, "step": 2554 }, { "epoch": 2.34, "learning_rate": 1.659310939975212e-05, "loss": 0.1308, "step": 2555 }, { "epoch": 2.34, "learning_rate": 1.659031887913817e-05, "loss": 0.1318, "step": 2556 }, { "epoch": 2.34, "learning_rate": 1.658752745101794e-05, "loss": 0.1391, "step": 2557 }, { "epoch": 2.34, "learning_rate": 1.658473511577581e-05, "loss": 0.1373, "step": 2558 }, { "epoch": 2.35, "learning_rate": 1.65819418737963e-05, "loss": 0.1232, "step": 2559 }, { "epoch": 2.35, "learning_rate": 1.6579147725464038e-05, "loss": 0.1306, "step": 2560 }, { "epoch": 2.35, "learning_rate": 1.6576352671163796e-05, "loss": 0.1212, "step": 2561 }, { "epoch": 2.35, "learning_rate": 1.6573556711280457e-05, "loss": 0.1262, "step": 2562 }, { "epoch": 2.35, "learning_rate": 1.6570759846199035e-05, "loss": 0.1361, "step": 2563 }, { "epoch": 2.35, "learning_rate": 1.6567962076304666e-05, "loss": 0.1294, "step": 2564 }, { "epoch": 2.35, "learning_rate": 1.6565163401982606e-05, "loss": 0.1312, "step": 2565 }, { "epoch": 2.35, "learning_rate": 1.656236382361825e-05, "loss": 0.1353, "step": 2566 }, { "epoch": 2.35, "learning_rate": 1.65595633415971e-05, "loss": 0.1285, "step": 2567 }, { "epoch": 2.35, "learning_rate": 1.6556761956304796e-05, "loss": 0.1277, "step": 2568 }, { "epoch": 2.35, "learning_rate": 1.6553959668127096e-05, "loss": 0.1354, "step": 2569 }, { "epoch": 2.36, "learning_rate": 1.6551156477449878e-05, "loss": 0.1259, "step": 2570 }, { "epoch": 2.36, "learning_rate": 1.654835238465916e-05, "loss": 0.1298, "step": 2571 }, { "epoch": 2.36, "learning_rate": 1.6545547390141063e-05, "loss": 0.1372, "step": 2572 }, { "epoch": 2.36, "learning_rate": 1.6542741494281847e-05, "loss": 0.1342, "step": 2573 }, { "epoch": 2.36, "learning_rate": 1.6539934697467895e-05, "loss": 0.1237, "step": 2574 }, { "epoch": 2.36, "learning_rate": 1.6537127000085708e-05, "loss": 0.1259, "step": 2575 }, { "epoch": 2.36, "learning_rate": 1.6534318402521913e-05, "loss": 0.129, "step": 2576 }, { "epoch": 2.36, "learning_rate": 1.6531508905163268e-05, "loss": 0.1416, "step": 2577 }, { "epoch": 2.36, "learning_rate": 1.6528698508396647e-05, "loss": 0.1369, "step": 2578 }, { "epoch": 2.36, "learning_rate": 1.6525887212609042e-05, "loss": 0.1284, "step": 2579 }, { "epoch": 2.36, "learning_rate": 1.6523075018187586e-05, "loss": 0.1232, "step": 2580 }, { "epoch": 2.37, "learning_rate": 1.6520261925519523e-05, "loss": 0.1326, "step": 2581 }, { "epoch": 2.37, "learning_rate": 1.6517447934992224e-05, "loss": 0.1311, "step": 2582 }, { "epoch": 2.37, "learning_rate": 1.6514633046993184e-05, "loss": 0.1367, "step": 2583 }, { "epoch": 2.37, "learning_rate": 1.6511817261910024e-05, "loss": 0.1329, "step": 2584 }, { "epoch": 2.37, "learning_rate": 1.6509000580130478e-05, "loss": 0.117, "step": 2585 }, { "epoch": 2.37, "learning_rate": 1.650618300204242e-05, "loss": 0.121, "step": 2586 }, { "epoch": 2.37, "learning_rate": 1.6503364528033838e-05, "loss": 0.135, "step": 2587 }, { "epoch": 2.37, "learning_rate": 1.650054515849284e-05, "loss": 0.1286, "step": 2588 }, { "epoch": 2.37, "learning_rate": 1.6497724893807664e-05, "loss": 0.1307, "step": 2589 }, { "epoch": 2.37, "learning_rate": 1.6494903734366664e-05, "loss": 0.1369, "step": 2590 }, { "epoch": 2.37, "learning_rate": 1.649208168055833e-05, "loss": 0.1315, "step": 2591 }, { "epoch": 2.38, "learning_rate": 1.6489258732771262e-05, "loss": 0.1247, "step": 2592 }, { "epoch": 2.38, "learning_rate": 1.648643489139419e-05, "loss": 0.1257, "step": 2593 }, { "epoch": 2.38, "learning_rate": 1.648361015681596e-05, "loss": 0.133, "step": 2594 }, { "epoch": 2.38, "learning_rate": 1.6480784529425556e-05, "loss": 0.1304, "step": 2595 }, { "epoch": 2.38, "learning_rate": 1.6477958009612068e-05, "loss": 0.1351, "step": 2596 }, { "epoch": 2.38, "learning_rate": 1.647513059776472e-05, "loss": 0.1297, "step": 2597 }, { "epoch": 2.38, "learning_rate": 1.647230229427285e-05, "loss": 0.138, "step": 2598 }, { "epoch": 2.38, "learning_rate": 1.6469473099525926e-05, "loss": 0.1208, "step": 2599 }, { "epoch": 2.38, "learning_rate": 1.6466643013913543e-05, "loss": 0.1304, "step": 2600 }, { "epoch": 2.38, "learning_rate": 1.6463812037825397e-05, "loss": 0.1175, "step": 2601 }, { "epoch": 2.38, "learning_rate": 1.6460980171651338e-05, "loss": 0.1369, "step": 2602 }, { "epoch": 2.39, "learning_rate": 1.645814741578131e-05, "loss": 0.1291, "step": 2603 }, { "epoch": 2.39, "learning_rate": 1.6455313770605402e-05, "loss": 0.1299, "step": 2604 }, { "epoch": 2.39, "learning_rate": 1.6452479236513806e-05, "loss": 0.1292, "step": 2605 }, { "epoch": 2.39, "learning_rate": 1.6449643813896852e-05, "loss": 0.1192, "step": 2606 }, { "epoch": 2.39, "learning_rate": 1.6446807503144986e-05, "loss": 0.126, "step": 2607 }, { "epoch": 2.39, "learning_rate": 1.644397030464877e-05, "loss": 0.1304, "step": 2608 }, { "epoch": 2.39, "learning_rate": 1.6441132218798905e-05, "loss": 0.1303, "step": 2609 }, { "epoch": 2.39, "learning_rate": 1.6438293245986193e-05, "loss": 0.1262, "step": 2610 }, { "epoch": 2.39, "learning_rate": 1.6435453386601575e-05, "loss": 0.1289, "step": 2611 }, { "epoch": 2.39, "learning_rate": 1.6432612641036105e-05, "loss": 0.1227, "step": 2612 }, { "epoch": 2.4, "learning_rate": 1.6429771009680964e-05, "loss": 0.127, "step": 2613 }, { "epoch": 2.4, "learning_rate": 1.6426928492927453e-05, "loss": 0.1355, "step": 2614 }, { "epoch": 2.4, "learning_rate": 1.6424085091166996e-05, "loss": 0.1456, "step": 2615 }, { "epoch": 2.4, "learning_rate": 1.6421240804791137e-05, "loss": 0.1381, "step": 2616 }, { "epoch": 2.4, "learning_rate": 1.6418395634191536e-05, "loss": 0.1338, "step": 2617 }, { "epoch": 2.4, "learning_rate": 1.6415549579759995e-05, "loss": 0.123, "step": 2618 }, { "epoch": 2.4, "learning_rate": 1.641270264188841e-05, "loss": 0.1322, "step": 2619 }, { "epoch": 2.4, "learning_rate": 1.640985482096882e-05, "loss": 0.1246, "step": 2620 }, { "epoch": 2.4, "learning_rate": 1.640700611739338e-05, "loss": 0.1405, "step": 2621 }, { "epoch": 2.4, "learning_rate": 1.640415653155436e-05, "loss": 0.1241, "step": 2622 }, { "epoch": 2.4, "learning_rate": 1.640130606384416e-05, "loss": 0.1261, "step": 2623 }, { "epoch": 2.41, "learning_rate": 1.6398454714655297e-05, "loss": 0.1285, "step": 2624 }, { "epoch": 2.41, "learning_rate": 1.6395602484380407e-05, "loss": 0.1364, "step": 2625 }, { "epoch": 2.41, "learning_rate": 1.6392749373412253e-05, "loss": 0.1401, "step": 2626 }, { "epoch": 2.41, "learning_rate": 1.6389895382143715e-05, "loss": 0.1356, "step": 2627 }, { "epoch": 2.41, "learning_rate": 1.63870405109678e-05, "loss": 0.1314, "step": 2628 }, { "epoch": 2.41, "learning_rate": 1.638418476027763e-05, "loss": 0.1292, "step": 2629 }, { "epoch": 2.41, "learning_rate": 1.6381328130466454e-05, "loss": 0.1383, "step": 2630 }, { "epoch": 2.41, "learning_rate": 1.637847062192763e-05, "loss": 0.1276, "step": 2631 }, { "epoch": 2.41, "learning_rate": 1.637561223505465e-05, "loss": 0.1218, "step": 2632 }, { "epoch": 2.41, "learning_rate": 1.6372752970241125e-05, "loss": 0.1341, "step": 2633 }, { "epoch": 2.41, "learning_rate": 1.6369892827880775e-05, "loss": 0.126, "step": 2634 }, { "epoch": 2.42, "learning_rate": 1.636703180836746e-05, "loss": 0.1187, "step": 2635 }, { "epoch": 2.42, "learning_rate": 1.6364169912095147e-05, "loss": 0.1238, "step": 2636 }, { "epoch": 2.42, "learning_rate": 1.6361307139457928e-05, "loss": 0.1243, "step": 2637 }, { "epoch": 2.42, "learning_rate": 1.6358443490850016e-05, "loss": 0.1392, "step": 2638 }, { "epoch": 2.42, "learning_rate": 1.635557896666574e-05, "loss": 0.1306, "step": 2639 }, { "epoch": 2.42, "learning_rate": 1.6352713567299553e-05, "loss": 0.1285, "step": 2640 }, { "epoch": 2.42, "learning_rate": 1.634984729314604e-05, "loss": 0.1266, "step": 2641 }, { "epoch": 2.42, "learning_rate": 1.634698014459988e-05, "loss": 0.1329, "step": 2642 }, { "epoch": 2.42, "learning_rate": 1.63441121220559e-05, "loss": 0.1381, "step": 2643 }, { "epoch": 2.42, "learning_rate": 1.634124322590903e-05, "loss": 0.1315, "step": 2644 }, { "epoch": 2.42, "learning_rate": 1.633837345655432e-05, "loss": 0.1328, "step": 2645 }, { "epoch": 2.43, "learning_rate": 1.6335502814386958e-05, "loss": 0.1336, "step": 2646 }, { "epoch": 2.43, "learning_rate": 1.633263129980223e-05, "loss": 0.1374, "step": 2647 }, { "epoch": 2.43, "learning_rate": 1.6329758913195557e-05, "loss": 0.1204, "step": 2648 }, { "epoch": 2.43, "learning_rate": 1.632688565496247e-05, "loss": 0.1428, "step": 2649 }, { "epoch": 2.43, "learning_rate": 1.632401152549863e-05, "loss": 0.1244, "step": 2650 }, { "epoch": 2.43, "learning_rate": 1.632113652519981e-05, "loss": 0.1233, "step": 2651 }, { "epoch": 2.43, "learning_rate": 1.631826065446191e-05, "loss": 0.1245, "step": 2652 }, { "epoch": 2.43, "learning_rate": 1.6315383913680945e-05, "loss": 0.1286, "step": 2653 }, { "epoch": 2.43, "learning_rate": 1.6312506303253045e-05, "loss": 0.1433, "step": 2654 }, { "epoch": 2.43, "learning_rate": 1.6309627823574472e-05, "loss": 0.1361, "step": 2655 }, { "epoch": 2.43, "learning_rate": 1.6306748475041595e-05, "loss": 0.1267, "step": 2656 }, { "epoch": 2.44, "learning_rate": 1.6303868258050916e-05, "loss": 0.1328, "step": 2657 }, { "epoch": 2.44, "learning_rate": 1.6300987172999042e-05, "loss": 0.1315, "step": 2658 }, { "epoch": 2.44, "learning_rate": 1.6298105220282715e-05, "loss": 0.1292, "step": 2659 }, { "epoch": 2.44, "learning_rate": 1.6295222400298782e-05, "loss": 0.115, "step": 2660 }, { "epoch": 2.44, "learning_rate": 1.6292338713444214e-05, "loss": 0.1399, "step": 2661 }, { "epoch": 2.44, "learning_rate": 1.628945416011611e-05, "loss": 0.132, "step": 2662 }, { "epoch": 2.44, "learning_rate": 1.6286568740711678e-05, "loss": 0.1233, "step": 2663 }, { "epoch": 2.44, "learning_rate": 1.628368245562825e-05, "loss": 0.1217, "step": 2664 }, { "epoch": 2.44, "learning_rate": 1.6280795305263276e-05, "loss": 0.1328, "step": 2665 }, { "epoch": 2.44, "learning_rate": 1.6277907290014318e-05, "loss": 0.1245, "step": 2666 }, { "epoch": 2.44, "learning_rate": 1.6275018410279075e-05, "loss": 0.1271, "step": 2667 }, { "epoch": 2.45, "learning_rate": 1.6272128666455348e-05, "loss": 0.1162, "step": 2668 }, { "epoch": 2.45, "learning_rate": 1.626923805894107e-05, "loss": 0.1296, "step": 2669 }, { "epoch": 2.45, "learning_rate": 1.626634658813427e-05, "loss": 0.1351, "step": 2670 }, { "epoch": 2.45, "learning_rate": 1.626345425443313e-05, "loss": 0.137, "step": 2671 }, { "epoch": 2.45, "learning_rate": 1.6260561058235926e-05, "loss": 0.1414, "step": 2672 }, { "epoch": 2.45, "learning_rate": 1.6257666999941058e-05, "loss": 0.129, "step": 2673 }, { "epoch": 2.45, "learning_rate": 1.6254772079947048e-05, "loss": 0.1316, "step": 2674 }, { "epoch": 2.45, "learning_rate": 1.6251876298652535e-05, "loss": 0.1283, "step": 2675 }, { "epoch": 2.45, "learning_rate": 1.6248979656456273e-05, "loss": 0.1325, "step": 2676 }, { "epoch": 2.45, "learning_rate": 1.6246082153757145e-05, "loss": 0.1314, "step": 2677 }, { "epoch": 2.45, "learning_rate": 1.624318379095414e-05, "loss": 0.1273, "step": 2678 }, { "epoch": 2.46, "learning_rate": 1.6240284568446373e-05, "loss": 0.1286, "step": 2679 }, { "epoch": 2.46, "learning_rate": 1.6237384486633077e-05, "loss": 0.134, "step": 2680 }, { "epoch": 2.46, "learning_rate": 1.62344835459136e-05, "loss": 0.1288, "step": 2681 }, { "epoch": 2.46, "learning_rate": 1.623158174668741e-05, "loss": 0.1242, "step": 2682 }, { "epoch": 2.46, "learning_rate": 1.6228679089354097e-05, "loss": 0.1327, "step": 2683 }, { "epoch": 2.46, "learning_rate": 1.6225775574313358e-05, "loss": 0.1201, "step": 2684 }, { "epoch": 2.46, "learning_rate": 1.622287120196502e-05, "loss": 0.1355, "step": 2685 }, { "epoch": 2.46, "learning_rate": 1.6219965972709023e-05, "loss": 0.1308, "step": 2686 }, { "epoch": 2.46, "learning_rate": 1.6217059886945427e-05, "loss": 0.1307, "step": 2687 }, { "epoch": 2.46, "learning_rate": 1.6214152945074412e-05, "loss": 0.1406, "step": 2688 }, { "epoch": 2.46, "learning_rate": 1.621124514749626e-05, "loss": 0.1414, "step": 2689 }, { "epoch": 2.47, "learning_rate": 1.6208336494611395e-05, "loss": 0.1278, "step": 2690 }, { "epoch": 2.47, "learning_rate": 1.6205426986820343e-05, "loss": 0.1239, "step": 2691 }, { "epoch": 2.47, "learning_rate": 1.6202516624523755e-05, "loss": 0.1286, "step": 2692 }, { "epoch": 2.47, "learning_rate": 1.619960540812239e-05, "loss": 0.1301, "step": 2693 }, { "epoch": 2.47, "learning_rate": 1.6196693338017132e-05, "loss": 0.1304, "step": 2694 }, { "epoch": 2.47, "learning_rate": 1.619378041460899e-05, "loss": 0.1305, "step": 2695 }, { "epoch": 2.47, "learning_rate": 1.6190866638299074e-05, "loss": 0.1355, "step": 2696 }, { "epoch": 2.47, "learning_rate": 1.618795200948862e-05, "loss": 0.1186, "step": 2697 }, { "epoch": 2.47, "learning_rate": 1.6185036528578985e-05, "loss": 0.1284, "step": 2698 }, { "epoch": 2.47, "learning_rate": 1.6182120195971637e-05, "loss": 0.1153, "step": 2699 }, { "epoch": 2.47, "learning_rate": 1.617920301206816e-05, "loss": 0.1406, "step": 2700 }, { "epoch": 2.48, "learning_rate": 1.6176284977270266e-05, "loss": 0.128, "step": 2701 }, { "epoch": 2.48, "learning_rate": 1.6173366091979776e-05, "loss": 0.1351, "step": 2702 }, { "epoch": 2.48, "learning_rate": 1.617044635659862e-05, "loss": 0.1252, "step": 2703 }, { "epoch": 2.48, "learning_rate": 1.6167525771528866e-05, "loss": 0.1362, "step": 2704 }, { "epoch": 2.48, "learning_rate": 1.6164604337172678e-05, "loss": 0.1185, "step": 2705 }, { "epoch": 2.48, "learning_rate": 1.6161682053932352e-05, "loss": 0.1324, "step": 2706 }, { "epoch": 2.48, "learning_rate": 1.6158758922210293e-05, "loss": 0.1286, "step": 2707 }, { "epoch": 2.48, "learning_rate": 1.6155834942409025e-05, "loss": 0.1384, "step": 2708 }, { "epoch": 2.48, "learning_rate": 1.6152910114931187e-05, "loss": 0.1264, "step": 2709 }, { "epoch": 2.48, "learning_rate": 1.614998444017954e-05, "loss": 0.1288, "step": 2710 }, { "epoch": 2.48, "learning_rate": 1.614705791855695e-05, "loss": 0.1265, "step": 2711 }, { "epoch": 2.49, "learning_rate": 1.614413055046642e-05, "loss": 0.1296, "step": 2712 }, { "epoch": 2.49, "learning_rate": 1.6141202336311046e-05, "loss": 0.1195, "step": 2713 }, { "epoch": 2.49, "learning_rate": 1.6138273276494062e-05, "loss": 0.1172, "step": 2714 }, { "epoch": 2.49, "learning_rate": 1.6135343371418796e-05, "loss": 0.1328, "step": 2715 }, { "epoch": 2.49, "learning_rate": 1.613241262148871e-05, "loss": 0.1294, "step": 2716 }, { "epoch": 2.49, "learning_rate": 1.6129481027107385e-05, "loss": 0.13, "step": 2717 }, { "epoch": 2.49, "learning_rate": 1.61265485886785e-05, "loss": 0.121, "step": 2718 }, { "epoch": 2.49, "learning_rate": 1.6123615306605863e-05, "loss": 0.1282, "step": 2719 }, { "epoch": 2.49, "learning_rate": 1.612068118129339e-05, "loss": 0.1346, "step": 2720 }, { "epoch": 2.49, "learning_rate": 1.6117746213145134e-05, "loss": 0.1155, "step": 2721 }, { "epoch": 2.49, "learning_rate": 1.6114810402565237e-05, "loss": 0.1297, "step": 2722 }, { "epoch": 2.5, "learning_rate": 1.6111873749957964e-05, "loss": 0.1367, "step": 2723 }, { "epoch": 2.5, "learning_rate": 1.6108936255727716e-05, "loss": 0.1305, "step": 2724 }, { "epoch": 2.5, "learning_rate": 1.6105997920278983e-05, "loss": 0.1397, "step": 2725 }, { "epoch": 2.5, "learning_rate": 1.6103058744016387e-05, "loss": 0.135, "step": 2726 }, { "epoch": 2.5, "learning_rate": 1.610011872734466e-05, "loss": 0.1168, "step": 2727 }, { "epoch": 2.5, "learning_rate": 1.609717787066865e-05, "loss": 0.1386, "step": 2728 }, { "epoch": 2.5, "learning_rate": 1.6094236174393324e-05, "loss": 0.1191, "step": 2729 }, { "epoch": 2.5, "learning_rate": 1.6091293638923756e-05, "loss": 0.1306, "step": 2730 }, { "epoch": 2.5, "learning_rate": 1.6088350264665155e-05, "loss": 0.1292, "step": 2731 }, { "epoch": 2.5, "learning_rate": 1.6085406052022815e-05, "loss": 0.1265, "step": 2732 }, { "epoch": 2.51, "learning_rate": 1.608246100140218e-05, "loss": 0.1321, "step": 2733 }, { "epoch": 2.51, "learning_rate": 1.6079515113208778e-05, "loss": 0.1235, "step": 2734 }, { "epoch": 2.51, "learning_rate": 1.6076568387848272e-05, "loss": 0.1307, "step": 2735 }, { "epoch": 2.51, "learning_rate": 1.6073620825726436e-05, "loss": 0.1309, "step": 2736 }, { "epoch": 2.51, "learning_rate": 1.607067242724916e-05, "loss": 0.1308, "step": 2737 }, { "epoch": 2.51, "learning_rate": 1.606772319282244e-05, "loss": 0.1231, "step": 2738 }, { "epoch": 2.51, "learning_rate": 1.60647731228524e-05, "loss": 0.1219, "step": 2739 }, { "epoch": 2.51, "learning_rate": 1.6061822217745272e-05, "loss": 0.119, "step": 2740 }, { "epoch": 2.51, "learning_rate": 1.60588704779074e-05, "loss": 0.1198, "step": 2741 }, { "epoch": 2.51, "learning_rate": 1.605591790374526e-05, "loss": 0.1205, "step": 2742 }, { "epoch": 2.51, "learning_rate": 1.6052964495665414e-05, "loss": 0.125, "step": 2743 }, { "epoch": 2.52, "learning_rate": 1.6050010254074564e-05, "loss": 0.1242, "step": 2744 }, { "epoch": 2.52, "learning_rate": 1.6047055179379512e-05, "loss": 0.1306, "step": 2745 }, { "epoch": 2.52, "learning_rate": 1.6044099271987194e-05, "loss": 0.1243, "step": 2746 }, { "epoch": 2.52, "learning_rate": 1.604114253230463e-05, "loss": 0.1203, "step": 2747 }, { "epoch": 2.52, "learning_rate": 1.603818496073898e-05, "loss": 0.1242, "step": 2748 }, { "epoch": 2.52, "learning_rate": 1.6035226557697514e-05, "loss": 0.1281, "step": 2749 }, { "epoch": 2.52, "learning_rate": 1.60322673235876e-05, "loss": 0.118, "step": 2750 }, { "epoch": 2.52, "learning_rate": 1.6029307258816753e-05, "loss": 0.1278, "step": 2751 }, { "epoch": 2.52, "learning_rate": 1.6026346363792565e-05, "loss": 0.1365, "step": 2752 }, { "epoch": 2.52, "learning_rate": 1.602338463892277e-05, "loss": 0.1334, "step": 2753 }, { "epoch": 2.52, "learning_rate": 1.6020422084615197e-05, "loss": 0.128, "step": 2754 }, { "epoch": 2.53, "learning_rate": 1.6017458701277807e-05, "loss": 0.1202, "step": 2755 }, { "epoch": 2.53, "learning_rate": 1.6014494489318663e-05, "loss": 0.1233, "step": 2756 }, { "epoch": 2.53, "learning_rate": 1.6011529449145946e-05, "loss": 0.1264, "step": 2757 }, { "epoch": 2.53, "learning_rate": 1.6008563581167953e-05, "loss": 0.1196, "step": 2758 }, { "epoch": 2.53, "learning_rate": 1.6005596885793087e-05, "loss": 0.1317, "step": 2759 }, { "epoch": 2.53, "learning_rate": 1.600262936342988e-05, "loss": 0.1195, "step": 2760 }, { "epoch": 2.53, "learning_rate": 1.5999661014486956e-05, "loss": 0.1221, "step": 2761 }, { "epoch": 2.53, "learning_rate": 1.5996691839373077e-05, "loss": 0.1184, "step": 2762 }, { "epoch": 2.53, "learning_rate": 1.59937218384971e-05, "loss": 0.1099, "step": 2763 }, { "epoch": 2.53, "learning_rate": 1.5990751012268007e-05, "loss": 0.129, "step": 2764 }, { "epoch": 2.53, "learning_rate": 1.5987779361094888e-05, "loss": 0.127, "step": 2765 }, { "epoch": 2.54, "learning_rate": 1.598480688538695e-05, "loss": 0.1245, "step": 2766 }, { "epoch": 2.54, "learning_rate": 1.5981833585553504e-05, "loss": 0.1331, "step": 2767 }, { "epoch": 2.54, "learning_rate": 1.5978859462003992e-05, "loss": 0.1265, "step": 2768 }, { "epoch": 2.54, "learning_rate": 1.5975884515147956e-05, "loss": 0.1226, "step": 2769 }, { "epoch": 2.54, "learning_rate": 1.5972908745395054e-05, "loss": 0.122, "step": 2770 }, { "epoch": 2.54, "learning_rate": 1.5969932153155055e-05, "loss": 0.1321, "step": 2771 }, { "epoch": 2.54, "learning_rate": 1.5966954738837847e-05, "loss": 0.1319, "step": 2772 }, { "epoch": 2.54, "learning_rate": 1.5963976502853434e-05, "loss": 0.1307, "step": 2773 }, { "epoch": 2.54, "learning_rate": 1.5960997445611925e-05, "loss": 0.1189, "step": 2774 }, { "epoch": 2.54, "learning_rate": 1.5958017567523535e-05, "loss": 0.1284, "step": 2775 }, { "epoch": 2.54, "learning_rate": 1.595503686899862e-05, "loss": 0.1279, "step": 2776 }, { "epoch": 2.55, "learning_rate": 1.5952055350447616e-05, "loss": 0.1167, "step": 2777 }, { "epoch": 2.55, "learning_rate": 1.5949073012281092e-05, "loss": 0.1288, "step": 2778 }, { "epoch": 2.55, "learning_rate": 1.5946089854909726e-05, "loss": 0.1264, "step": 2779 }, { "epoch": 2.55, "learning_rate": 1.5943105878744306e-05, "loss": 0.1342, "step": 2780 }, { "epoch": 2.55, "learning_rate": 1.5940121084195737e-05, "loss": 0.1262, "step": 2781 }, { "epoch": 2.55, "learning_rate": 1.5937135471675033e-05, "loss": 0.1216, "step": 2782 }, { "epoch": 2.55, "learning_rate": 1.5934149041593317e-05, "loss": 0.1279, "step": 2783 }, { "epoch": 2.55, "learning_rate": 1.593116179436183e-05, "loss": 0.1285, "step": 2784 }, { "epoch": 2.55, "learning_rate": 1.5928173730391934e-05, "loss": 0.1288, "step": 2785 }, { "epoch": 2.55, "learning_rate": 1.5925184850095084e-05, "loss": 0.1267, "step": 2786 }, { "epoch": 2.55, "learning_rate": 1.5922195153882862e-05, "loss": 0.1249, "step": 2787 }, { "epoch": 2.56, "learning_rate": 1.5919204642166956e-05, "loss": 0.1142, "step": 2788 }, { "epoch": 2.56, "learning_rate": 1.591621331535917e-05, "loss": 0.1256, "step": 2789 }, { "epoch": 2.56, "learning_rate": 1.5913221173871418e-05, "loss": 0.1248, "step": 2790 }, { "epoch": 2.56, "learning_rate": 1.5910228218115723e-05, "loss": 0.1106, "step": 2791 }, { "epoch": 2.56, "learning_rate": 1.5907234448504236e-05, "loss": 0.1282, "step": 2792 }, { "epoch": 2.56, "learning_rate": 1.5904239865449188e-05, "loss": 0.1321, "step": 2793 }, { "epoch": 2.56, "learning_rate": 1.5901244469362962e-05, "loss": 0.1197, "step": 2794 }, { "epoch": 2.56, "learning_rate": 1.589824826065802e-05, "loss": 0.1164, "step": 2795 }, { "epoch": 2.56, "learning_rate": 1.5895251239746953e-05, "loss": 0.133, "step": 2796 }, { "epoch": 2.56, "learning_rate": 1.589225340704246e-05, "loss": 0.132, "step": 2797 }, { "epoch": 2.56, "learning_rate": 1.5889254762957353e-05, "loss": 0.1133, "step": 2798 }, { "epoch": 2.57, "learning_rate": 1.5886255307904548e-05, "loss": 0.1209, "step": 2799 }, { "epoch": 2.57, "learning_rate": 1.5883255042297083e-05, "loss": 0.1204, "step": 2800 }, { "epoch": 2.57, "learning_rate": 1.5880253966548106e-05, "loss": 0.1334, "step": 2801 }, { "epoch": 2.57, "learning_rate": 1.5877252081070865e-05, "loss": 0.1274, "step": 2802 }, { "epoch": 2.57, "learning_rate": 1.587424938627874e-05, "loss": 0.1213, "step": 2803 }, { "epoch": 2.57, "learning_rate": 1.58712458825852e-05, "loss": 0.1134, "step": 2804 }, { "epoch": 2.57, "learning_rate": 1.5868241570403846e-05, "loss": 0.118, "step": 2805 }, { "epoch": 2.57, "learning_rate": 1.5865236450148372e-05, "loss": 0.1181, "step": 2806 }, { "epoch": 2.57, "learning_rate": 1.58622305222326e-05, "loss": 0.1308, "step": 2807 }, { "epoch": 2.57, "learning_rate": 1.585922378707045e-05, "loss": 0.1168, "step": 2808 }, { "epoch": 2.57, "learning_rate": 1.585621624507596e-05, "loss": 0.122, "step": 2809 }, { "epoch": 2.58, "learning_rate": 1.5853207896663276e-05, "loss": 0.1156, "step": 2810 }, { "epoch": 2.58, "learning_rate": 1.5850198742246658e-05, "loss": 0.1172, "step": 2811 }, { "epoch": 2.58, "learning_rate": 1.5847188782240473e-05, "loss": 0.1241, "step": 2812 }, { "epoch": 2.58, "learning_rate": 1.5844178017059203e-05, "loss": 0.1257, "step": 2813 }, { "epoch": 2.58, "learning_rate": 1.584116644711744e-05, "loss": 0.123, "step": 2814 }, { "epoch": 2.58, "learning_rate": 1.5838154072829886e-05, "loss": 0.1333, "step": 2815 }, { "epoch": 2.58, "learning_rate": 1.583514089461135e-05, "loss": 0.1336, "step": 2816 }, { "epoch": 2.58, "learning_rate": 1.583212691287676e-05, "loss": 0.1208, "step": 2817 }, { "epoch": 2.58, "learning_rate": 1.582911212804115e-05, "loss": 0.1279, "step": 2818 }, { "epoch": 2.58, "learning_rate": 1.5826096540519665e-05, "loss": 0.1262, "step": 2819 }, { "epoch": 2.58, "learning_rate": 1.5823080150727558e-05, "loss": 0.1218, "step": 2820 }, { "epoch": 2.59, "learning_rate": 1.582006295908019e-05, "loss": 0.13, "step": 2821 }, { "epoch": 2.59, "learning_rate": 1.5817044965993048e-05, "loss": 0.1198, "step": 2822 }, { "epoch": 2.59, "learning_rate": 1.581402617188171e-05, "loss": 0.1255, "step": 2823 }, { "epoch": 2.59, "learning_rate": 1.5811006577161882e-05, "loss": 0.1178, "step": 2824 }, { "epoch": 2.59, "learning_rate": 1.580798618224936e-05, "loss": 0.1213, "step": 2825 }, { "epoch": 2.59, "learning_rate": 1.5804964987560064e-05, "loss": 0.1258, "step": 2826 }, { "epoch": 2.59, "learning_rate": 1.5801942993510026e-05, "loss": 0.1292, "step": 2827 }, { "epoch": 2.59, "learning_rate": 1.5798920200515382e-05, "loss": 0.1186, "step": 2828 }, { "epoch": 2.59, "learning_rate": 1.5795896608992377e-05, "loss": 0.1325, "step": 2829 }, { "epoch": 2.59, "learning_rate": 1.5792872219357372e-05, "loss": 0.1164, "step": 2830 }, { "epoch": 2.59, "learning_rate": 1.5789847032026832e-05, "loss": 0.1231, "step": 2831 }, { "epoch": 2.6, "learning_rate": 1.5786821047417335e-05, "loss": 0.1355, "step": 2832 }, { "epoch": 2.6, "learning_rate": 1.5783794265945567e-05, "loss": 0.1216, "step": 2833 }, { "epoch": 2.6, "learning_rate": 1.5780766688028322e-05, "loss": 0.1215, "step": 2834 }, { "epoch": 2.6, "learning_rate": 1.5777738314082514e-05, "loss": 0.133, "step": 2835 }, { "epoch": 2.6, "learning_rate": 1.577470914452515e-05, "loss": 0.1226, "step": 2836 }, { "epoch": 2.6, "learning_rate": 1.5771679179773364e-05, "loss": 0.1187, "step": 2837 }, { "epoch": 2.6, "learning_rate": 1.5768648420244387e-05, "loss": 0.1335, "step": 2838 }, { "epoch": 2.6, "learning_rate": 1.5765616866355562e-05, "loss": 0.1333, "step": 2839 }, { "epoch": 2.6, "learning_rate": 1.5762584518524348e-05, "loss": 0.127, "step": 2840 }, { "epoch": 2.6, "learning_rate": 1.57595513771683e-05, "loss": 0.1187, "step": 2841 }, { "epoch": 2.6, "learning_rate": 1.57565174427051e-05, "loss": 0.1213, "step": 2842 }, { "epoch": 2.61, "learning_rate": 1.575348271555252e-05, "loss": 0.1063, "step": 2843 }, { "epoch": 2.61, "learning_rate": 1.575044719612846e-05, "loss": 0.1275, "step": 2844 }, { "epoch": 2.61, "learning_rate": 1.5747410884850914e-05, "loss": 0.1299, "step": 2845 }, { "epoch": 2.61, "learning_rate": 1.5744373782137993e-05, "loss": 0.1273, "step": 2846 }, { "epoch": 2.61, "learning_rate": 1.5741335888407912e-05, "loss": 0.1344, "step": 2847 }, { "epoch": 2.61, "learning_rate": 1.5738297204079005e-05, "loss": 0.1197, "step": 2848 }, { "epoch": 2.61, "learning_rate": 1.5735257729569702e-05, "loss": 0.1332, "step": 2849 }, { "epoch": 2.61, "learning_rate": 1.5732217465298547e-05, "loss": 0.133, "step": 2850 }, { "epoch": 2.61, "learning_rate": 1.57291764116842e-05, "loss": 0.1271, "step": 2851 }, { "epoch": 2.61, "learning_rate": 1.5726134569145413e-05, "loss": 0.1175, "step": 2852 }, { "epoch": 2.62, "learning_rate": 1.5723091938101065e-05, "loss": 0.1239, "step": 2853 }, { "epoch": 2.62, "learning_rate": 1.572004851897013e-05, "loss": 0.1299, "step": 2854 }, { "epoch": 2.62, "learning_rate": 1.57170043121717e-05, "loss": 0.1192, "step": 2855 }, { "epoch": 2.62, "learning_rate": 1.571395931812497e-05, "loss": 0.1192, "step": 2856 }, { "epoch": 2.62, "learning_rate": 1.5710913537249242e-05, "loss": 0.1197, "step": 2857 }, { "epoch": 2.62, "learning_rate": 1.5707866969963933e-05, "loss": 0.1275, "step": 2858 }, { "epoch": 2.62, "learning_rate": 1.570481961668856e-05, "loss": 0.1243, "step": 2859 }, { "epoch": 2.62, "learning_rate": 1.5701771477842755e-05, "loss": 0.1174, "step": 2860 }, { "epoch": 2.62, "learning_rate": 1.569872255384625e-05, "loss": 0.1279, "step": 2861 }, { "epoch": 2.62, "learning_rate": 1.5695672845118903e-05, "loss": 0.1273, "step": 2862 }, { "epoch": 2.62, "learning_rate": 1.5692622352080662e-05, "loss": 0.1231, "step": 2863 }, { "epoch": 2.63, "learning_rate": 1.5689571075151582e-05, "loss": 0.1328, "step": 2864 }, { "epoch": 2.63, "learning_rate": 1.5686519014751843e-05, "loss": 0.1147, "step": 2865 }, { "epoch": 2.63, "learning_rate": 1.568346617130172e-05, "loss": 0.1245, "step": 2866 }, { "epoch": 2.63, "learning_rate": 1.5680412545221596e-05, "loss": 0.1276, "step": 2867 }, { "epoch": 2.63, "learning_rate": 1.5677358136931958e-05, "loss": 0.1312, "step": 2868 }, { "epoch": 2.63, "learning_rate": 1.567430294685342e-05, "loss": 0.1147, "step": 2869 }, { "epoch": 2.63, "learning_rate": 1.5671246975406685e-05, "loss": 0.1231, "step": 2870 }, { "epoch": 2.63, "learning_rate": 1.5668190223012567e-05, "loss": 0.1213, "step": 2871 }, { "epoch": 2.63, "learning_rate": 1.5665132690091994e-05, "loss": 0.11, "step": 2872 }, { "epoch": 2.63, "learning_rate": 1.5662074377065992e-05, "loss": 0.1227, "step": 2873 }, { "epoch": 2.63, "learning_rate": 1.5659015284355704e-05, "loss": 0.1233, "step": 2874 }, { "epoch": 2.64, "learning_rate": 1.5655955412382375e-05, "loss": 0.1235, "step": 2875 }, { "epoch": 2.64, "learning_rate": 1.565289476156736e-05, "loss": 0.1241, "step": 2876 }, { "epoch": 2.64, "learning_rate": 1.5649833332332117e-05, "loss": 0.1266, "step": 2877 }, { "epoch": 2.64, "learning_rate": 1.5646771125098213e-05, "loss": 0.1228, "step": 2878 }, { "epoch": 2.64, "learning_rate": 1.5643708140287327e-05, "loss": 0.1303, "step": 2879 }, { "epoch": 2.64, "learning_rate": 1.5640644378321236e-05, "loss": 0.1234, "step": 2880 }, { "epoch": 2.64, "learning_rate": 1.5637579839621833e-05, "loss": 0.1271, "step": 2881 }, { "epoch": 2.64, "learning_rate": 1.5634514524611112e-05, "loss": 0.1157, "step": 2882 }, { "epoch": 2.64, "learning_rate": 1.5631448433711175e-05, "loss": 0.1243, "step": 2883 }, { "epoch": 2.64, "learning_rate": 1.5628381567344236e-05, "loss": 0.1194, "step": 2884 }, { "epoch": 2.64, "learning_rate": 1.5625313925932606e-05, "loss": 0.1257, "step": 2885 }, { "epoch": 2.65, "learning_rate": 1.5622245509898714e-05, "loss": 0.1321, "step": 2886 }, { "epoch": 2.65, "learning_rate": 1.5619176319665083e-05, "loss": 0.1211, "step": 2887 }, { "epoch": 2.65, "learning_rate": 1.561610635565436e-05, "loss": 0.122, "step": 2888 }, { "epoch": 2.65, "learning_rate": 1.5613035618289274e-05, "loss": 0.1352, "step": 2889 }, { "epoch": 2.65, "learning_rate": 1.5609964107992685e-05, "loss": 0.1255, "step": 2890 }, { "epoch": 2.65, "learning_rate": 1.5606891825187547e-05, "loss": 0.1127, "step": 2891 }, { "epoch": 2.65, "learning_rate": 1.5603818770296923e-05, "loss": 0.1283, "step": 2892 }, { "epoch": 2.65, "learning_rate": 1.5600744943743976e-05, "loss": 0.13, "step": 2893 }, { "epoch": 2.65, "learning_rate": 1.5597670345951982e-05, "loss": 0.1308, "step": 2894 }, { "epoch": 2.65, "learning_rate": 1.5594594977344333e-05, "loss": 0.1281, "step": 2895 }, { "epoch": 2.65, "learning_rate": 1.5591518838344507e-05, "loss": 0.1334, "step": 2896 }, { "epoch": 2.66, "learning_rate": 1.5588441929376097e-05, "loss": 0.1218, "step": 2897 }, { "epoch": 2.66, "learning_rate": 1.5585364250862803e-05, "loss": 0.1212, "step": 2898 }, { "epoch": 2.66, "learning_rate": 1.5582285803228436e-05, "loss": 0.1231, "step": 2899 }, { "epoch": 2.66, "learning_rate": 1.55792065868969e-05, "loss": 0.1243, "step": 2900 }, { "epoch": 2.66, "learning_rate": 1.5576126602292212e-05, "loss": 0.1285, "step": 2901 }, { "epoch": 2.66, "learning_rate": 1.55730458498385e-05, "loss": 0.1273, "step": 2902 }, { "epoch": 2.66, "learning_rate": 1.5569964329959994e-05, "loss": 0.1144, "step": 2903 }, { "epoch": 2.66, "learning_rate": 1.5566882043081023e-05, "loss": 0.1161, "step": 2904 }, { "epoch": 2.66, "learning_rate": 1.5563798989626027e-05, "loss": 0.1201, "step": 2905 }, { "epoch": 2.66, "learning_rate": 1.556071517001956e-05, "loss": 0.1229, "step": 2906 }, { "epoch": 2.66, "learning_rate": 1.5557630584686258e-05, "loss": 0.1137, "step": 2907 }, { "epoch": 2.67, "learning_rate": 1.5554545234050892e-05, "loss": 0.1189, "step": 2908 }, { "epoch": 2.67, "learning_rate": 1.5551459118538313e-05, "loss": 0.1368, "step": 2909 }, { "epoch": 2.67, "learning_rate": 1.5548372238573496e-05, "loss": 0.1115, "step": 2910 }, { "epoch": 2.67, "learning_rate": 1.554528459458151e-05, "loss": 0.1223, "step": 2911 }, { "epoch": 2.67, "learning_rate": 1.5542196186987535e-05, "loss": 0.1116, "step": 2912 }, { "epoch": 2.67, "learning_rate": 1.553910701621685e-05, "loss": 0.1144, "step": 2913 }, { "epoch": 2.67, "learning_rate": 1.5536017082694846e-05, "loss": 0.1283, "step": 2914 }, { "epoch": 2.67, "learning_rate": 1.5532926386847017e-05, "loss": 0.1221, "step": 2915 }, { "epoch": 2.67, "learning_rate": 1.5529834929098955e-05, "loss": 0.1287, "step": 2916 }, { "epoch": 2.67, "learning_rate": 1.5526742709876365e-05, "loss": 0.1259, "step": 2917 }, { "epoch": 2.67, "learning_rate": 1.552364972960506e-05, "loss": 0.1134, "step": 2918 }, { "epoch": 2.68, "learning_rate": 1.5520555988710946e-05, "loss": 0.1289, "step": 2919 }, { "epoch": 2.68, "learning_rate": 1.5517461487620047e-05, "loss": 0.1255, "step": 2920 }, { "epoch": 2.68, "learning_rate": 1.5514366226758478e-05, "loss": 0.118, "step": 2921 }, { "epoch": 2.68, "learning_rate": 1.5511270206552467e-05, "loss": 0.1208, "step": 2922 }, { "epoch": 2.68, "learning_rate": 1.5508173427428347e-05, "loss": 0.1187, "step": 2923 }, { "epoch": 2.68, "learning_rate": 1.5505075889812553e-05, "loss": 0.1165, "step": 2924 }, { "epoch": 2.68, "learning_rate": 1.5501977594131625e-05, "loss": 0.1252, "step": 2925 }, { "epoch": 2.68, "learning_rate": 1.549887854081221e-05, "loss": 0.117, "step": 2926 }, { "epoch": 2.68, "learning_rate": 1.549577873028105e-05, "loss": 0.1189, "step": 2927 }, { "epoch": 2.68, "learning_rate": 1.5492678162965004e-05, "loss": 0.1171, "step": 2928 }, { "epoch": 2.68, "learning_rate": 1.5489576839291028e-05, "loss": 0.1201, "step": 2929 }, { "epoch": 2.69, "learning_rate": 1.5486474759686182e-05, "loss": 0.1175, "step": 2930 }, { "epoch": 2.69, "learning_rate": 1.5483371924577633e-05, "loss": 0.1134, "step": 2931 }, { "epoch": 2.69, "learning_rate": 1.548026833439265e-05, "loss": 0.1203, "step": 2932 }, { "epoch": 2.69, "learning_rate": 1.5477163989558608e-05, "loss": 0.116, "step": 2933 }, { "epoch": 2.69, "learning_rate": 1.547405889050298e-05, "loss": 0.1203, "step": 2934 }, { "epoch": 2.69, "learning_rate": 1.5470953037653348e-05, "loss": 0.1165, "step": 2935 }, { "epoch": 2.69, "learning_rate": 1.5467846431437404e-05, "loss": 0.1269, "step": 2936 }, { "epoch": 2.69, "learning_rate": 1.5464739072282927e-05, "loss": 0.1255, "step": 2937 }, { "epoch": 2.69, "learning_rate": 1.5461630960617822e-05, "loss": 0.1166, "step": 2938 }, { "epoch": 2.69, "learning_rate": 1.545852209687007e-05, "loss": 0.125, "step": 2939 }, { "epoch": 2.69, "learning_rate": 1.5455412481467782e-05, "loss": 0.1135, "step": 2940 }, { "epoch": 2.7, "learning_rate": 1.5452302114839155e-05, "loss": 0.1198, "step": 2941 }, { "epoch": 2.7, "learning_rate": 1.5449190997412497e-05, "loss": 0.1325, "step": 2942 }, { "epoch": 2.7, "learning_rate": 1.5446079129616224e-05, "loss": 0.1283, "step": 2943 }, { "epoch": 2.7, "learning_rate": 1.544296651187884e-05, "loss": 0.1137, "step": 2944 }, { "epoch": 2.7, "learning_rate": 1.5439853144628967e-05, "loss": 0.1184, "step": 2945 }, { "epoch": 2.7, "learning_rate": 1.5436739028295326e-05, "loss": 0.1295, "step": 2946 }, { "epoch": 2.7, "learning_rate": 1.5433624163306736e-05, "loss": 0.1151, "step": 2947 }, { "epoch": 2.7, "learning_rate": 1.5430508550092123e-05, "loss": 0.1202, "step": 2948 }, { "epoch": 2.7, "learning_rate": 1.542739218908052e-05, "loss": 0.1131, "step": 2949 }, { "epoch": 2.7, "learning_rate": 1.5424275080701056e-05, "loss": 0.1092, "step": 2950 }, { "epoch": 2.7, "learning_rate": 1.5421157225382968e-05, "loss": 0.1182, "step": 2951 }, { "epoch": 2.71, "learning_rate": 1.5418038623555594e-05, "loss": 0.1162, "step": 2952 }, { "epoch": 2.71, "learning_rate": 1.541491927564837e-05, "loss": 0.1155, "step": 2953 }, { "epoch": 2.71, "learning_rate": 1.5411799182090844e-05, "loss": 0.1107, "step": 2954 }, { "epoch": 2.71, "learning_rate": 1.5408678343312662e-05, "loss": 0.123, "step": 2955 }, { "epoch": 2.71, "learning_rate": 1.540555675974357e-05, "loss": 0.1263, "step": 2956 }, { "epoch": 2.71, "learning_rate": 1.5402434431813422e-05, "loss": 0.1248, "step": 2957 }, { "epoch": 2.71, "learning_rate": 1.539931135995217e-05, "loss": 0.1266, "step": 2958 }, { "epoch": 2.71, "learning_rate": 1.5396187544589874e-05, "loss": 0.1293, "step": 2959 }, { "epoch": 2.71, "learning_rate": 1.5393062986156685e-05, "loss": 0.1291, "step": 2960 }, { "epoch": 2.71, "learning_rate": 1.538993768508287e-05, "loss": 0.1214, "step": 2961 }, { "epoch": 2.71, "learning_rate": 1.5386811641798785e-05, "loss": 0.1269, "step": 2962 }, { "epoch": 2.72, "learning_rate": 1.5383684856734906e-05, "loss": 0.1175, "step": 2963 }, { "epoch": 2.72, "learning_rate": 1.5380557330321794e-05, "loss": 0.1232, "step": 2964 }, { "epoch": 2.72, "learning_rate": 1.5377429062990122e-05, "loss": 0.1172, "step": 2965 }, { "epoch": 2.72, "learning_rate": 1.5374300055170657e-05, "loss": 0.1203, "step": 2966 }, { "epoch": 2.72, "learning_rate": 1.5371170307294274e-05, "loss": 0.1227, "step": 2967 }, { "epoch": 2.72, "learning_rate": 1.536803981979195e-05, "loss": 0.1222, "step": 2968 }, { "epoch": 2.72, "learning_rate": 1.536490859309476e-05, "loss": 0.1237, "step": 2969 }, { "epoch": 2.72, "learning_rate": 1.536177662763389e-05, "loss": 0.1238, "step": 2970 }, { "epoch": 2.72, "learning_rate": 1.5358643923840614e-05, "loss": 0.1232, "step": 2971 }, { "epoch": 2.72, "learning_rate": 1.5355510482146315e-05, "loss": 0.1162, "step": 2972 }, { "epoch": 2.73, "learning_rate": 1.535237630298248e-05, "loss": 0.1213, "step": 2973 }, { "epoch": 2.73, "learning_rate": 1.534924138678069e-05, "loss": 0.1157, "step": 2974 }, { "epoch": 2.73, "learning_rate": 1.534610573397264e-05, "loss": 0.1215, "step": 2975 }, { "epoch": 2.73, "learning_rate": 1.5342969344990106e-05, "loss": 0.1196, "step": 2976 }, { "epoch": 2.73, "learning_rate": 1.533983222026499e-05, "loss": 0.1198, "step": 2977 }, { "epoch": 2.73, "learning_rate": 1.533669436022928e-05, "loss": 0.1134, "step": 2978 }, { "epoch": 2.73, "learning_rate": 1.533355576531507e-05, "loss": 0.1244, "step": 2979 }, { "epoch": 2.73, "learning_rate": 1.5330416435954548e-05, "loss": 0.1224, "step": 2980 }, { "epoch": 2.73, "learning_rate": 1.532727637258001e-05, "loss": 0.1207, "step": 2981 }, { "epoch": 2.73, "learning_rate": 1.532413557562386e-05, "loss": 0.1251, "step": 2982 }, { "epoch": 2.73, "learning_rate": 1.5320994045518583e-05, "loss": 0.1295, "step": 2983 }, { "epoch": 2.74, "learning_rate": 1.5317851782696787e-05, "loss": 0.1097, "step": 2984 }, { "epoch": 2.74, "learning_rate": 1.5314708787591166e-05, "loss": 0.1276, "step": 2985 }, { "epoch": 2.74, "learning_rate": 1.531156506063452e-05, "loss": 0.1159, "step": 2986 }, { "epoch": 2.74, "learning_rate": 1.530842060225975e-05, "loss": 0.1123, "step": 2987 }, { "epoch": 2.74, "learning_rate": 1.530527541289986e-05, "loss": 0.1285, "step": 2988 }, { "epoch": 2.74, "learning_rate": 1.5302129492987944e-05, "loss": 0.1245, "step": 2989 }, { "epoch": 2.74, "learning_rate": 1.5298982842957216e-05, "loss": 0.1274, "step": 2990 }, { "epoch": 2.74, "learning_rate": 1.529583546324097e-05, "loss": 0.1211, "step": 2991 }, { "epoch": 2.74, "learning_rate": 1.529268735427261e-05, "loss": 0.1119, "step": 2992 }, { "epoch": 2.74, "learning_rate": 1.528953851648564e-05, "loss": 0.1341, "step": 2993 }, { "epoch": 2.74, "learning_rate": 1.5286388950313667e-05, "loss": 0.1176, "step": 2994 }, { "epoch": 2.75, "learning_rate": 1.52832386561904e-05, "loss": 0.1252, "step": 2995 }, { "epoch": 2.75, "learning_rate": 1.528008763454963e-05, "loss": 0.1187, "step": 2996 }, { "epoch": 2.75, "learning_rate": 1.527693588582528e-05, "loss": 0.127, "step": 2997 }, { "epoch": 2.75, "learning_rate": 1.5273783410451336e-05, "loss": 0.1145, "step": 2998 }, { "epoch": 2.75, "learning_rate": 1.5270630208861917e-05, "loss": 0.1278, "step": 2999 }, { "epoch": 2.75, "learning_rate": 1.5267476281491224e-05, "loss": 0.1163, "step": 3000 }, { "epoch": 2.75, "learning_rate": 1.526432162877356e-05, "loss": 0.1355, "step": 3001 }, { "epoch": 2.75, "learning_rate": 1.5261166251143333e-05, "loss": 0.1256, "step": 3002 }, { "epoch": 2.75, "learning_rate": 1.5258010149035043e-05, "loss": 0.1171, "step": 3003 }, { "epoch": 2.75, "learning_rate": 1.5254853322883301e-05, "loss": 0.1216, "step": 3004 }, { "epoch": 2.75, "learning_rate": 1.5251695773122807e-05, "loss": 0.1217, "step": 3005 }, { "epoch": 2.76, "learning_rate": 1.5248537500188366e-05, "loss": 0.126, "step": 3006 }, { "epoch": 2.76, "learning_rate": 1.5245378504514881e-05, "loss": 0.1332, "step": 3007 }, { "epoch": 2.76, "learning_rate": 1.5242218786537355e-05, "loss": 0.1198, "step": 3008 }, { "epoch": 2.76, "learning_rate": 1.5239058346690892e-05, "loss": 0.1108, "step": 3009 }, { "epoch": 2.76, "learning_rate": 1.523589718541069e-05, "loss": 0.1281, "step": 3010 }, { "epoch": 2.76, "learning_rate": 1.5232735303132058e-05, "loss": 0.1162, "step": 3011 }, { "epoch": 2.76, "learning_rate": 1.5229572700290387e-05, "loss": 0.1118, "step": 3012 }, { "epoch": 2.76, "learning_rate": 1.5226409377321182e-05, "loss": 0.1248, "step": 3013 }, { "epoch": 2.76, "learning_rate": 1.5223245334660044e-05, "loss": 0.1139, "step": 3014 }, { "epoch": 2.76, "learning_rate": 1.5220080572742665e-05, "loss": 0.1238, "step": 3015 }, { "epoch": 2.76, "learning_rate": 1.5216915092004847e-05, "loss": 0.1132, "step": 3016 }, { "epoch": 2.77, "learning_rate": 1.5213748892882483e-05, "loss": 0.1136, "step": 3017 }, { "epoch": 2.77, "learning_rate": 1.5210581975811569e-05, "loss": 0.1193, "step": 3018 }, { "epoch": 2.77, "learning_rate": 1.5207414341228198e-05, "loss": 0.119, "step": 3019 }, { "epoch": 2.77, "learning_rate": 1.5204245989568564e-05, "loss": 0.1142, "step": 3020 }, { "epoch": 2.77, "learning_rate": 1.5201076921268958e-05, "loss": 0.1158, "step": 3021 }, { "epoch": 2.77, "learning_rate": 1.5197907136765768e-05, "loss": 0.1193, "step": 3022 }, { "epoch": 2.77, "learning_rate": 1.5194736636495487e-05, "loss": 0.1177, "step": 3023 }, { "epoch": 2.77, "learning_rate": 1.5191565420894696e-05, "loss": 0.132, "step": 3024 }, { "epoch": 2.77, "learning_rate": 1.5188393490400088e-05, "loss": 0.1278, "step": 3025 }, { "epoch": 2.77, "learning_rate": 1.5185220845448441e-05, "loss": 0.1165, "step": 3026 }, { "epoch": 2.77, "learning_rate": 1.5182047486476643e-05, "loss": 0.1207, "step": 3027 }, { "epoch": 2.78, "learning_rate": 1.5178873413921666e-05, "loss": 0.1221, "step": 3028 }, { "epoch": 2.78, "learning_rate": 1.51756986282206e-05, "loss": 0.1119, "step": 3029 }, { "epoch": 2.78, "learning_rate": 1.5172523129810615e-05, "loss": 0.132, "step": 3030 }, { "epoch": 2.78, "learning_rate": 1.5169346919128986e-05, "loss": 0.1137, "step": 3031 }, { "epoch": 2.78, "learning_rate": 1.5166169996613093e-05, "loss": 0.1155, "step": 3032 }, { "epoch": 2.78, "learning_rate": 1.5162992362700406e-05, "loss": 0.1261, "step": 3033 }, { "epoch": 2.78, "learning_rate": 1.5159814017828488e-05, "loss": 0.1254, "step": 3034 }, { "epoch": 2.78, "learning_rate": 1.5156634962435014e-05, "loss": 0.1172, "step": 3035 }, { "epoch": 2.78, "learning_rate": 1.5153455196957746e-05, "loss": 0.1212, "step": 3036 }, { "epoch": 2.78, "learning_rate": 1.5150274721834546e-05, "loss": 0.13, "step": 3037 }, { "epoch": 2.78, "learning_rate": 1.5147093537503375e-05, "loss": 0.1168, "step": 3038 }, { "epoch": 2.79, "learning_rate": 1.5143911644402291e-05, "loss": 0.1059, "step": 3039 }, { "epoch": 2.79, "learning_rate": 1.5140729042969453e-05, "loss": 0.1203, "step": 3040 }, { "epoch": 2.79, "learning_rate": 1.5137545733643115e-05, "loss": 0.1201, "step": 3041 }, { "epoch": 2.79, "learning_rate": 1.5134361716861624e-05, "loss": 0.1339, "step": 3042 }, { "epoch": 2.79, "learning_rate": 1.5131176993063431e-05, "loss": 0.1158, "step": 3043 }, { "epoch": 2.79, "learning_rate": 1.512799156268708e-05, "loss": 0.116, "step": 3044 }, { "epoch": 2.79, "learning_rate": 1.5124805426171215e-05, "loss": 0.1219, "step": 3045 }, { "epoch": 2.79, "learning_rate": 1.512161858395458e-05, "loss": 0.1198, "step": 3046 }, { "epoch": 2.79, "learning_rate": 1.5118431036476007e-05, "loss": 0.1164, "step": 3047 }, { "epoch": 2.79, "learning_rate": 1.5115242784174436e-05, "loss": 0.1159, "step": 3048 }, { "epoch": 2.79, "learning_rate": 1.511205382748889e-05, "loss": 0.1252, "step": 3049 }, { "epoch": 2.8, "learning_rate": 1.5108864166858506e-05, "loss": 0.1227, "step": 3050 }, { "epoch": 2.8, "learning_rate": 1.5105673802722509e-05, "loss": 0.1233, "step": 3051 }, { "epoch": 2.8, "learning_rate": 1.5102482735520218e-05, "loss": 0.1181, "step": 3052 }, { "epoch": 2.8, "learning_rate": 1.5099290965691054e-05, "loss": 0.1095, "step": 3053 }, { "epoch": 2.8, "learning_rate": 1.5096098493674531e-05, "loss": 0.1221, "step": 3054 }, { "epoch": 2.8, "learning_rate": 1.5092905319910268e-05, "loss": 0.1295, "step": 3055 }, { "epoch": 2.8, "learning_rate": 1.5089711444837964e-05, "loss": 0.1385, "step": 3056 }, { "epoch": 2.8, "learning_rate": 1.5086516868897438e-05, "loss": 0.1196, "step": 3057 }, { "epoch": 2.8, "learning_rate": 1.5083321592528584e-05, "loss": 0.1251, "step": 3058 }, { "epoch": 2.8, "learning_rate": 1.5080125616171402e-05, "loss": 0.1165, "step": 3059 }, { "epoch": 2.8, "learning_rate": 1.5076928940265987e-05, "loss": 0.1252, "step": 3060 }, { "epoch": 2.81, "learning_rate": 1.5073731565252532e-05, "loss": 0.1267, "step": 3061 }, { "epoch": 2.81, "learning_rate": 1.5070533491571325e-05, "loss": 0.1186, "step": 3062 }, { "epoch": 2.81, "learning_rate": 1.5067334719662747e-05, "loss": 0.1149, "step": 3063 }, { "epoch": 2.81, "learning_rate": 1.5064135249967286e-05, "loss": 0.119, "step": 3064 }, { "epoch": 2.81, "learning_rate": 1.5060935082925507e-05, "loss": 0.1356, "step": 3065 }, { "epoch": 2.81, "learning_rate": 1.5057734218978098e-05, "loss": 0.1128, "step": 3066 }, { "epoch": 2.81, "learning_rate": 1.505453265856581e-05, "loss": 0.1183, "step": 3067 }, { "epoch": 2.81, "learning_rate": 1.505133040212952e-05, "loss": 0.1215, "step": 3068 }, { "epoch": 2.81, "learning_rate": 1.5048127450110182e-05, "loss": 0.1159, "step": 3069 }, { "epoch": 2.81, "learning_rate": 1.5044923802948853e-05, "loss": 0.1236, "step": 3070 }, { "epoch": 2.81, "learning_rate": 1.5041719461086683e-05, "loss": 0.1183, "step": 3071 }, { "epoch": 2.82, "learning_rate": 1.5038514424964925e-05, "loss": 0.1212, "step": 3072 }, { "epoch": 2.82, "learning_rate": 1.5035308695024914e-05, "loss": 0.1142, "step": 3073 }, { "epoch": 2.82, "learning_rate": 1.50321022717081e-05, "loss": 0.117, "step": 3074 }, { "epoch": 2.82, "learning_rate": 1.5028895155456003e-05, "loss": 0.1194, "step": 3075 }, { "epoch": 2.82, "learning_rate": 1.5025687346710262e-05, "loss": 0.1273, "step": 3076 }, { "epoch": 2.82, "learning_rate": 1.5022478845912601e-05, "loss": 0.1206, "step": 3077 }, { "epoch": 2.82, "learning_rate": 1.5019269653504834e-05, "loss": 0.1222, "step": 3078 }, { "epoch": 2.82, "learning_rate": 1.501605976992888e-05, "loss": 0.115, "step": 3079 }, { "epoch": 2.82, "learning_rate": 1.501284919562675e-05, "loss": 0.1086, "step": 3080 }, { "epoch": 2.82, "learning_rate": 1.500963793104055e-05, "loss": 0.1158, "step": 3081 }, { "epoch": 2.82, "learning_rate": 1.5006425976612477e-05, "loss": 0.1084, "step": 3082 }, { "epoch": 2.83, "learning_rate": 1.5003213332784831e-05, "loss": 0.1221, "step": 3083 }, { "epoch": 2.83, "learning_rate": 1.5000000000000002e-05, "loss": 0.1168, "step": 3084 }, { "epoch": 2.83, "learning_rate": 1.4996785978700472e-05, "loss": 0.1186, "step": 3085 }, { "epoch": 2.83, "learning_rate": 1.4993571269328824e-05, "loss": 0.1309, "step": 3086 }, { "epoch": 2.83, "learning_rate": 1.499035587232773e-05, "loss": 0.1152, "step": 3087 }, { "epoch": 2.83, "learning_rate": 1.4987139788139962e-05, "loss": 0.1072, "step": 3088 }, { "epoch": 2.83, "learning_rate": 1.4983923017208386e-05, "loss": 0.1216, "step": 3089 }, { "epoch": 2.83, "learning_rate": 1.4980705559975957e-05, "loss": 0.1081, "step": 3090 }, { "epoch": 2.83, "learning_rate": 1.497748741688573e-05, "loss": 0.113, "step": 3091 }, { "epoch": 2.83, "learning_rate": 1.4974268588380851e-05, "loss": 0.1096, "step": 3092 }, { "epoch": 2.84, "learning_rate": 1.4971049074904569e-05, "loss": 0.1195, "step": 3093 }, { "epoch": 2.84, "learning_rate": 1.496782887690021e-05, "loss": 0.1124, "step": 3094 }, { "epoch": 2.84, "learning_rate": 1.4964607994811213e-05, "loss": 0.1248, "step": 3095 }, { "epoch": 2.84, "learning_rate": 1.4961386429081098e-05, "loss": 0.1125, "step": 3096 }, { "epoch": 2.84, "learning_rate": 1.4958164180153488e-05, "loss": 0.1147, "step": 3097 }, { "epoch": 2.84, "learning_rate": 1.4954941248472092e-05, "loss": 0.1186, "step": 3098 }, { "epoch": 2.84, "learning_rate": 1.495171763448072e-05, "loss": 0.1186, "step": 3099 }, { "epoch": 2.84, "learning_rate": 1.4948493338623276e-05, "loss": 0.1238, "step": 3100 }, { "epoch": 2.84, "learning_rate": 1.4945268361343747e-05, "loss": 0.1124, "step": 3101 }, { "epoch": 2.84, "learning_rate": 1.494204270308623e-05, "loss": 0.1268, "step": 3102 }, { "epoch": 2.84, "learning_rate": 1.4938816364294903e-05, "loss": 0.1223, "step": 3103 }, { "epoch": 2.85, "learning_rate": 1.493558934541404e-05, "loss": 0.1222, "step": 3104 }, { "epoch": 2.85, "learning_rate": 1.4932361646888021e-05, "loss": 0.1136, "step": 3105 }, { "epoch": 2.85, "learning_rate": 1.4929133269161298e-05, "loss": 0.1247, "step": 3106 }, { "epoch": 2.85, "learning_rate": 1.4925904212678435e-05, "loss": 0.1272, "step": 3107 }, { "epoch": 2.85, "learning_rate": 1.492267447788408e-05, "loss": 0.1252, "step": 3108 }, { "epoch": 2.85, "learning_rate": 1.4919444065222979e-05, "loss": 0.1183, "step": 3109 }, { "epoch": 2.85, "learning_rate": 1.4916212975139966e-05, "loss": 0.1207, "step": 3110 }, { "epoch": 2.85, "learning_rate": 1.4912981208079973e-05, "loss": 0.1247, "step": 3111 }, { "epoch": 2.85, "learning_rate": 1.4909748764488025e-05, "loss": 0.1184, "step": 3112 }, { "epoch": 2.85, "learning_rate": 1.4906515644809238e-05, "loss": 0.1156, "step": 3113 }, { "epoch": 2.85, "learning_rate": 1.4903281849488824e-05, "loss": 0.1112, "step": 3114 }, { "epoch": 2.86, "learning_rate": 1.4900047378972083e-05, "loss": 0.1237, "step": 3115 }, { "epoch": 2.86, "learning_rate": 1.4896812233704412e-05, "loss": 0.1306, "step": 3116 }, { "epoch": 2.86, "learning_rate": 1.4893576414131303e-05, "loss": 0.1131, "step": 3117 }, { "epoch": 2.86, "learning_rate": 1.4890339920698334e-05, "loss": 0.1202, "step": 3118 }, { "epoch": 2.86, "learning_rate": 1.4887102753851182e-05, "loss": 0.1203, "step": 3119 }, { "epoch": 2.86, "learning_rate": 1.4883864914035611e-05, "loss": 0.125, "step": 3120 }, { "epoch": 2.86, "learning_rate": 1.4880626401697487e-05, "loss": 0.1238, "step": 3121 }, { "epoch": 2.86, "learning_rate": 1.4877387217282757e-05, "loss": 0.1059, "step": 3122 }, { "epoch": 2.86, "learning_rate": 1.4874147361237472e-05, "loss": 0.1168, "step": 3123 }, { "epoch": 2.86, "learning_rate": 1.487090683400776e-05, "loss": 0.109, "step": 3124 }, { "epoch": 2.86, "learning_rate": 1.4867665636039863e-05, "loss": 0.12, "step": 3125 }, { "epoch": 2.87, "learning_rate": 1.4864423767780097e-05, "loss": 0.1224, "step": 3126 }, { "epoch": 2.87, "learning_rate": 1.4861181229674877e-05, "loss": 0.1218, "step": 3127 }, { "epoch": 2.87, "learning_rate": 1.4857938022170707e-05, "loss": 0.1032, "step": 3128 }, { "epoch": 2.87, "learning_rate": 1.4854694145714194e-05, "loss": 0.1229, "step": 3129 }, { "epoch": 2.87, "learning_rate": 1.4851449600752027e-05, "loss": 0.1251, "step": 3130 }, { "epoch": 2.87, "learning_rate": 1.4848204387730982e-05, "loss": 0.1236, "step": 3131 }, { "epoch": 2.87, "learning_rate": 1.4844958507097944e-05, "loss": 0.1204, "step": 3132 }, { "epoch": 2.87, "learning_rate": 1.4841711959299874e-05, "loss": 0.1096, "step": 3133 }, { "epoch": 2.87, "learning_rate": 1.4838464744783836e-05, "loss": 0.1177, "step": 3134 }, { "epoch": 2.87, "learning_rate": 1.4835216863996974e-05, "loss": 0.1006, "step": 3135 }, { "epoch": 2.87, "learning_rate": 1.4831968317386537e-05, "loss": 0.1266, "step": 3136 }, { "epoch": 2.88, "learning_rate": 1.482871910539986e-05, "loss": 0.1167, "step": 3137 }, { "epoch": 2.88, "learning_rate": 1.4825469228484365e-05, "loss": 0.1101, "step": 3138 }, { "epoch": 2.88, "learning_rate": 1.4822218687087571e-05, "loss": 0.1185, "step": 3139 }, { "epoch": 2.88, "learning_rate": 1.4818967481657088e-05, "loss": 0.1165, "step": 3140 }, { "epoch": 2.88, "learning_rate": 1.4815715612640613e-05, "loss": 0.1229, "step": 3141 }, { "epoch": 2.88, "learning_rate": 1.4812463080485944e-05, "loss": 0.1159, "step": 3142 }, { "epoch": 2.88, "learning_rate": 1.480920988564096e-05, "loss": 0.1113, "step": 3143 }, { "epoch": 2.88, "learning_rate": 1.4805956028553636e-05, "loss": 0.1207, "step": 3144 }, { "epoch": 2.88, "learning_rate": 1.4802701509672038e-05, "loss": 0.124, "step": 3145 }, { "epoch": 2.88, "learning_rate": 1.4799446329444325e-05, "loss": 0.1126, "step": 3146 }, { "epoch": 2.88, "learning_rate": 1.4796190488318743e-05, "loss": 0.1224, "step": 3147 }, { "epoch": 2.89, "learning_rate": 1.479293398674363e-05, "loss": 0.114, "step": 3148 }, { "epoch": 2.89, "learning_rate": 1.4789676825167415e-05, "loss": 0.131, "step": 3149 }, { "epoch": 2.89, "learning_rate": 1.478641900403862e-05, "loss": 0.1157, "step": 3150 }, { "epoch": 2.89, "learning_rate": 1.4783160523805859e-05, "loss": 0.1178, "step": 3151 }, { "epoch": 2.89, "learning_rate": 1.4779901384917833e-05, "loss": 0.1124, "step": 3152 }, { "epoch": 2.89, "learning_rate": 1.477664158782333e-05, "loss": 0.1149, "step": 3153 }, { "epoch": 2.89, "learning_rate": 1.4773381132971243e-05, "loss": 0.1177, "step": 3154 }, { "epoch": 2.89, "learning_rate": 1.4770120020810538e-05, "loss": 0.1087, "step": 3155 }, { "epoch": 2.89, "learning_rate": 1.4766858251790284e-05, "loss": 0.1289, "step": 3156 }, { "epoch": 2.89, "learning_rate": 1.4763595826359636e-05, "loss": 0.113, "step": 3157 }, { "epoch": 2.89, "learning_rate": 1.4760332744967836e-05, "loss": 0.1133, "step": 3158 }, { "epoch": 2.9, "learning_rate": 1.4757069008064223e-05, "loss": 0.112, "step": 3159 }, { "epoch": 2.9, "learning_rate": 1.4753804616098224e-05, "loss": 0.1264, "step": 3160 }, { "epoch": 2.9, "learning_rate": 1.475053956951935e-05, "loss": 0.1016, "step": 3161 }, { "epoch": 2.9, "learning_rate": 1.4747273868777216e-05, "loss": 0.1107, "step": 3162 }, { "epoch": 2.9, "learning_rate": 1.4744007514321511e-05, "loss": 0.1211, "step": 3163 }, { "epoch": 2.9, "learning_rate": 1.4740740506602027e-05, "loss": 0.1324, "step": 3164 }, { "epoch": 2.9, "learning_rate": 1.4737472846068638e-05, "loss": 0.1097, "step": 3165 }, { "epoch": 2.9, "learning_rate": 1.4734204533171312e-05, "loss": 0.1189, "step": 3166 }, { "epoch": 2.9, "learning_rate": 1.4730935568360103e-05, "loss": 0.1128, "step": 3167 }, { "epoch": 2.9, "learning_rate": 1.4727665952085155e-05, "loss": 0.1083, "step": 3168 }, { "epoch": 2.9, "learning_rate": 1.472439568479671e-05, "loss": 0.1162, "step": 3169 }, { "epoch": 2.91, "learning_rate": 1.472112476694509e-05, "loss": 0.1167, "step": 3170 }, { "epoch": 2.91, "learning_rate": 1.471785319898071e-05, "loss": 0.1157, "step": 3171 }, { "epoch": 2.91, "learning_rate": 1.4714580981354077e-05, "loss": 0.124, "step": 3172 }, { "epoch": 2.91, "learning_rate": 1.4711308114515781e-05, "loss": 0.1193, "step": 3173 }, { "epoch": 2.91, "learning_rate": 1.4708034598916507e-05, "loss": 0.1099, "step": 3174 }, { "epoch": 2.91, "learning_rate": 1.4704760435007029e-05, "loss": 0.1223, "step": 3175 }, { "epoch": 2.91, "learning_rate": 1.4701485623238205e-05, "loss": 0.1211, "step": 3176 }, { "epoch": 2.91, "learning_rate": 1.4698210164060992e-05, "loss": 0.1086, "step": 3177 }, { "epoch": 2.91, "learning_rate": 1.4694934057926427e-05, "loss": 0.1178, "step": 3178 }, { "epoch": 2.91, "learning_rate": 1.4691657305285639e-05, "loss": 0.1273, "step": 3179 }, { "epoch": 2.91, "learning_rate": 1.4688379906589849e-05, "loss": 0.1232, "step": 3180 }, { "epoch": 2.92, "learning_rate": 1.468510186229036e-05, "loss": 0.1216, "step": 3181 }, { "epoch": 2.92, "learning_rate": 1.4681823172838572e-05, "loss": 0.1148, "step": 3182 }, { "epoch": 2.92, "learning_rate": 1.467854383868597e-05, "loss": 0.1168, "step": 3183 }, { "epoch": 2.92, "learning_rate": 1.4675263860284128e-05, "loss": 0.1014, "step": 3184 }, { "epoch": 2.92, "learning_rate": 1.4671983238084707e-05, "loss": 0.1199, "step": 3185 }, { "epoch": 2.92, "learning_rate": 1.4668701972539459e-05, "loss": 0.1108, "step": 3186 }, { "epoch": 2.92, "learning_rate": 1.4665420064100226e-05, "loss": 0.1112, "step": 3187 }, { "epoch": 2.92, "learning_rate": 1.4662137513218932e-05, "loss": 0.1101, "step": 3188 }, { "epoch": 2.92, "learning_rate": 1.46588543203476e-05, "loss": 0.1119, "step": 3189 }, { "epoch": 2.92, "learning_rate": 1.4655570485938332e-05, "loss": 0.1157, "step": 3190 }, { "epoch": 2.92, "learning_rate": 1.465228601044332e-05, "loss": 0.1113, "step": 3191 }, { "epoch": 2.93, "learning_rate": 1.4649000894314845e-05, "loss": 0.126, "step": 3192 }, { "epoch": 2.93, "learning_rate": 1.4645715138005281e-05, "loss": 0.1166, "step": 3193 }, { "epoch": 2.93, "learning_rate": 1.4642428741967084e-05, "loss": 0.1178, "step": 3194 }, { "epoch": 2.93, "learning_rate": 1.4639141706652804e-05, "loss": 0.1111, "step": 3195 }, { "epoch": 2.93, "learning_rate": 1.4635854032515073e-05, "loss": 0.1096, "step": 3196 }, { "epoch": 2.93, "learning_rate": 1.463256572000661e-05, "loss": 0.1103, "step": 3197 }, { "epoch": 2.93, "learning_rate": 1.4629276769580233e-05, "loss": 0.113, "step": 3198 }, { "epoch": 2.93, "learning_rate": 1.462598718168883e-05, "loss": 0.1208, "step": 3199 }, { "epoch": 2.93, "learning_rate": 1.4622696956785399e-05, "loss": 0.1211, "step": 3200 }, { "epoch": 2.93, "learning_rate": 1.4619406095323002e-05, "loss": 0.1241, "step": 3201 }, { "epoch": 2.93, "learning_rate": 1.4616114597754805e-05, "loss": 0.1163, "step": 3202 }, { "epoch": 2.94, "learning_rate": 1.461282246453406e-05, "loss": 0.1162, "step": 3203 }, { "epoch": 2.94, "learning_rate": 1.4609529696114098e-05, "loss": 0.1203, "step": 3204 }, { "epoch": 2.94, "learning_rate": 1.4606236292948347e-05, "loss": 0.1208, "step": 3205 }, { "epoch": 2.94, "learning_rate": 1.4602942255490318e-05, "loss": 0.1249, "step": 3206 }, { "epoch": 2.94, "learning_rate": 1.4599647584193608e-05, "loss": 0.1177, "step": 3207 }, { "epoch": 2.94, "learning_rate": 1.4596352279511903e-05, "loss": 0.1157, "step": 3208 }, { "epoch": 2.94, "learning_rate": 1.4593056341898975e-05, "loss": 0.12, "step": 3209 }, { "epoch": 2.94, "learning_rate": 1.458975977180869e-05, "loss": 0.1121, "step": 3210 }, { "epoch": 2.94, "learning_rate": 1.4586462569694988e-05, "loss": 0.1261, "step": 3211 }, { "epoch": 2.94, "learning_rate": 1.4583164736011908e-05, "loss": 0.1255, "step": 3212 }, { "epoch": 2.95, "learning_rate": 1.457986627121357e-05, "loss": 0.1139, "step": 3213 }, { "epoch": 2.95, "learning_rate": 1.4576567175754183e-05, "loss": 0.1134, "step": 3214 }, { "epoch": 2.95, "learning_rate": 1.4573267450088042e-05, "loss": 0.1197, "step": 3215 }, { "epoch": 2.95, "learning_rate": 1.4569967094669528e-05, "loss": 0.1142, "step": 3216 }, { "epoch": 2.95, "learning_rate": 1.4566666109953109e-05, "loss": 0.1349, "step": 3217 }, { "epoch": 2.95, "learning_rate": 1.4563364496393342e-05, "loss": 0.1199, "step": 3218 }, { "epoch": 2.95, "learning_rate": 1.456006225444487e-05, "loss": 0.1135, "step": 3219 }, { "epoch": 2.95, "learning_rate": 1.4556759384562418e-05, "loss": 0.1224, "step": 3220 }, { "epoch": 2.95, "learning_rate": 1.4553455887200802e-05, "loss": 0.1083, "step": 3221 }, { "epoch": 2.95, "learning_rate": 1.4550151762814926e-05, "loss": 0.1126, "step": 3222 }, { "epoch": 2.95, "learning_rate": 1.4546847011859775e-05, "loss": 0.1077, "step": 3223 }, { "epoch": 2.96, "learning_rate": 1.4543541634790424e-05, "loss": 0.1026, "step": 3224 }, { "epoch": 2.96, "learning_rate": 1.4540235632062029e-05, "loss": 0.1087, "step": 3225 }, { "epoch": 2.96, "learning_rate": 1.4536929004129845e-05, "loss": 0.1224, "step": 3226 }, { "epoch": 2.96, "learning_rate": 1.4533621751449196e-05, "loss": 0.1211, "step": 3227 }, { "epoch": 2.96, "learning_rate": 1.4530313874475503e-05, "loss": 0.1069, "step": 3228 }, { "epoch": 2.96, "learning_rate": 1.4527005373664274e-05, "loss": 0.1182, "step": 3229 }, { "epoch": 2.96, "learning_rate": 1.4523696249471095e-05, "loss": 0.121, "step": 3230 }, { "epoch": 2.96, "learning_rate": 1.4520386502351645e-05, "loss": 0.1129, "step": 3231 }, { "epoch": 2.96, "learning_rate": 1.4517076132761686e-05, "loss": 0.1132, "step": 3232 }, { "epoch": 2.96, "learning_rate": 1.4513765141157059e-05, "loss": 0.1161, "step": 3233 }, { "epoch": 2.96, "learning_rate": 1.4510453527993704e-05, "loss": 0.1009, "step": 3234 }, { "epoch": 2.97, "learning_rate": 1.450714129372764e-05, "loss": 0.1191, "step": 3235 }, { "epoch": 2.97, "learning_rate": 1.450382843881497e-05, "loss": 0.1066, "step": 3236 }, { "epoch": 2.97, "learning_rate": 1.4500514963711883e-05, "loss": 0.1145, "step": 3237 }, { "epoch": 2.97, "learning_rate": 1.4497200868874653e-05, "loss": 0.1233, "step": 3238 }, { "epoch": 2.97, "learning_rate": 1.4493886154759644e-05, "loss": 0.1237, "step": 3239 }, { "epoch": 2.97, "learning_rate": 1.4490570821823298e-05, "loss": 0.1138, "step": 3240 }, { "epoch": 2.97, "learning_rate": 1.448725487052215e-05, "loss": 0.1189, "step": 3241 }, { "epoch": 2.97, "learning_rate": 1.4483938301312817e-05, "loss": 0.1235, "step": 3242 }, { "epoch": 2.97, "learning_rate": 1.4480621114651993e-05, "loss": 0.1134, "step": 3243 }, { "epoch": 2.97, "learning_rate": 1.4477303310996474e-05, "loss": 0.122, "step": 3244 }, { "epoch": 2.97, "learning_rate": 1.4473984890803123e-05, "loss": 0.1075, "step": 3245 }, { "epoch": 2.98, "learning_rate": 1.4470665854528902e-05, "loss": 0.1178, "step": 3246 }, { "epoch": 2.98, "learning_rate": 1.4467346202630847e-05, "loss": 0.1112, "step": 3247 }, { "epoch": 2.98, "learning_rate": 1.446402593556609e-05, "loss": 0.1218, "step": 3248 }, { "epoch": 2.98, "learning_rate": 1.4460705053791838e-05, "loss": 0.1152, "step": 3249 }, { "epoch": 2.98, "learning_rate": 1.4457383557765385e-05, "loss": 0.1301, "step": 3250 }, { "epoch": 2.98, "learning_rate": 1.445406144794411e-05, "loss": 0.1195, "step": 3251 }, { "epoch": 2.98, "learning_rate": 1.4450738724785482e-05, "loss": 0.1107, "step": 3252 }, { "epoch": 2.98, "learning_rate": 1.4447415388747048e-05, "loss": 0.1088, "step": 3253 }, { "epoch": 2.98, "learning_rate": 1.444409144028644e-05, "loss": 0.1108, "step": 3254 }, { "epoch": 2.98, "learning_rate": 1.4440766879861377e-05, "loss": 0.1039, "step": 3255 }, { "epoch": 2.98, "learning_rate": 1.4437441707929657e-05, "loss": 0.102, "step": 3256 }, { "epoch": 2.99, "learning_rate": 1.4434115924949168e-05, "loss": 0.12, "step": 3257 }, { "epoch": 2.99, "learning_rate": 1.4430789531377882e-05, "loss": 0.1151, "step": 3258 }, { "epoch": 2.99, "learning_rate": 1.4427462527673852e-05, "loss": 0.1128, "step": 3259 }, { "epoch": 2.99, "learning_rate": 1.4424134914295216e-05, "loss": 0.1094, "step": 3260 }, { "epoch": 2.99, "learning_rate": 1.4420806691700195e-05, "loss": 0.1173, "step": 3261 }, { "epoch": 2.99, "learning_rate": 1.4417477860347098e-05, "loss": 0.1083, "step": 3262 }, { "epoch": 2.99, "learning_rate": 1.4414148420694312e-05, "loss": 0.1132, "step": 3263 }, { "epoch": 2.99, "learning_rate": 1.4410818373200313e-05, "loss": 0.1077, "step": 3264 }, { "epoch": 2.99, "learning_rate": 1.4407487718323653e-05, "loss": 0.1232, "step": 3265 }, { "epoch": 2.99, "learning_rate": 1.4404156456522979e-05, "loss": 0.1124, "step": 3266 }, { "epoch": 2.99, "learning_rate": 1.4400824588257016e-05, "loss": 0.1028, "step": 3267 }, { "epoch": 3.0, "learning_rate": 1.4397492113984569e-05, "loss": 0.1157, "step": 3268 }, { "epoch": 3.0, "learning_rate": 1.4394159034164528e-05, "loss": 0.1199, "step": 3269 }, { "epoch": 3.0, "learning_rate": 1.439082534925587e-05, "loss": 0.1126, "step": 3270 }, { "epoch": 3.0, "learning_rate": 1.4387491059717653e-05, "loss": 0.113, "step": 3271 }, { "epoch": 3.0, "learning_rate": 1.4384156166009018e-05, "loss": 0.1237, "step": 3272 }, { "epoch": 3.0, "learning_rate": 1.4380820668589193e-05, "loss": 0.1042, "step": 3273 }, { "epoch": 3.0, "learning_rate": 1.4377484567917482e-05, "loss": 0.1018, "step": 3274 }, { "epoch": 3.0, "learning_rate": 1.4374147864453275e-05, "loss": 0.116, "step": 3275 }, { "epoch": 3.0, "learning_rate": 1.4370810558656049e-05, "loss": 0.1099, "step": 3276 }, { "epoch": 3.0, "learning_rate": 1.4367472650985362e-05, "loss": 0.1025, "step": 3277 }, { "epoch": 3.0, "learning_rate": 1.436413414190085e-05, "loss": 0.1095, "step": 3278 }, { "epoch": 3.01, "learning_rate": 1.4360795031862233e-05, "loss": 0.0995, "step": 3279 }, { "epoch": 3.01, "learning_rate": 1.4357455321329327e-05, "loss": 0.0998, "step": 3280 }, { "epoch": 3.01, "learning_rate": 1.4354115010762009e-05, "loss": 0.1187, "step": 3281 }, { "epoch": 3.01, "learning_rate": 1.4350774100620252e-05, "loss": 0.1066, "step": 3282 }, { "epoch": 3.01, "learning_rate": 1.4347432591364114e-05, "loss": 0.0976, "step": 3283 }, { "epoch": 3.01, "learning_rate": 1.4344090483453728e-05, "loss": 0.1002, "step": 3284 }, { "epoch": 3.01, "learning_rate": 1.434074777734931e-05, "loss": 0.1006, "step": 3285 }, { "epoch": 3.01, "learning_rate": 1.433740447351116e-05, "loss": 0.0952, "step": 3286 }, { "epoch": 3.01, "learning_rate": 1.4334060572399668e-05, "loss": 0.1022, "step": 3287 }, { "epoch": 3.01, "learning_rate": 1.4330716074475287e-05, "loss": 0.0952, "step": 3288 }, { "epoch": 3.01, "learning_rate": 1.4327370980198579e-05, "loss": 0.1035, "step": 3289 }, { "epoch": 3.02, "learning_rate": 1.4324025290030158e-05, "loss": 0.0958, "step": 3290 }, { "epoch": 3.02, "learning_rate": 1.4320679004430747e-05, "loss": 0.0995, "step": 3291 }, { "epoch": 3.02, "learning_rate": 1.4317332123861133e-05, "loss": 0.1003, "step": 3292 }, { "epoch": 3.02, "learning_rate": 1.4313984648782192e-05, "loss": 0.0999, "step": 3293 }, { "epoch": 3.02, "learning_rate": 1.4310636579654887e-05, "loss": 0.1039, "step": 3294 }, { "epoch": 3.02, "learning_rate": 1.4307287916940249e-05, "loss": 0.1026, "step": 3295 }, { "epoch": 3.02, "learning_rate": 1.4303938661099404e-05, "loss": 0.104, "step": 3296 }, { "epoch": 3.02, "learning_rate": 1.430058881259355e-05, "loss": 0.0983, "step": 3297 }, { "epoch": 3.02, "learning_rate": 1.4297238371883974e-05, "loss": 0.1031, "step": 3298 }, { "epoch": 3.02, "learning_rate": 1.4293887339432045e-05, "loss": 0.1102, "step": 3299 }, { "epoch": 3.02, "learning_rate": 1.42905357156992e-05, "loss": 0.101, "step": 3300 }, { "epoch": 3.03, "learning_rate": 1.428718350114698e-05, "loss": 0.1022, "step": 3301 }, { "epoch": 3.03, "learning_rate": 1.4283830696236984e-05, "loss": 0.1083, "step": 3302 }, { "epoch": 3.03, "learning_rate": 1.4280477301430908e-05, "loss": 0.109, "step": 3303 }, { "epoch": 3.03, "learning_rate": 1.4277123317190524e-05, "loss": 0.0957, "step": 3304 }, { "epoch": 3.03, "learning_rate": 1.4273768743977687e-05, "loss": 0.1081, "step": 3305 }, { "epoch": 3.03, "learning_rate": 1.4270413582254328e-05, "loss": 0.0971, "step": 3306 }, { "epoch": 3.03, "learning_rate": 1.4267057832482465e-05, "loss": 0.1011, "step": 3307 }, { "epoch": 3.03, "learning_rate": 1.4263701495124193e-05, "loss": 0.1091, "step": 3308 }, { "epoch": 3.03, "learning_rate": 1.426034457064169e-05, "loss": 0.1007, "step": 3309 }, { "epoch": 3.03, "learning_rate": 1.425698705949722e-05, "loss": 0.1012, "step": 3310 }, { "epoch": 3.03, "learning_rate": 1.4253628962153112e-05, "loss": 0.1034, "step": 3311 }, { "epoch": 3.04, "learning_rate": 1.4250270279071794e-05, "loss": 0.1029, "step": 3312 }, { "epoch": 3.04, "learning_rate": 1.424691101071576e-05, "loss": 0.1015, "step": 3313 }, { "epoch": 3.04, "learning_rate": 1.4243551157547598e-05, "loss": 0.0983, "step": 3314 }, { "epoch": 3.04, "learning_rate": 1.4240190720029964e-05, "loss": 0.1009, "step": 3315 }, { "epoch": 3.04, "learning_rate": 1.4236829698625602e-05, "loss": 0.115, "step": 3316 }, { "epoch": 3.04, "learning_rate": 1.4233468093797336e-05, "loss": 0.1091, "step": 3317 }, { "epoch": 3.04, "learning_rate": 1.4230105906008067e-05, "loss": 0.1031, "step": 3318 }, { "epoch": 3.04, "learning_rate": 1.4226743135720782e-05, "loss": 0.0996, "step": 3319 }, { "epoch": 3.04, "learning_rate": 1.422337978339854e-05, "loss": 0.1015, "step": 3320 }, { "epoch": 3.04, "learning_rate": 1.4220015849504485e-05, "loss": 0.1108, "step": 3321 }, { "epoch": 3.04, "learning_rate": 1.421665133450184e-05, "loss": 0.0966, "step": 3322 }, { "epoch": 3.05, "learning_rate": 1.4213286238853911e-05, "loss": 0.0992, "step": 3323 }, { "epoch": 3.05, "learning_rate": 1.4209920563024081e-05, "loss": 0.1136, "step": 3324 }, { "epoch": 3.05, "learning_rate": 1.4206554307475816e-05, "loss": 0.0932, "step": 3325 }, { "epoch": 3.05, "learning_rate": 1.4203187472672654e-05, "loss": 0.1009, "step": 3326 }, { "epoch": 3.05, "learning_rate": 1.419982005907822e-05, "loss": 0.0929, "step": 3327 }, { "epoch": 3.05, "learning_rate": 1.4196452067156217e-05, "loss": 0.1023, "step": 3328 }, { "epoch": 3.05, "learning_rate": 1.4193083497370426e-05, "loss": 0.0952, "step": 3329 }, { "epoch": 3.05, "learning_rate": 1.418971435018471e-05, "loss": 0.1027, "step": 3330 }, { "epoch": 3.05, "learning_rate": 1.4186344626063017e-05, "loss": 0.1056, "step": 3331 }, { "epoch": 3.05, "learning_rate": 1.4182974325469356e-05, "loss": 0.1009, "step": 3332 }, { "epoch": 3.05, "learning_rate": 1.4179603448867836e-05, "loss": 0.097, "step": 3333 }, { "epoch": 3.06, "learning_rate": 1.4176231996722634e-05, "loss": 0.1105, "step": 3334 }, { "epoch": 3.06, "learning_rate": 1.4172859969498008e-05, "loss": 0.1097, "step": 3335 }, { "epoch": 3.06, "learning_rate": 1.4169487367658295e-05, "loss": 0.0967, "step": 3336 }, { "epoch": 3.06, "learning_rate": 1.4166114191667914e-05, "loss": 0.0994, "step": 3337 }, { "epoch": 3.06, "learning_rate": 1.416274044199136e-05, "loss": 0.1015, "step": 3338 }, { "epoch": 3.06, "learning_rate": 1.4159366119093214e-05, "loss": 0.1022, "step": 3339 }, { "epoch": 3.06, "learning_rate": 1.4155991223438123e-05, "loss": 0.1069, "step": 3340 }, { "epoch": 3.06, "learning_rate": 1.4152615755490822e-05, "loss": 0.0928, "step": 3341 }, { "epoch": 3.06, "learning_rate": 1.4149239715716125e-05, "loss": 0.106, "step": 3342 }, { "epoch": 3.06, "learning_rate": 1.4145863104578921e-05, "loss": 0.0979, "step": 3343 }, { "epoch": 3.07, "learning_rate": 1.414248592254418e-05, "loss": 0.0908, "step": 3344 }, { "epoch": 3.07, "learning_rate": 1.4139108170076947e-05, "loss": 0.1192, "step": 3345 }, { "epoch": 3.07, "learning_rate": 1.413572984764235e-05, "loss": 0.1002, "step": 3346 }, { "epoch": 3.07, "learning_rate": 1.4132350955705598e-05, "loss": 0.1032, "step": 3347 }, { "epoch": 3.07, "learning_rate": 1.4128971494731968e-05, "loss": 0.098, "step": 3348 }, { "epoch": 3.07, "learning_rate": 1.412559146518683e-05, "loss": 0.1073, "step": 3349 }, { "epoch": 3.07, "learning_rate": 1.4122210867535613e-05, "loss": 0.1103, "step": 3350 }, { "epoch": 3.07, "learning_rate": 1.4118829702243844e-05, "loss": 0.1016, "step": 3351 }, { "epoch": 3.07, "learning_rate": 1.4115447969777114e-05, "loss": 0.1077, "step": 3352 }, { "epoch": 3.07, "learning_rate": 1.4112065670601103e-05, "loss": 0.0953, "step": 3353 }, { "epoch": 3.07, "learning_rate": 1.4108682805181558e-05, "loss": 0.102, "step": 3354 }, { "epoch": 3.08, "learning_rate": 1.4105299373984312e-05, "loss": 0.1035, "step": 3355 }, { "epoch": 3.08, "learning_rate": 1.4101915377475275e-05, "loss": 0.1059, "step": 3356 }, { "epoch": 3.08, "learning_rate": 1.4098530816120428e-05, "loss": 0.1018, "step": 3357 }, { "epoch": 3.08, "learning_rate": 1.4095145690385843e-05, "loss": 0.0995, "step": 3358 }, { "epoch": 3.08, "learning_rate": 1.4091760000737656e-05, "loss": 0.1, "step": 3359 }, { "epoch": 3.08, "learning_rate": 1.4088373747642086e-05, "loss": 0.0912, "step": 3360 }, { "epoch": 3.08, "learning_rate": 1.4084986931565433e-05, "loss": 0.1018, "step": 3361 }, { "epoch": 3.08, "learning_rate": 1.4081599552974068e-05, "loss": 0.0928, "step": 3362 }, { "epoch": 3.08, "learning_rate": 1.407821161233445e-05, "loss": 0.0962, "step": 3363 }, { "epoch": 3.08, "learning_rate": 1.40748231101131e-05, "loss": 0.1001, "step": 3364 }, { "epoch": 3.08, "learning_rate": 1.407143404677663e-05, "loss": 0.1077, "step": 3365 }, { "epoch": 3.09, "learning_rate": 1.4068044422791722e-05, "loss": 0.1007, "step": 3366 }, { "epoch": 3.09, "learning_rate": 1.406465423862514e-05, "loss": 0.1012, "step": 3367 }, { "epoch": 3.09, "learning_rate": 1.4061263494743717e-05, "loss": 0.0959, "step": 3368 }, { "epoch": 3.09, "learning_rate": 1.4057872191614375e-05, "loss": 0.0954, "step": 3369 }, { "epoch": 3.09, "learning_rate": 1.40544803297041e-05, "loss": 0.1063, "step": 3370 }, { "epoch": 3.09, "learning_rate": 1.4051087909479963e-05, "loss": 0.0977, "step": 3371 }, { "epoch": 3.09, "learning_rate": 1.4047694931409112e-05, "loss": 0.1039, "step": 3372 }, { "epoch": 3.09, "learning_rate": 1.404430139595877e-05, "loss": 0.0992, "step": 3373 }, { "epoch": 3.09, "learning_rate": 1.4040907303596238e-05, "loss": 0.1075, "step": 3374 }, { "epoch": 3.09, "learning_rate": 1.403751265478889e-05, "loss": 0.1093, "step": 3375 }, { "epoch": 3.09, "learning_rate": 1.4034117450004178e-05, "loss": 0.0862, "step": 3376 }, { "epoch": 3.1, "learning_rate": 1.4030721689709634e-05, "loss": 0.1029, "step": 3377 }, { "epoch": 3.1, "learning_rate": 1.4027325374372866e-05, "loss": 0.0963, "step": 3378 }, { "epoch": 3.1, "learning_rate": 1.4023928504461552e-05, "loss": 0.0965, "step": 3379 }, { "epoch": 3.1, "learning_rate": 1.4020531080443453e-05, "loss": 0.0971, "step": 3380 }, { "epoch": 3.1, "learning_rate": 1.4017133102786405e-05, "loss": 0.0873, "step": 3381 }, { "epoch": 3.1, "learning_rate": 1.4013734571958319e-05, "loss": 0.0946, "step": 3382 }, { "epoch": 3.1, "learning_rate": 1.4010335488427184e-05, "loss": 0.0958, "step": 3383 }, { "epoch": 3.1, "learning_rate": 1.4006935852661061e-05, "loss": 0.1061, "step": 3384 }, { "epoch": 3.1, "learning_rate": 1.4003535665128092e-05, "loss": 0.091, "step": 3385 }, { "epoch": 3.1, "learning_rate": 1.4000134926296491e-05, "loss": 0.108, "step": 3386 }, { "epoch": 3.1, "learning_rate": 1.3996733636634554e-05, "loss": 0.1056, "step": 3387 }, { "epoch": 3.11, "learning_rate": 1.3993331796610643e-05, "loss": 0.0978, "step": 3388 }, { "epoch": 3.11, "learning_rate": 1.3989929406693206e-05, "loss": 0.1067, "step": 3389 }, { "epoch": 3.11, "learning_rate": 1.398652646735076e-05, "loss": 0.1116, "step": 3390 }, { "epoch": 3.11, "learning_rate": 1.3983122979051897e-05, "loss": 0.0902, "step": 3391 }, { "epoch": 3.11, "learning_rate": 1.3979718942265292e-05, "loss": 0.0997, "step": 3392 }, { "epoch": 3.11, "learning_rate": 1.3976314357459688e-05, "loss": 0.1143, "step": 3393 }, { "epoch": 3.11, "learning_rate": 1.397290922510391e-05, "loss": 0.1093, "step": 3394 }, { "epoch": 3.11, "learning_rate": 1.396950354566685e-05, "loss": 0.1042, "step": 3395 }, { "epoch": 3.11, "learning_rate": 1.396609731961748e-05, "loss": 0.1003, "step": 3396 }, { "epoch": 3.11, "learning_rate": 1.3962690547424853e-05, "loss": 0.1128, "step": 3397 }, { "epoch": 3.11, "learning_rate": 1.3959283229558084e-05, "loss": 0.1005, "step": 3398 }, { "epoch": 3.12, "learning_rate": 1.395587536648638e-05, "loss": 0.0998, "step": 3399 }, { "epoch": 3.12, "learning_rate": 1.3952466958679006e-05, "loss": 0.1067, "step": 3400 }, { "epoch": 3.12, "learning_rate": 1.3949058006605312e-05, "loss": 0.0959, "step": 3401 }, { "epoch": 3.12, "learning_rate": 1.394564851073472e-05, "loss": 0.1019, "step": 3402 }, { "epoch": 3.12, "learning_rate": 1.3942238471536729e-05, "loss": 0.0983, "step": 3403 }, { "epoch": 3.12, "learning_rate": 1.393882788948091e-05, "loss": 0.104, "step": 3404 }, { "epoch": 3.12, "learning_rate": 1.3935416765036913e-05, "loss": 0.0991, "step": 3405 }, { "epoch": 3.12, "learning_rate": 1.3932005098674458e-05, "loss": 0.1143, "step": 3406 }, { "epoch": 3.12, "learning_rate": 1.392859289086334e-05, "loss": 0.0997, "step": 3407 }, { "epoch": 3.12, "learning_rate": 1.3925180142073433e-05, "loss": 0.1068, "step": 3408 }, { "epoch": 3.12, "learning_rate": 1.3921766852774678e-05, "loss": 0.0988, "step": 3409 }, { "epoch": 3.13, "learning_rate": 1.3918353023437104e-05, "loss": 0.0991, "step": 3410 }, { "epoch": 3.13, "learning_rate": 1.3914938654530793e-05, "loss": 0.1057, "step": 3411 }, { "epoch": 3.13, "learning_rate": 1.3911523746525924e-05, "loss": 0.1141, "step": 3412 }, { "epoch": 3.13, "learning_rate": 1.3908108299892735e-05, "loss": 0.0966, "step": 3413 }, { "epoch": 3.13, "learning_rate": 1.3904692315101544e-05, "loss": 0.1013, "step": 3414 }, { "epoch": 3.13, "learning_rate": 1.3901275792622744e-05, "loss": 0.0964, "step": 3415 }, { "epoch": 3.13, "learning_rate": 1.3897858732926794e-05, "loss": 0.0966, "step": 3416 }, { "epoch": 3.13, "learning_rate": 1.3894441136484242e-05, "loss": 0.0992, "step": 3417 }, { "epoch": 3.13, "learning_rate": 1.3891023003765694e-05, "loss": 0.1223, "step": 3418 }, { "epoch": 3.13, "learning_rate": 1.3887604335241837e-05, "loss": 0.0985, "step": 3419 }, { "epoch": 3.13, "learning_rate": 1.388418513138344e-05, "loss": 0.0984, "step": 3420 }, { "epoch": 3.14, "learning_rate": 1.3880765392661327e-05, "loss": 0.1025, "step": 3421 }, { "epoch": 3.14, "learning_rate": 1.3877345119546415e-05, "loss": 0.1002, "step": 3422 }, { "epoch": 3.14, "learning_rate": 1.3873924312509677e-05, "loss": 0.1085, "step": 3423 }, { "epoch": 3.14, "learning_rate": 1.3870502972022175e-05, "loss": 0.1082, "step": 3424 }, { "epoch": 3.14, "learning_rate": 1.3867081098555032e-05, "loss": 0.099, "step": 3425 }, { "epoch": 3.14, "learning_rate": 1.3863658692579457e-05, "loss": 0.0942, "step": 3426 }, { "epoch": 3.14, "learning_rate": 1.3860235754566718e-05, "loss": 0.0972, "step": 3427 }, { "epoch": 3.14, "learning_rate": 1.3856812284988167e-05, "loss": 0.1147, "step": 3428 }, { "epoch": 3.14, "learning_rate": 1.385338828431523e-05, "loss": 0.1073, "step": 3429 }, { "epoch": 3.14, "learning_rate": 1.3849963753019395e-05, "loss": 0.1042, "step": 3430 }, { "epoch": 3.14, "learning_rate": 1.3846538691572236e-05, "loss": 0.1143, "step": 3431 }, { "epoch": 3.15, "learning_rate": 1.3843113100445387e-05, "loss": 0.107, "step": 3432 }, { "epoch": 3.15, "learning_rate": 1.3839686980110569e-05, "loss": 0.1037, "step": 3433 }, { "epoch": 3.15, "learning_rate": 1.3836260331039563e-05, "loss": 0.1165, "step": 3434 }, { "epoch": 3.15, "learning_rate": 1.3832833153704233e-05, "loss": 0.1123, "step": 3435 }, { "epoch": 3.15, "learning_rate": 1.3829405448576511e-05, "loss": 0.1017, "step": 3436 }, { "epoch": 3.15, "learning_rate": 1.3825977216128398e-05, "loss": 0.0904, "step": 3437 }, { "epoch": 3.15, "learning_rate": 1.3822548456831978e-05, "loss": 0.0992, "step": 3438 }, { "epoch": 3.15, "learning_rate": 1.3819119171159399e-05, "loss": 0.108, "step": 3439 }, { "epoch": 3.15, "learning_rate": 1.381568935958288e-05, "loss": 0.1018, "step": 3440 }, { "epoch": 3.15, "learning_rate": 1.3812259022574717e-05, "loss": 0.1055, "step": 3441 }, { "epoch": 3.15, "learning_rate": 1.3808828160607283e-05, "loss": 0.1124, "step": 3442 }, { "epoch": 3.16, "learning_rate": 1.380539677415301e-05, "loss": 0.1083, "step": 3443 }, { "epoch": 3.16, "learning_rate": 1.3801964863684416e-05, "loss": 0.116, "step": 3444 }, { "epoch": 3.16, "learning_rate": 1.3798532429674086e-05, "loss": 0.107, "step": 3445 }, { "epoch": 3.16, "learning_rate": 1.3795099472594669e-05, "loss": 0.0985, "step": 3446 }, { "epoch": 3.16, "learning_rate": 1.3791665992918899e-05, "loss": 0.1037, "step": 3447 }, { "epoch": 3.16, "learning_rate": 1.3788231991119577e-05, "loss": 0.11, "step": 3448 }, { "epoch": 3.16, "learning_rate": 1.3784797467669568e-05, "loss": 0.0965, "step": 3449 }, { "epoch": 3.16, "learning_rate": 1.3781362423041824e-05, "loss": 0.1126, "step": 3450 }, { "epoch": 3.16, "learning_rate": 1.3777926857709359e-05, "loss": 0.1061, "step": 3451 }, { "epoch": 3.16, "learning_rate": 1.377449077214526e-05, "loss": 0.0914, "step": 3452 }, { "epoch": 3.16, "learning_rate": 1.3771054166822682e-05, "loss": 0.1108, "step": 3453 }, { "epoch": 3.17, "learning_rate": 1.3767617042214862e-05, "loss": 0.1131, "step": 3454 }, { "epoch": 3.17, "learning_rate": 1.3764179398795097e-05, "loss": 0.0959, "step": 3455 }, { "epoch": 3.17, "learning_rate": 1.3760741237036765e-05, "loss": 0.1018, "step": 3456 }, { "epoch": 3.17, "learning_rate": 1.3757302557413306e-05, "loss": 0.1059, "step": 3457 }, { "epoch": 3.17, "learning_rate": 1.3753863360398243e-05, "loss": 0.102, "step": 3458 }, { "epoch": 3.17, "learning_rate": 1.3750423646465153e-05, "loss": 0.1058, "step": 3459 }, { "epoch": 3.17, "learning_rate": 1.3746983416087708e-05, "loss": 0.1022, "step": 3460 }, { "epoch": 3.17, "learning_rate": 1.3743542669739629e-05, "loss": 0.1052, "step": 3461 }, { "epoch": 3.17, "learning_rate": 1.3740101407894716e-05, "loss": 0.0968, "step": 3462 }, { "epoch": 3.17, "learning_rate": 1.3736659631026847e-05, "loss": 0.1051, "step": 3463 }, { "epoch": 3.18, "learning_rate": 1.3733217339609962e-05, "loss": 0.0937, "step": 3464 }, { "epoch": 3.18, "learning_rate": 1.3729774534118074e-05, "loss": 0.0968, "step": 3465 }, { "epoch": 3.18, "learning_rate": 1.3726331215025266e-05, "loss": 0.1086, "step": 3466 }, { "epoch": 3.18, "learning_rate": 1.3722887382805697e-05, "loss": 0.0982, "step": 3467 }, { "epoch": 3.18, "learning_rate": 1.3719443037933588e-05, "loss": 0.0949, "step": 3468 }, { "epoch": 3.18, "learning_rate": 1.371599818088324e-05, "loss": 0.0992, "step": 3469 }, { "epoch": 3.18, "learning_rate": 1.3712552812129017e-05, "loss": 0.11, "step": 3470 }, { "epoch": 3.18, "learning_rate": 1.370910693214536e-05, "loss": 0.1104, "step": 3471 }, { "epoch": 3.18, "learning_rate": 1.370566054140677e-05, "loss": 0.1163, "step": 3472 }, { "epoch": 3.18, "learning_rate": 1.3702213640387833e-05, "loss": 0.1039, "step": 3473 }, { "epoch": 3.18, "learning_rate": 1.3698766229563193e-05, "loss": 0.0981, "step": 3474 }, { "epoch": 3.19, "learning_rate": 1.369531830940757e-05, "loss": 0.1024, "step": 3475 }, { "epoch": 3.19, "learning_rate": 1.3691869880395751e-05, "loss": 0.1123, "step": 3476 }, { "epoch": 3.19, "learning_rate": 1.3688420943002599e-05, "loss": 0.1055, "step": 3477 }, { "epoch": 3.19, "learning_rate": 1.3684971497703035e-05, "loss": 0.1053, "step": 3478 }, { "epoch": 3.19, "learning_rate": 1.3681521544972067e-05, "loss": 0.0898, "step": 3479 }, { "epoch": 3.19, "learning_rate": 1.3678071085284757e-05, "loss": 0.1054, "step": 3480 }, { "epoch": 3.19, "learning_rate": 1.3674620119116247e-05, "loss": 0.1088, "step": 3481 }, { "epoch": 3.19, "learning_rate": 1.3671168646941742e-05, "loss": 0.0852, "step": 3482 }, { "epoch": 3.19, "learning_rate": 1.3667716669236524e-05, "loss": 0.1067, "step": 3483 }, { "epoch": 3.19, "learning_rate": 1.3664264186475935e-05, "loss": 0.106, "step": 3484 }, { "epoch": 3.19, "learning_rate": 1.3660811199135398e-05, "loss": 0.1068, "step": 3485 }, { "epoch": 3.2, "learning_rate": 1.3657357707690398e-05, "loss": 0.1035, "step": 3486 }, { "epoch": 3.2, "learning_rate": 1.3653903712616486e-05, "loss": 0.0985, "step": 3487 }, { "epoch": 3.2, "learning_rate": 1.3650449214389292e-05, "loss": 0.1067, "step": 3488 }, { "epoch": 3.2, "learning_rate": 1.364699421348451e-05, "loss": 0.1004, "step": 3489 }, { "epoch": 3.2, "learning_rate": 1.3643538710377901e-05, "loss": 0.1021, "step": 3490 }, { "epoch": 3.2, "learning_rate": 1.3640082705545305e-05, "loss": 0.0939, "step": 3491 }, { "epoch": 3.2, "learning_rate": 1.3636626199462615e-05, "loss": 0.1093, "step": 3492 }, { "epoch": 3.2, "learning_rate": 1.3633169192605806e-05, "loss": 0.1012, "step": 3493 }, { "epoch": 3.2, "learning_rate": 1.3629711685450922e-05, "loss": 0.1033, "step": 3494 }, { "epoch": 3.2, "learning_rate": 1.3626253678474067e-05, "loss": 0.1071, "step": 3495 }, { "epoch": 3.2, "learning_rate": 1.3622795172151417e-05, "loss": 0.0958, "step": 3496 }, { "epoch": 3.21, "learning_rate": 1.3619336166959225e-05, "loss": 0.1043, "step": 3497 }, { "epoch": 3.21, "learning_rate": 1.36158766633738e-05, "loss": 0.1097, "step": 3498 }, { "epoch": 3.21, "learning_rate": 1.3612416661871532e-05, "loss": 0.0944, "step": 3499 }, { "epoch": 3.21, "learning_rate": 1.3608956162928865e-05, "loss": 0.1057, "step": 3500 }, { "epoch": 3.21, "learning_rate": 1.3605495167022327e-05, "loss": 0.1047, "step": 3501 }, { "epoch": 3.21, "learning_rate": 1.3602033674628508e-05, "loss": 0.1089, "step": 3502 }, { "epoch": 3.21, "learning_rate": 1.3598571686224059e-05, "loss": 0.1134, "step": 3503 }, { "epoch": 3.21, "learning_rate": 1.359510920228571e-05, "loss": 0.1133, "step": 3504 }, { "epoch": 3.21, "learning_rate": 1.3591646223290256e-05, "loss": 0.0999, "step": 3505 }, { "epoch": 3.21, "learning_rate": 1.3588182749714558e-05, "loss": 0.105, "step": 3506 }, { "epoch": 3.21, "learning_rate": 1.3584718782035543e-05, "loss": 0.1014, "step": 3507 }, { "epoch": 3.22, "learning_rate": 1.3581254320730217e-05, "loss": 0.1079, "step": 3508 }, { "epoch": 3.22, "learning_rate": 1.3577789366275644e-05, "loss": 0.1055, "step": 3509 }, { "epoch": 3.22, "learning_rate": 1.3574323919148956e-05, "loss": 0.1111, "step": 3510 }, { "epoch": 3.22, "learning_rate": 1.3570857979827357e-05, "loss": 0.1042, "step": 3511 }, { "epoch": 3.22, "learning_rate": 1.3567391548788113e-05, "loss": 0.1072, "step": 3512 }, { "epoch": 3.22, "learning_rate": 1.3563924626508566e-05, "loss": 0.0985, "step": 3513 }, { "epoch": 3.22, "learning_rate": 1.3560457213466122e-05, "loss": 0.1113, "step": 3514 }, { "epoch": 3.22, "learning_rate": 1.3556989310138253e-05, "loss": 0.1115, "step": 3515 }, { "epoch": 3.22, "learning_rate": 1.3553520917002498e-05, "loss": 0.0991, "step": 3516 }, { "epoch": 3.22, "learning_rate": 1.3550052034536467e-05, "loss": 0.0988, "step": 3517 }, { "epoch": 3.22, "learning_rate": 1.3546582663217835e-05, "loss": 0.1065, "step": 3518 }, { "epoch": 3.23, "learning_rate": 1.3543112803524343e-05, "loss": 0.0993, "step": 3519 }, { "epoch": 3.23, "learning_rate": 1.3539642455933803e-05, "loss": 0.1062, "step": 3520 }, { "epoch": 3.23, "learning_rate": 1.3536171620924088e-05, "loss": 0.1087, "step": 3521 }, { "epoch": 3.23, "learning_rate": 1.3532700298973147e-05, "loss": 0.1015, "step": 3522 }, { "epoch": 3.23, "learning_rate": 1.352922849055899e-05, "loss": 0.0999, "step": 3523 }, { "epoch": 3.23, "learning_rate": 1.3525756196159691e-05, "loss": 0.0961, "step": 3524 }, { "epoch": 3.23, "learning_rate": 1.3522283416253403e-05, "loss": 0.1019, "step": 3525 }, { "epoch": 3.23, "learning_rate": 1.351881015131833e-05, "loss": 0.112, "step": 3526 }, { "epoch": 3.23, "learning_rate": 1.351533640183276e-05, "loss": 0.0995, "step": 3527 }, { "epoch": 3.23, "learning_rate": 1.351186216827503e-05, "loss": 0.0969, "step": 3528 }, { "epoch": 3.23, "learning_rate": 1.3508387451123555e-05, "loss": 0.1072, "step": 3529 }, { "epoch": 3.24, "learning_rate": 1.3504912250856811e-05, "loss": 0.0982, "step": 3530 }, { "epoch": 3.24, "learning_rate": 1.3501436567953353e-05, "loss": 0.0945, "step": 3531 }, { "epoch": 3.24, "learning_rate": 1.349796040289178e-05, "loss": 0.1056, "step": 3532 }, { "epoch": 3.24, "learning_rate": 1.3494483756150775e-05, "loss": 0.1014, "step": 3533 }, { "epoch": 3.24, "learning_rate": 1.3491006628209088e-05, "loss": 0.0927, "step": 3534 }, { "epoch": 3.24, "learning_rate": 1.3487529019545523e-05, "loss": 0.0895, "step": 3535 }, { "epoch": 3.24, "learning_rate": 1.3484050930638958e-05, "loss": 0.1058, "step": 3536 }, { "epoch": 3.24, "learning_rate": 1.348057236196834e-05, "loss": 0.1113, "step": 3537 }, { "epoch": 3.24, "learning_rate": 1.3477093314012677e-05, "loss": 0.0976, "step": 3538 }, { "epoch": 3.24, "learning_rate": 1.3473613787251039e-05, "loss": 0.0995, "step": 3539 }, { "epoch": 3.24, "learning_rate": 1.3470133782162572e-05, "loss": 0.1021, "step": 3540 }, { "epoch": 3.25, "learning_rate": 1.3466653299226482e-05, "loss": 0.1028, "step": 3541 }, { "epoch": 3.25, "learning_rate": 1.3463172338922041e-05, "loss": 0.1054, "step": 3542 }, { "epoch": 3.25, "learning_rate": 1.3459690901728589e-05, "loss": 0.1148, "step": 3543 }, { "epoch": 3.25, "learning_rate": 1.3456208988125528e-05, "loss": 0.1042, "step": 3544 }, { "epoch": 3.25, "learning_rate": 1.345272659859233e-05, "loss": 0.0998, "step": 3545 }, { "epoch": 3.25, "learning_rate": 1.3449243733608526e-05, "loss": 0.0868, "step": 3546 }, { "epoch": 3.25, "learning_rate": 1.3445760393653722e-05, "loss": 0.1015, "step": 3547 }, { "epoch": 3.25, "learning_rate": 1.344227657920758e-05, "loss": 0.0943, "step": 3548 }, { "epoch": 3.25, "learning_rate": 1.3438792290749833e-05, "loss": 0.113, "step": 3549 }, { "epoch": 3.25, "learning_rate": 1.3435307528760283e-05, "loss": 0.0967, "step": 3550 }, { "epoch": 3.25, "learning_rate": 1.3431822293718782e-05, "loss": 0.1011, "step": 3551 }, { "epoch": 3.26, "learning_rate": 1.3428336586105264e-05, "loss": 0.1025, "step": 3552 }, { "epoch": 3.26, "learning_rate": 1.3424850406399719e-05, "loss": 0.0939, "step": 3553 }, { "epoch": 3.26, "learning_rate": 1.3421363755082207e-05, "loss": 0.1036, "step": 3554 }, { "epoch": 3.26, "learning_rate": 1.3417876632632845e-05, "loss": 0.0949, "step": 3555 }, { "epoch": 3.26, "learning_rate": 1.341438903953182e-05, "loss": 0.0929, "step": 3556 }, { "epoch": 3.26, "learning_rate": 1.3410900976259393e-05, "loss": 0.1079, "step": 3557 }, { "epoch": 3.26, "learning_rate": 1.3407412443295868e-05, "loss": 0.0874, "step": 3558 }, { "epoch": 3.26, "learning_rate": 1.3403923441121635e-05, "loss": 0.0942, "step": 3559 }, { "epoch": 3.26, "learning_rate": 1.3400433970217137e-05, "loss": 0.0967, "step": 3560 }, { "epoch": 3.26, "learning_rate": 1.3396944031062886e-05, "loss": 0.102, "step": 3561 }, { "epoch": 3.26, "learning_rate": 1.3393453624139454e-05, "loss": 0.1044, "step": 3562 }, { "epoch": 3.27, "learning_rate": 1.3389962749927482e-05, "loss": 0.0953, "step": 3563 }, { "epoch": 3.27, "learning_rate": 1.3386471408907675e-05, "loss": 0.0954, "step": 3564 }, { "epoch": 3.27, "learning_rate": 1.3382979601560796e-05, "loss": 0.1038, "step": 3565 }, { "epoch": 3.27, "learning_rate": 1.3379487328367684e-05, "loss": 0.1049, "step": 3566 }, { "epoch": 3.27, "learning_rate": 1.3375994589809232e-05, "loss": 0.1056, "step": 3567 }, { "epoch": 3.27, "learning_rate": 1.3372501386366399e-05, "loss": 0.1063, "step": 3568 }, { "epoch": 3.27, "learning_rate": 1.3369007718520208e-05, "loss": 0.0947, "step": 3569 }, { "epoch": 3.27, "learning_rate": 1.3365513586751755e-05, "loss": 0.1131, "step": 3570 }, { "epoch": 3.27, "learning_rate": 1.3362018991542183e-05, "loss": 0.0948, "step": 3571 }, { "epoch": 3.27, "learning_rate": 1.3358523933372714e-05, "loss": 0.1093, "step": 3572 }, { "epoch": 3.27, "learning_rate": 1.3355028412724625e-05, "loss": 0.1073, "step": 3573 }, { "epoch": 3.28, "learning_rate": 1.3351532430079258e-05, "loss": 0.0997, "step": 3574 }, { "epoch": 3.28, "learning_rate": 1.3348035985918027e-05, "loss": 0.0883, "step": 3575 }, { "epoch": 3.28, "learning_rate": 1.3344539080722395e-05, "loss": 0.0978, "step": 3576 }, { "epoch": 3.28, "learning_rate": 1.3341041714973901e-05, "loss": 0.0879, "step": 3577 }, { "epoch": 3.28, "learning_rate": 1.3337543889154139e-05, "loss": 0.1042, "step": 3578 }, { "epoch": 3.28, "learning_rate": 1.3334045603744772e-05, "loss": 0.0892, "step": 3579 }, { "epoch": 3.28, "learning_rate": 1.3330546859227524e-05, "loss": 0.1011, "step": 3580 }, { "epoch": 3.28, "learning_rate": 1.3327047656084182e-05, "loss": 0.1039, "step": 3581 }, { "epoch": 3.28, "learning_rate": 1.3323547994796597e-05, "loss": 0.104, "step": 3582 }, { "epoch": 3.28, "learning_rate": 1.3320047875846682e-05, "loss": 0.0976, "step": 3583 }, { "epoch": 3.29, "learning_rate": 1.3316547299716419e-05, "loss": 0.1003, "step": 3584 }, { "epoch": 3.29, "learning_rate": 1.3313046266887837e-05, "loss": 0.0966, "step": 3585 }, { "epoch": 3.29, "learning_rate": 1.3309544777843046e-05, "loss": 0.1013, "step": 3586 }, { "epoch": 3.29, "learning_rate": 1.330604283306421e-05, "loss": 0.1075, "step": 3587 }, { "epoch": 3.29, "learning_rate": 1.3302540433033557e-05, "loss": 0.0995, "step": 3588 }, { "epoch": 3.29, "learning_rate": 1.3299037578233376e-05, "loss": 0.0961, "step": 3589 }, { "epoch": 3.29, "learning_rate": 1.3295534269146021e-05, "loss": 0.1004, "step": 3590 }, { "epoch": 3.29, "learning_rate": 1.3292030506253913e-05, "loss": 0.0999, "step": 3591 }, { "epoch": 3.29, "learning_rate": 1.3288526290039524e-05, "loss": 0.0942, "step": 3592 }, { "epoch": 3.29, "learning_rate": 1.3285021620985397e-05, "loss": 0.1077, "step": 3593 }, { "epoch": 3.29, "learning_rate": 1.3281516499574134e-05, "loss": 0.0974, "step": 3594 }, { "epoch": 3.3, "learning_rate": 1.3278010926288405e-05, "loss": 0.1051, "step": 3595 }, { "epoch": 3.3, "learning_rate": 1.3274504901610935e-05, "loss": 0.1077, "step": 3596 }, { "epoch": 3.3, "learning_rate": 1.3270998426024508e-05, "loss": 0.1013, "step": 3597 }, { "epoch": 3.3, "learning_rate": 1.3267491500011987e-05, "loss": 0.0952, "step": 3598 }, { "epoch": 3.3, "learning_rate": 1.3263984124056278e-05, "loss": 0.0959, "step": 3599 }, { "epoch": 3.3, "learning_rate": 1.3260476298640362e-05, "loss": 0.1013, "step": 3600 }, { "epoch": 3.3, "learning_rate": 1.3256968024247271e-05, "loss": 0.1042, "step": 3601 }, { "epoch": 3.3, "learning_rate": 1.325345930136011e-05, "loss": 0.0938, "step": 3602 }, { "epoch": 3.3, "learning_rate": 1.3249950130462037e-05, "loss": 0.1073, "step": 3603 }, { "epoch": 3.3, "learning_rate": 1.324644051203628e-05, "loss": 0.105, "step": 3604 }, { "epoch": 3.3, "learning_rate": 1.3242930446566117e-05, "loss": 0.0897, "step": 3605 }, { "epoch": 3.31, "learning_rate": 1.3239419934534899e-05, "loss": 0.1059, "step": 3606 }, { "epoch": 3.31, "learning_rate": 1.3235908976426036e-05, "loss": 0.1108, "step": 3607 }, { "epoch": 3.31, "learning_rate": 1.323239757272299e-05, "loss": 0.1016, "step": 3608 }, { "epoch": 3.31, "learning_rate": 1.3228885723909297e-05, "loss": 0.0949, "step": 3609 }, { "epoch": 3.31, "learning_rate": 1.3225373430468547e-05, "loss": 0.0996, "step": 3610 }, { "epoch": 3.31, "learning_rate": 1.3221860692884396e-05, "loss": 0.0969, "step": 3611 }, { "epoch": 3.31, "learning_rate": 1.3218347511640552e-05, "loss": 0.0934, "step": 3612 }, { "epoch": 3.31, "learning_rate": 1.3214833887220794e-05, "loss": 0.0959, "step": 3613 }, { "epoch": 3.31, "learning_rate": 1.3211319820108964e-05, "loss": 0.0985, "step": 3614 }, { "epoch": 3.31, "learning_rate": 1.3207805310788948e-05, "loss": 0.1146, "step": 3615 }, { "epoch": 3.31, "learning_rate": 1.3204290359744716e-05, "loss": 0.1, "step": 3616 }, { "epoch": 3.32, "learning_rate": 1.3200774967460276e-05, "loss": 0.1002, "step": 3617 }, { "epoch": 3.32, "learning_rate": 1.3197259134419718e-05, "loss": 0.1028, "step": 3618 }, { "epoch": 3.32, "learning_rate": 1.3193742861107174e-05, "loss": 0.0857, "step": 3619 }, { "epoch": 3.32, "learning_rate": 1.3190226148006853e-05, "loss": 0.0969, "step": 3620 }, { "epoch": 3.32, "learning_rate": 1.318670899560301e-05, "loss": 0.0993, "step": 3621 }, { "epoch": 3.32, "learning_rate": 1.318319140437997e-05, "loss": 0.0948, "step": 3622 }, { "epoch": 3.32, "learning_rate": 1.3179673374822114e-05, "loss": 0.092, "step": 3623 }, { "epoch": 3.32, "learning_rate": 1.317615490741389e-05, "loss": 0.0921, "step": 3624 }, { "epoch": 3.32, "learning_rate": 1.3172636002639793e-05, "loss": 0.0902, "step": 3625 }, { "epoch": 3.32, "learning_rate": 1.3169116660984396e-05, "loss": 0.1148, "step": 3626 }, { "epoch": 3.32, "learning_rate": 1.3165596882932317e-05, "loss": 0.0963, "step": 3627 }, { "epoch": 3.33, "learning_rate": 1.316207666896824e-05, "loss": 0.107, "step": 3628 }, { "epoch": 3.33, "learning_rate": 1.315855601957691e-05, "loss": 0.1122, "step": 3629 }, { "epoch": 3.33, "learning_rate": 1.3155034935243131e-05, "loss": 0.1006, "step": 3630 }, { "epoch": 3.33, "learning_rate": 1.3151513416451766e-05, "loss": 0.0969, "step": 3631 }, { "epoch": 3.33, "learning_rate": 1.3147991463687738e-05, "loss": 0.0941, "step": 3632 }, { "epoch": 3.33, "learning_rate": 1.3144469077436033e-05, "loss": 0.0988, "step": 3633 }, { "epoch": 3.33, "learning_rate": 1.3140946258181693e-05, "loss": 0.0999, "step": 3634 }, { "epoch": 3.33, "learning_rate": 1.3137423006409815e-05, "loss": 0.0964, "step": 3635 }, { "epoch": 3.33, "learning_rate": 1.3133899322605568e-05, "loss": 0.1035, "step": 3636 }, { "epoch": 3.33, "learning_rate": 1.3130375207254172e-05, "loss": 0.1036, "step": 3637 }, { "epoch": 3.33, "learning_rate": 1.3126850660840904e-05, "loss": 0.0938, "step": 3638 }, { "epoch": 3.34, "learning_rate": 1.3123325683851112e-05, "loss": 0.0956, "step": 3639 }, { "epoch": 3.34, "learning_rate": 1.3119800276770187e-05, "loss": 0.111, "step": 3640 }, { "epoch": 3.34, "learning_rate": 1.3116274440083593e-05, "loss": 0.0961, "step": 3641 }, { "epoch": 3.34, "learning_rate": 1.3112748174276848e-05, "loss": 0.0987, "step": 3642 }, { "epoch": 3.34, "learning_rate": 1.310922147983553e-05, "loss": 0.1005, "step": 3643 }, { "epoch": 3.34, "learning_rate": 1.3105694357245274e-05, "loss": 0.0986, "step": 3644 }, { "epoch": 3.34, "learning_rate": 1.310216680699177e-05, "loss": 0.1054, "step": 3645 }, { "epoch": 3.34, "learning_rate": 1.309863882956078e-05, "loss": 0.0977, "step": 3646 }, { "epoch": 3.34, "learning_rate": 1.3095110425438114e-05, "loss": 0.0975, "step": 3647 }, { "epoch": 3.34, "learning_rate": 1.3091581595109641e-05, "loss": 0.1087, "step": 3648 }, { "epoch": 3.34, "learning_rate": 1.3088052339061296e-05, "loss": 0.0994, "step": 3649 }, { "epoch": 3.35, "learning_rate": 1.3084522657779066e-05, "loss": 0.0867, "step": 3650 }, { "epoch": 3.35, "learning_rate": 1.3080992551748997e-05, "loss": 0.0844, "step": 3651 }, { "epoch": 3.35, "learning_rate": 1.3077462021457197e-05, "loss": 0.0964, "step": 3652 }, { "epoch": 3.35, "learning_rate": 1.307393106738983e-05, "loss": 0.0957, "step": 3653 }, { "epoch": 3.35, "learning_rate": 1.3070399690033115e-05, "loss": 0.0913, "step": 3654 }, { "epoch": 3.35, "learning_rate": 1.3066867889873341e-05, "loss": 0.1061, "step": 3655 }, { "epoch": 3.35, "learning_rate": 1.3063335667396837e-05, "loss": 0.113, "step": 3656 }, { "epoch": 3.35, "learning_rate": 1.305980302309001e-05, "loss": 0.094, "step": 3657 }, { "epoch": 3.35, "learning_rate": 1.3056269957439309e-05, "loss": 0.1015, "step": 3658 }, { "epoch": 3.35, "learning_rate": 1.3052736470931253e-05, "loss": 0.0952, "step": 3659 }, { "epoch": 3.35, "learning_rate": 1.3049202564052408e-05, "loss": 0.1191, "step": 3660 }, { "epoch": 3.36, "learning_rate": 1.3045668237289406e-05, "loss": 0.0927, "step": 3661 }, { "epoch": 3.36, "learning_rate": 1.3042133491128934e-05, "loss": 0.0978, "step": 3662 }, { "epoch": 3.36, "learning_rate": 1.3038598326057737e-05, "loss": 0.1043, "step": 3663 }, { "epoch": 3.36, "learning_rate": 1.3035062742562618e-05, "loss": 0.1044, "step": 3664 }, { "epoch": 3.36, "learning_rate": 1.3031526741130435e-05, "loss": 0.0957, "step": 3665 }, { "epoch": 3.36, "learning_rate": 1.3027990322248112e-05, "loss": 0.0979, "step": 3666 }, { "epoch": 3.36, "learning_rate": 1.3024453486402617e-05, "loss": 0.1069, "step": 3667 }, { "epoch": 3.36, "learning_rate": 1.3020916234080985e-05, "loss": 0.1007, "step": 3668 }, { "epoch": 3.36, "learning_rate": 1.3017378565770304e-05, "loss": 0.1078, "step": 3669 }, { "epoch": 3.36, "learning_rate": 1.3013840481957729e-05, "loss": 0.095, "step": 3670 }, { "epoch": 3.36, "learning_rate": 1.3010301983130458e-05, "loss": 0.0963, "step": 3671 }, { "epoch": 3.37, "learning_rate": 1.3006763069775751e-05, "loss": 0.1003, "step": 3672 }, { "epoch": 3.37, "learning_rate": 1.3003223742380935e-05, "loss": 0.1011, "step": 3673 }, { "epoch": 3.37, "learning_rate": 1.2999684001433378e-05, "loss": 0.0969, "step": 3674 }, { "epoch": 3.37, "learning_rate": 1.2996143847420517e-05, "loss": 0.0974, "step": 3675 }, { "epoch": 3.37, "learning_rate": 1.2992603280829838e-05, "loss": 0.0947, "step": 3676 }, { "epoch": 3.37, "learning_rate": 1.2989062302148892e-05, "loss": 0.1045, "step": 3677 }, { "epoch": 3.37, "learning_rate": 1.2985520911865277e-05, "loss": 0.0926, "step": 3678 }, { "epoch": 3.37, "learning_rate": 1.2981979110466655e-05, "loss": 0.1067, "step": 3679 }, { "epoch": 3.37, "learning_rate": 1.2978436898440747e-05, "loss": 0.1013, "step": 3680 }, { "epoch": 3.37, "learning_rate": 1.2974894276275319e-05, "loss": 0.0923, "step": 3681 }, { "epoch": 3.37, "learning_rate": 1.2971351244458203e-05, "loss": 0.1014, "step": 3682 }, { "epoch": 3.38, "learning_rate": 1.2967807803477285e-05, "loss": 0.0932, "step": 3683 }, { "epoch": 3.38, "learning_rate": 1.2964263953820511e-05, "loss": 0.1034, "step": 3684 }, { "epoch": 3.38, "learning_rate": 1.2960719695975875e-05, "loss": 0.1054, "step": 3685 }, { "epoch": 3.38, "learning_rate": 1.295717503043143e-05, "loss": 0.1038, "step": 3686 }, { "epoch": 3.38, "learning_rate": 1.2953629957675293e-05, "loss": 0.0991, "step": 3687 }, { "epoch": 3.38, "learning_rate": 1.2950084478195626e-05, "loss": 0.0881, "step": 3688 }, { "epoch": 3.38, "learning_rate": 1.2946538592480655e-05, "loss": 0.0989, "step": 3689 }, { "epoch": 3.38, "learning_rate": 1.2942992301018655e-05, "loss": 0.0957, "step": 3690 }, { "epoch": 3.38, "learning_rate": 1.2939445604297965e-05, "loss": 0.0992, "step": 3691 }, { "epoch": 3.38, "learning_rate": 1.2935898502806974e-05, "loss": 0.0989, "step": 3692 }, { "epoch": 3.38, "learning_rate": 1.2932350997034129e-05, "loss": 0.1039, "step": 3693 }, { "epoch": 3.39, "learning_rate": 1.2928803087467928e-05, "loss": 0.099, "step": 3694 }, { "epoch": 3.39, "learning_rate": 1.2925254774596931e-05, "loss": 0.1105, "step": 3695 }, { "epoch": 3.39, "learning_rate": 1.2921706058909757e-05, "loss": 0.0974, "step": 3696 }, { "epoch": 3.39, "learning_rate": 1.2918156940895066e-05, "loss": 0.1013, "step": 3697 }, { "epoch": 3.39, "learning_rate": 1.2914607421041586e-05, "loss": 0.1083, "step": 3698 }, { "epoch": 3.39, "learning_rate": 1.2911057499838096e-05, "loss": 0.0996, "step": 3699 }, { "epoch": 3.39, "learning_rate": 1.290750717777343e-05, "loss": 0.1, "step": 3700 }, { "epoch": 3.39, "learning_rate": 1.290395645533648e-05, "loss": 0.0957, "step": 3701 }, { "epoch": 3.39, "learning_rate": 1.2900405333016183e-05, "loss": 0.0969, "step": 3702 }, { "epoch": 3.39, "learning_rate": 1.2896853811301553e-05, "loss": 0.1002, "step": 3703 }, { "epoch": 3.4, "learning_rate": 1.2893301890681631e-05, "loss": 0.0991, "step": 3704 }, { "epoch": 3.4, "learning_rate": 1.2889749571645536e-05, "loss": 0.1052, "step": 3705 }, { "epoch": 3.4, "learning_rate": 1.2886196854682428e-05, "loss": 0.1082, "step": 3706 }, { "epoch": 3.4, "learning_rate": 1.2882643740281531e-05, "loss": 0.1025, "step": 3707 }, { "epoch": 3.4, "learning_rate": 1.2879090228932112e-05, "loss": 0.1013, "step": 3708 }, { "epoch": 3.4, "learning_rate": 1.2875536321123508e-05, "loss": 0.0941, "step": 3709 }, { "epoch": 3.4, "learning_rate": 1.2871982017345097e-05, "loss": 0.094, "step": 3710 }, { "epoch": 3.4, "learning_rate": 1.2868427318086319e-05, "loss": 0.1013, "step": 3711 }, { "epoch": 3.4, "learning_rate": 1.2864872223836667e-05, "loss": 0.0913, "step": 3712 }, { "epoch": 3.4, "learning_rate": 1.2861316735085687e-05, "loss": 0.1112, "step": 3713 }, { "epoch": 3.4, "learning_rate": 1.2857760852322983e-05, "loss": 0.0943, "step": 3714 }, { "epoch": 3.41, "learning_rate": 1.2854204576038206e-05, "loss": 0.0982, "step": 3715 }, { "epoch": 3.41, "learning_rate": 1.285064790672107e-05, "loss": 0.1034, "step": 3716 }, { "epoch": 3.41, "learning_rate": 1.2847090844861335e-05, "loss": 0.1038, "step": 3717 }, { "epoch": 3.41, "learning_rate": 1.2843533390948821e-05, "loss": 0.09, "step": 3718 }, { "epoch": 3.41, "learning_rate": 1.28399755454734e-05, "loss": 0.0888, "step": 3719 }, { "epoch": 3.41, "learning_rate": 1.2836417308924997e-05, "loss": 0.1064, "step": 3720 }, { "epoch": 3.41, "learning_rate": 1.2832858681793594e-05, "loss": 0.1004, "step": 3721 }, { "epoch": 3.41, "learning_rate": 1.282929966456922e-05, "loss": 0.0912, "step": 3722 }, { "epoch": 3.41, "learning_rate": 1.2825740257741966e-05, "loss": 0.1017, "step": 3723 }, { "epoch": 3.41, "learning_rate": 1.2822180461801969e-05, "loss": 0.1061, "step": 3724 }, { "epoch": 3.41, "learning_rate": 1.2818620277239431e-05, "loss": 0.1038, "step": 3725 }, { "epoch": 3.42, "learning_rate": 1.2815059704544589e-05, "loss": 0.09, "step": 3726 }, { "epoch": 3.42, "learning_rate": 1.281149874420775e-05, "loss": 0.1005, "step": 3727 }, { "epoch": 3.42, "learning_rate": 1.2807937396719273e-05, "loss": 0.0985, "step": 3728 }, { "epoch": 3.42, "learning_rate": 1.2804375662569557e-05, "loss": 0.1061, "step": 3729 }, { "epoch": 3.42, "learning_rate": 1.2800813542249073e-05, "loss": 0.0976, "step": 3730 }, { "epoch": 3.42, "learning_rate": 1.2797251036248326e-05, "loss": 0.1026, "step": 3731 }, { "epoch": 3.42, "learning_rate": 1.279368814505789e-05, "loss": 0.1027, "step": 3732 }, { "epoch": 3.42, "learning_rate": 1.2790124869168382e-05, "loss": 0.1001, "step": 3733 }, { "epoch": 3.42, "learning_rate": 1.2786561209070476e-05, "loss": 0.1039, "step": 3734 }, { "epoch": 3.42, "learning_rate": 1.27829971652549e-05, "loss": 0.0868, "step": 3735 }, { "epoch": 3.42, "learning_rate": 1.2779432738212437e-05, "loss": 0.1014, "step": 3736 }, { "epoch": 3.43, "learning_rate": 1.277586792843391e-05, "loss": 0.0944, "step": 3737 }, { "epoch": 3.43, "learning_rate": 1.2772302736410209e-05, "loss": 0.1002, "step": 3738 }, { "epoch": 3.43, "learning_rate": 1.2768737162632274e-05, "loss": 0.0977, "step": 3739 }, { "epoch": 3.43, "learning_rate": 1.2765171207591092e-05, "loss": 0.1029, "step": 3740 }, { "epoch": 3.43, "learning_rate": 1.2761604871777704e-05, "loss": 0.0956, "step": 3741 }, { "epoch": 3.43, "learning_rate": 1.2758038155683205e-05, "loss": 0.1008, "step": 3742 }, { "epoch": 3.43, "learning_rate": 1.2754471059798746e-05, "loss": 0.1039, "step": 3743 }, { "epoch": 3.43, "learning_rate": 1.2750903584615527e-05, "loss": 0.0943, "step": 3744 }, { "epoch": 3.43, "learning_rate": 1.2747335730624792e-05, "loss": 0.0909, "step": 3745 }, { "epoch": 3.43, "learning_rate": 1.2743767498317855e-05, "loss": 0.1039, "step": 3746 }, { "epoch": 3.43, "learning_rate": 1.2740198888186063e-05, "loss": 0.096, "step": 3747 }, { "epoch": 3.44, "learning_rate": 1.2736629900720832e-05, "loss": 0.1031, "step": 3748 }, { "epoch": 3.44, "learning_rate": 1.2733060536413615e-05, "loss": 0.0979, "step": 3749 }, { "epoch": 3.44, "learning_rate": 1.272949079575593e-05, "loss": 0.0968, "step": 3750 }, { "epoch": 3.44, "learning_rate": 1.2725920679239337e-05, "loss": 0.1051, "step": 3751 }, { "epoch": 3.44, "learning_rate": 1.2722350187355452e-05, "loss": 0.0893, "step": 3752 }, { "epoch": 3.44, "learning_rate": 1.2718779320595945e-05, "loss": 0.1003, "step": 3753 }, { "epoch": 3.44, "learning_rate": 1.271520807945253e-05, "loss": 0.0972, "step": 3754 }, { "epoch": 3.44, "learning_rate": 1.2711636464416985e-05, "loss": 0.0968, "step": 3755 }, { "epoch": 3.44, "learning_rate": 1.2708064475981123e-05, "loss": 0.0993, "step": 3756 }, { "epoch": 3.44, "learning_rate": 1.2704492114636822e-05, "loss": 0.1047, "step": 3757 }, { "epoch": 3.44, "learning_rate": 1.2700919380876004e-05, "loss": 0.0952, "step": 3758 }, { "epoch": 3.45, "learning_rate": 1.2697346275190649e-05, "loss": 0.1012, "step": 3759 }, { "epoch": 3.45, "learning_rate": 1.2693772798072785e-05, "loss": 0.0978, "step": 3760 }, { "epoch": 3.45, "learning_rate": 1.2690198950014483e-05, "loss": 0.0955, "step": 3761 }, { "epoch": 3.45, "learning_rate": 1.2686624731507882e-05, "loss": 0.0961, "step": 3762 }, { "epoch": 3.45, "learning_rate": 1.2683050143045153e-05, "loss": 0.0969, "step": 3763 }, { "epoch": 3.45, "learning_rate": 1.2679475185118535e-05, "loss": 0.0973, "step": 3764 }, { "epoch": 3.45, "learning_rate": 1.2675899858220303e-05, "loss": 0.0887, "step": 3765 }, { "epoch": 3.45, "learning_rate": 1.2672324162842797e-05, "loss": 0.0876, "step": 3766 }, { "epoch": 3.45, "learning_rate": 1.2668748099478396e-05, "loss": 0.1008, "step": 3767 }, { "epoch": 3.45, "learning_rate": 1.2665171668619536e-05, "loss": 0.0922, "step": 3768 }, { "epoch": 3.45, "learning_rate": 1.2661594870758702e-05, "loss": 0.1047, "step": 3769 }, { "epoch": 3.46, "learning_rate": 1.265801770638843e-05, "loss": 0.0921, "step": 3770 }, { "epoch": 3.46, "learning_rate": 1.2654440176001308e-05, "loss": 0.1016, "step": 3771 }, { "epoch": 3.46, "learning_rate": 1.2650862280089966e-05, "loss": 0.0972, "step": 3772 }, { "epoch": 3.46, "learning_rate": 1.26472840191471e-05, "loss": 0.0889, "step": 3773 }, { "epoch": 3.46, "learning_rate": 1.2643705393665436e-05, "loss": 0.0976, "step": 3774 }, { "epoch": 3.46, "learning_rate": 1.264012640413777e-05, "loss": 0.1002, "step": 3775 }, { "epoch": 3.46, "learning_rate": 1.2636547051056937e-05, "loss": 0.1, "step": 3776 }, { "epoch": 3.46, "learning_rate": 1.2632967334915825e-05, "loss": 0.0852, "step": 3777 }, { "epoch": 3.46, "learning_rate": 1.2629387256207366e-05, "loss": 0.0878, "step": 3778 }, { "epoch": 3.46, "learning_rate": 1.2625806815424558e-05, "loss": 0.0901, "step": 3779 }, { "epoch": 3.46, "learning_rate": 1.2622226013060428e-05, "loss": 0.095, "step": 3780 }, { "epoch": 3.47, "learning_rate": 1.2618644849608068e-05, "loss": 0.0962, "step": 3781 }, { "epoch": 3.47, "learning_rate": 1.2615063325560614e-05, "loss": 0.0915, "step": 3782 }, { "epoch": 3.47, "learning_rate": 1.2611481441411253e-05, "loss": 0.0957, "step": 3783 }, { "epoch": 3.47, "learning_rate": 1.260789919765322e-05, "loss": 0.1082, "step": 3784 }, { "epoch": 3.47, "learning_rate": 1.2604316594779803e-05, "loss": 0.0986, "step": 3785 }, { "epoch": 3.47, "learning_rate": 1.260073363328433e-05, "loss": 0.095, "step": 3786 }, { "epoch": 3.47, "learning_rate": 1.2597150313660194e-05, "loss": 0.0946, "step": 3787 }, { "epoch": 3.47, "learning_rate": 1.259356663640082e-05, "loss": 0.1058, "step": 3788 }, { "epoch": 3.47, "learning_rate": 1.25899826019997e-05, "loss": 0.1005, "step": 3789 }, { "epoch": 3.47, "learning_rate": 1.2586398210950359e-05, "loss": 0.0966, "step": 3790 }, { "epoch": 3.47, "learning_rate": 1.2582813463746381e-05, "loss": 0.0989, "step": 3791 }, { "epoch": 3.48, "learning_rate": 1.2579228360881399e-05, "loss": 0.102, "step": 3792 }, { "epoch": 3.48, "learning_rate": 1.2575642902849085e-05, "loss": 0.0981, "step": 3793 }, { "epoch": 3.48, "learning_rate": 1.2572057090143174e-05, "loss": 0.1057, "step": 3794 }, { "epoch": 3.48, "learning_rate": 1.256847092325744e-05, "loss": 0.0997, "step": 3795 }, { "epoch": 3.48, "learning_rate": 1.2564884402685711e-05, "loss": 0.1119, "step": 3796 }, { "epoch": 3.48, "learning_rate": 1.2561297528921855e-05, "loss": 0.1041, "step": 3797 }, { "epoch": 3.48, "learning_rate": 1.2557710302459803e-05, "loss": 0.0884, "step": 3798 }, { "epoch": 3.48, "learning_rate": 1.255412272379352e-05, "loss": 0.0895, "step": 3799 }, { "epoch": 3.48, "learning_rate": 1.2550534793417031e-05, "loss": 0.0929, "step": 3800 }, { "epoch": 3.48, "learning_rate": 1.2546946511824403e-05, "loss": 0.0953, "step": 3801 }, { "epoch": 3.48, "learning_rate": 1.254335787950975e-05, "loss": 0.0941, "step": 3802 }, { "epoch": 3.49, "learning_rate": 1.2539768896967244e-05, "loss": 0.0938, "step": 3803 }, { "epoch": 3.49, "learning_rate": 1.253617956469109e-05, "loss": 0.0943, "step": 3804 }, { "epoch": 3.49, "learning_rate": 1.2532589883175561e-05, "loss": 0.0954, "step": 3805 }, { "epoch": 3.49, "learning_rate": 1.2528999852914955e-05, "loss": 0.1134, "step": 3806 }, { "epoch": 3.49, "learning_rate": 1.2525409474403633e-05, "loss": 0.0945, "step": 3807 }, { "epoch": 3.49, "learning_rate": 1.2521818748136005e-05, "loss": 0.1091, "step": 3808 }, { "epoch": 3.49, "learning_rate": 1.251822767460652e-05, "loss": 0.0951, "step": 3809 }, { "epoch": 3.49, "learning_rate": 1.2514636254309683e-05, "loss": 0.0966, "step": 3810 }, { "epoch": 3.49, "learning_rate": 1.2511044487740041e-05, "loss": 0.0951, "step": 3811 }, { "epoch": 3.49, "learning_rate": 1.2507452375392189e-05, "loss": 0.099, "step": 3812 }, { "epoch": 3.49, "learning_rate": 1.2503859917760775e-05, "loss": 0.1055, "step": 3813 }, { "epoch": 3.5, "learning_rate": 1.250026711534049e-05, "loss": 0.0975, "step": 3814 }, { "epoch": 3.5, "learning_rate": 1.249667396862607e-05, "loss": 0.0979, "step": 3815 }, { "epoch": 3.5, "learning_rate": 1.2493080478112305e-05, "loss": 0.0838, "step": 3816 }, { "epoch": 3.5, "learning_rate": 1.2489486644294034e-05, "loss": 0.0924, "step": 3817 }, { "epoch": 3.5, "learning_rate": 1.2485892467666128e-05, "loss": 0.1002, "step": 3818 }, { "epoch": 3.5, "learning_rate": 1.2482297948723522e-05, "loss": 0.0972, "step": 3819 }, { "epoch": 3.5, "learning_rate": 1.2478703087961193e-05, "loss": 0.0943, "step": 3820 }, { "epoch": 3.5, "learning_rate": 1.247510788587416e-05, "loss": 0.0928, "step": 3821 }, { "epoch": 3.5, "learning_rate": 1.2471512342957496e-05, "loss": 0.101, "step": 3822 }, { "epoch": 3.5, "learning_rate": 1.2467916459706316e-05, "loss": 0.0939, "step": 3823 }, { "epoch": 3.51, "learning_rate": 1.2464320236615784e-05, "loss": 0.1003, "step": 3824 }, { "epoch": 3.51, "learning_rate": 1.246072367418111e-05, "loss": 0.0886, "step": 3825 }, { "epoch": 3.51, "learning_rate": 1.2457126772897555e-05, "loss": 0.0969, "step": 3826 }, { "epoch": 3.51, "learning_rate": 1.2453529533260415e-05, "loss": 0.0963, "step": 3827 }, { "epoch": 3.51, "learning_rate": 1.244993195576505e-05, "loss": 0.1087, "step": 3828 }, { "epoch": 3.51, "learning_rate": 1.2446334040906849e-05, "loss": 0.0905, "step": 3829 }, { "epoch": 3.51, "learning_rate": 1.2442735789181262e-05, "loss": 0.0876, "step": 3830 }, { "epoch": 3.51, "learning_rate": 1.2439137201083772e-05, "loss": 0.1091, "step": 3831 }, { "epoch": 3.51, "learning_rate": 1.2435538277109919e-05, "loss": 0.0948, "step": 3832 }, { "epoch": 3.51, "learning_rate": 1.243193901775529e-05, "loss": 0.1153, "step": 3833 }, { "epoch": 3.51, "learning_rate": 1.2428339423515504e-05, "loss": 0.0947, "step": 3834 }, { "epoch": 3.52, "learning_rate": 1.2424739494886245e-05, "loss": 0.0938, "step": 3835 }, { "epoch": 3.52, "learning_rate": 1.2421139232363223e-05, "loss": 0.1023, "step": 3836 }, { "epoch": 3.52, "learning_rate": 1.2417538636442217e-05, "loss": 0.0905, "step": 3837 }, { "epoch": 3.52, "learning_rate": 1.241393770761903e-05, "loss": 0.0965, "step": 3838 }, { "epoch": 3.52, "learning_rate": 1.2410336446389527e-05, "loss": 0.0921, "step": 3839 }, { "epoch": 3.52, "learning_rate": 1.2406734853249604e-05, "loss": 0.086, "step": 3840 }, { "epoch": 3.52, "learning_rate": 1.240313292869522e-05, "loss": 0.0941, "step": 3841 }, { "epoch": 3.52, "learning_rate": 1.2399530673222368e-05, "loss": 0.0941, "step": 3842 }, { "epoch": 3.52, "learning_rate": 1.2395928087327085e-05, "loss": 0.0991, "step": 3843 }, { "epoch": 3.52, "learning_rate": 1.239232517150546e-05, "loss": 0.0975, "step": 3844 }, { "epoch": 3.52, "learning_rate": 1.2388721926253631e-05, "loss": 0.1019, "step": 3845 }, { "epoch": 3.53, "learning_rate": 1.2385118352067766e-05, "loss": 0.0897, "step": 3846 }, { "epoch": 3.53, "learning_rate": 1.2381514449444094e-05, "loss": 0.1093, "step": 3847 }, { "epoch": 3.53, "learning_rate": 1.2377910218878882e-05, "loss": 0.092, "step": 3848 }, { "epoch": 3.53, "learning_rate": 1.237430566086844e-05, "loss": 0.0982, "step": 3849 }, { "epoch": 3.53, "learning_rate": 1.237070077590913e-05, "loss": 0.0852, "step": 3850 }, { "epoch": 3.53, "learning_rate": 1.2367095564497354e-05, "loss": 0.09, "step": 3851 }, { "epoch": 3.53, "learning_rate": 1.236349002712956e-05, "loss": 0.0866, "step": 3852 }, { "epoch": 3.53, "learning_rate": 1.235988416430224e-05, "loss": 0.0939, "step": 3853 }, { "epoch": 3.53, "learning_rate": 1.2356277976511933e-05, "loss": 0.1003, "step": 3854 }, { "epoch": 3.53, "learning_rate": 1.2352671464255221e-05, "loss": 0.0876, "step": 3855 }, { "epoch": 3.53, "learning_rate": 1.2349064628028733e-05, "loss": 0.1007, "step": 3856 }, { "epoch": 3.54, "learning_rate": 1.2345457468329138e-05, "loss": 0.0876, "step": 3857 }, { "epoch": 3.54, "learning_rate": 1.2341849985653155e-05, "loss": 0.0975, "step": 3858 }, { "epoch": 3.54, "learning_rate": 1.2338242180497545e-05, "loss": 0.0937, "step": 3859 }, { "epoch": 3.54, "learning_rate": 1.2334634053359115e-05, "loss": 0.0875, "step": 3860 }, { "epoch": 3.54, "learning_rate": 1.233102560473471e-05, "loss": 0.0929, "step": 3861 }, { "epoch": 3.54, "learning_rate": 1.2327416835121228e-05, "loss": 0.0974, "step": 3862 }, { "epoch": 3.54, "learning_rate": 1.2323807745015604e-05, "loss": 0.1072, "step": 3863 }, { "epoch": 3.54, "learning_rate": 1.2320198334914822e-05, "loss": 0.1093, "step": 3864 }, { "epoch": 3.54, "learning_rate": 1.231658860531591e-05, "loss": 0.0924, "step": 3865 }, { "epoch": 3.54, "learning_rate": 1.2312978556715934e-05, "loss": 0.1009, "step": 3866 }, { "epoch": 3.54, "learning_rate": 1.230936818961201e-05, "loss": 0.1005, "step": 3867 }, { "epoch": 3.55, "learning_rate": 1.23057575045013e-05, "loss": 0.1135, "step": 3868 }, { "epoch": 3.55, "learning_rate": 1.2302146501881002e-05, "loss": 0.0858, "step": 3869 }, { "epoch": 3.55, "learning_rate": 1.2298535182248363e-05, "loss": 0.1069, "step": 3870 }, { "epoch": 3.55, "learning_rate": 1.2294923546100672e-05, "loss": 0.0946, "step": 3871 }, { "epoch": 3.55, "learning_rate": 1.2291311593935263e-05, "loss": 0.1073, "step": 3872 }, { "epoch": 3.55, "learning_rate": 1.228769932624951e-05, "loss": 0.0812, "step": 3873 }, { "epoch": 3.55, "learning_rate": 1.2284086743540837e-05, "loss": 0.0907, "step": 3874 }, { "epoch": 3.55, "learning_rate": 1.2280473846306701e-05, "loss": 0.0921, "step": 3875 }, { "epoch": 3.55, "learning_rate": 1.2276860635044616e-05, "loss": 0.0959, "step": 3876 }, { "epoch": 3.55, "learning_rate": 1.2273247110252124e-05, "loss": 0.1026, "step": 3877 }, { "epoch": 3.55, "learning_rate": 1.2269633272426826e-05, "loss": 0.1101, "step": 3878 }, { "epoch": 3.56, "learning_rate": 1.2266019122066351e-05, "loss": 0.0978, "step": 3879 }, { "epoch": 3.56, "learning_rate": 1.226240465966838e-05, "loss": 0.1066, "step": 3880 }, { "epoch": 3.56, "learning_rate": 1.225878988573064e-05, "loss": 0.0911, "step": 3881 }, { "epoch": 3.56, "learning_rate": 1.225517480075089e-05, "loss": 0.1018, "step": 3882 }, { "epoch": 3.56, "learning_rate": 1.2251559405226943e-05, "loss": 0.0859, "step": 3883 }, { "epoch": 3.56, "learning_rate": 1.2247943699656644e-05, "loss": 0.091, "step": 3884 }, { "epoch": 3.56, "learning_rate": 1.2244327684537889e-05, "loss": 0.0961, "step": 3885 }, { "epoch": 3.56, "learning_rate": 1.2240711360368614e-05, "loss": 0.0966, "step": 3886 }, { "epoch": 3.56, "learning_rate": 1.2237094727646796e-05, "loss": 0.0923, "step": 3887 }, { "epoch": 3.56, "learning_rate": 1.2233477786870463e-05, "loss": 0.0983, "step": 3888 }, { "epoch": 3.56, "learning_rate": 1.2229860538537666e-05, "loss": 0.1025, "step": 3889 }, { "epoch": 3.57, "learning_rate": 1.2226242983146519e-05, "loss": 0.0953, "step": 3890 }, { "epoch": 3.57, "learning_rate": 1.222262512119517e-05, "loss": 0.105, "step": 3891 }, { "epoch": 3.57, "learning_rate": 1.221900695318181e-05, "loss": 0.0992, "step": 3892 }, { "epoch": 3.57, "learning_rate": 1.2215388479604665e-05, "loss": 0.0996, "step": 3893 }, { "epoch": 3.57, "learning_rate": 1.2211769700962015e-05, "loss": 0.1008, "step": 3894 }, { "epoch": 3.57, "learning_rate": 1.2208150617752175e-05, "loss": 0.0935, "step": 3895 }, { "epoch": 3.57, "learning_rate": 1.2204531230473504e-05, "loss": 0.1034, "step": 3896 }, { "epoch": 3.57, "learning_rate": 1.22009115396244e-05, "loss": 0.0832, "step": 3897 }, { "epoch": 3.57, "learning_rate": 1.2197291545703307e-05, "loss": 0.1018, "step": 3898 }, { "epoch": 3.57, "learning_rate": 1.219367124920871e-05, "loss": 0.0911, "step": 3899 }, { "epoch": 3.57, "learning_rate": 1.2190050650639131e-05, "loss": 0.0894, "step": 3900 }, { "epoch": 3.58, "learning_rate": 1.2186429750493143e-05, "loss": 0.0982, "step": 3901 }, { "epoch": 3.58, "learning_rate": 1.2182808549269345e-05, "loss": 0.0915, "step": 3902 }, { "epoch": 3.58, "learning_rate": 1.2179187047466396e-05, "loss": 0.1026, "step": 3903 }, { "epoch": 3.58, "learning_rate": 1.2175565245582983e-05, "loss": 0.0818, "step": 3904 }, { "epoch": 3.58, "learning_rate": 1.2171943144117842e-05, "loss": 0.0949, "step": 3905 }, { "epoch": 3.58, "learning_rate": 1.2168320743569744e-05, "loss": 0.0848, "step": 3906 }, { "epoch": 3.58, "learning_rate": 1.2164698044437503e-05, "loss": 0.1084, "step": 3907 }, { "epoch": 3.58, "learning_rate": 1.216107504721998e-05, "loss": 0.0932, "step": 3908 }, { "epoch": 3.58, "learning_rate": 1.2157451752416068e-05, "loss": 0.0955, "step": 3909 }, { "epoch": 3.58, "learning_rate": 1.2153828160524707e-05, "loss": 0.0955, "step": 3910 }, { "epoch": 3.58, "learning_rate": 1.2150204272044879e-05, "loss": 0.0939, "step": 3911 }, { "epoch": 3.59, "learning_rate": 1.21465800874756e-05, "loss": 0.1001, "step": 3912 }, { "epoch": 3.59, "learning_rate": 1.2142955607315929e-05, "loss": 0.1093, "step": 3913 }, { "epoch": 3.59, "learning_rate": 1.2139330832064975e-05, "loss": 0.1039, "step": 3914 }, { "epoch": 3.59, "learning_rate": 1.2135705762221876e-05, "loss": 0.0896, "step": 3915 }, { "epoch": 3.59, "learning_rate": 1.2132080398285812e-05, "loss": 0.0881, "step": 3916 }, { "epoch": 3.59, "learning_rate": 1.2128454740756015e-05, "loss": 0.1, "step": 3917 }, { "epoch": 3.59, "learning_rate": 1.2124828790131738e-05, "loss": 0.0846, "step": 3918 }, { "epoch": 3.59, "learning_rate": 1.2121202546912292e-05, "loss": 0.104, "step": 3919 }, { "epoch": 3.59, "learning_rate": 1.211757601159702e-05, "loss": 0.0882, "step": 3920 }, { "epoch": 3.59, "learning_rate": 1.21139491846853e-05, "loss": 0.0807, "step": 3921 }, { "epoch": 3.59, "learning_rate": 1.211032206667657e-05, "loss": 0.1012, "step": 3922 }, { "epoch": 3.6, "learning_rate": 1.2106694658070284e-05, "loss": 0.0919, "step": 3923 }, { "epoch": 3.6, "learning_rate": 1.2103066959365952e-05, "loss": 0.1055, "step": 3924 }, { "epoch": 3.6, "learning_rate": 1.2099438971063115e-05, "loss": 0.101, "step": 3925 }, { "epoch": 3.6, "learning_rate": 1.2095810693661363e-05, "loss": 0.0938, "step": 3926 }, { "epoch": 3.6, "learning_rate": 1.2092182127660313e-05, "loss": 0.0852, "step": 3927 }, { "epoch": 3.6, "learning_rate": 1.2088553273559637e-05, "loss": 0.1001, "step": 3928 }, { "epoch": 3.6, "learning_rate": 1.2084924131859033e-05, "loss": 0.099, "step": 3929 }, { "epoch": 3.6, "learning_rate": 1.2081294703058247e-05, "loss": 0.0914, "step": 3930 }, { "epoch": 3.6, "learning_rate": 1.2077664987657064e-05, "loss": 0.0914, "step": 3931 }, { "epoch": 3.6, "learning_rate": 1.2074034986155298e-05, "loss": 0.1027, "step": 3932 }, { "epoch": 3.6, "learning_rate": 1.2070404699052822e-05, "loss": 0.1032, "step": 3933 }, { "epoch": 3.61, "learning_rate": 1.206677412684953e-05, "loss": 0.095, "step": 3934 }, { "epoch": 3.61, "learning_rate": 1.2063143270045368e-05, "loss": 0.0929, "step": 3935 }, { "epoch": 3.61, "learning_rate": 1.205951212914031e-05, "loss": 0.0917, "step": 3936 }, { "epoch": 3.61, "learning_rate": 1.2055880704634377e-05, "loss": 0.0934, "step": 3937 }, { "epoch": 3.61, "learning_rate": 1.205224899702763e-05, "loss": 0.0884, "step": 3938 }, { "epoch": 3.61, "learning_rate": 1.2048617006820158e-05, "loss": 0.0923, "step": 3939 }, { "epoch": 3.61, "learning_rate": 1.2044984734512107e-05, "loss": 0.092, "step": 3940 }, { "epoch": 3.61, "learning_rate": 1.2041352180603643e-05, "loss": 0.1004, "step": 3941 }, { "epoch": 3.61, "learning_rate": 1.2037719345594984e-05, "loss": 0.0818, "step": 3942 }, { "epoch": 3.61, "learning_rate": 1.203408622998638e-05, "loss": 0.0934, "step": 3943 }, { "epoch": 3.62, "learning_rate": 1.2030452834278122e-05, "loss": 0.0917, "step": 3944 }, { "epoch": 3.62, "learning_rate": 1.2026819158970539e-05, "loss": 0.0941, "step": 3945 }, { "epoch": 3.62, "learning_rate": 1.2023185204563999e-05, "loss": 0.0976, "step": 3946 }, { "epoch": 3.62, "learning_rate": 1.201955097155891e-05, "loss": 0.1007, "step": 3947 }, { "epoch": 3.62, "learning_rate": 1.2015916460455714e-05, "loss": 0.107, "step": 3948 }, { "epoch": 3.62, "learning_rate": 1.2012281671754898e-05, "loss": 0.0947, "step": 3949 }, { "epoch": 3.62, "learning_rate": 1.2008646605956975e-05, "loss": 0.089, "step": 3950 }, { "epoch": 3.62, "learning_rate": 1.2005011263562514e-05, "loss": 0.088, "step": 3951 }, { "epoch": 3.62, "learning_rate": 1.2001375645072105e-05, "loss": 0.0895, "step": 3952 }, { "epoch": 3.62, "learning_rate": 1.1997739750986384e-05, "loss": 0.0958, "step": 3953 }, { "epoch": 3.62, "learning_rate": 1.199410358180603e-05, "loss": 0.0996, "step": 3954 }, { "epoch": 3.63, "learning_rate": 1.1990467138031747e-05, "loss": 0.0818, "step": 3955 }, { "epoch": 3.63, "learning_rate": 1.198683042016429e-05, "loss": 0.1025, "step": 3956 }, { "epoch": 3.63, "learning_rate": 1.198319342870444e-05, "loss": 0.1024, "step": 3957 }, { "epoch": 3.63, "learning_rate": 1.197955616415303e-05, "loss": 0.0981, "step": 3958 }, { "epoch": 3.63, "learning_rate": 1.1975918627010914e-05, "loss": 0.1024, "step": 3959 }, { "epoch": 3.63, "learning_rate": 1.1972280817778993e-05, "loss": 0.1041, "step": 3960 }, { "epoch": 3.63, "learning_rate": 1.1968642736958208e-05, "loss": 0.1101, "step": 3961 }, { "epoch": 3.63, "learning_rate": 1.1965004385049528e-05, "loss": 0.1051, "step": 3962 }, { "epoch": 3.63, "learning_rate": 1.1961365762553968e-05, "loss": 0.0976, "step": 3963 }, { "epoch": 3.63, "learning_rate": 1.1957726869972577e-05, "loss": 0.0909, "step": 3964 }, { "epoch": 3.63, "learning_rate": 1.195408770780644e-05, "loss": 0.1049, "step": 3965 }, { "epoch": 3.64, "learning_rate": 1.1950448276556684e-05, "loss": 0.0876, "step": 3966 }, { "epoch": 3.64, "learning_rate": 1.1946808576724465e-05, "loss": 0.0999, "step": 3967 }, { "epoch": 3.64, "learning_rate": 1.1943168608810977e-05, "loss": 0.0855, "step": 3968 }, { "epoch": 3.64, "learning_rate": 1.1939528373317464e-05, "loss": 0.1046, "step": 3969 }, { "epoch": 3.64, "learning_rate": 1.193588787074519e-05, "loss": 0.0849, "step": 3970 }, { "epoch": 3.64, "learning_rate": 1.1932247101595467e-05, "loss": 0.0952, "step": 3971 }, { "epoch": 3.64, "learning_rate": 1.1928606066369637e-05, "loss": 0.1009, "step": 3972 }, { "epoch": 3.64, "learning_rate": 1.1924964765569081e-05, "loss": 0.1006, "step": 3973 }, { "epoch": 3.64, "learning_rate": 1.1921323199695219e-05, "loss": 0.1002, "step": 3974 }, { "epoch": 3.64, "learning_rate": 1.1917681369249501e-05, "loss": 0.1038, "step": 3975 }, { "epoch": 3.64, "learning_rate": 1.1914039274733422e-05, "loss": 0.0843, "step": 3976 }, { "epoch": 3.65, "learning_rate": 1.1910396916648508e-05, "loss": 0.0898, "step": 3977 }, { "epoch": 3.65, "learning_rate": 1.1906754295496324e-05, "loss": 0.1036, "step": 3978 }, { "epoch": 3.65, "learning_rate": 1.1903111411778465e-05, "loss": 0.0992, "step": 3979 }, { "epoch": 3.65, "learning_rate": 1.1899468265996573e-05, "loss": 0.0926, "step": 3980 }, { "epoch": 3.65, "learning_rate": 1.1895824858652316e-05, "loss": 0.0906, "step": 3981 }, { "epoch": 3.65, "learning_rate": 1.18921811902474e-05, "loss": 0.0775, "step": 3982 }, { "epoch": 3.65, "learning_rate": 1.1888537261283575e-05, "loss": 0.0863, "step": 3983 }, { "epoch": 3.65, "learning_rate": 1.1884893072262615e-05, "loss": 0.1005, "step": 3984 }, { "epoch": 3.65, "learning_rate": 1.188124862368634e-05, "loss": 0.0893, "step": 3985 }, { "epoch": 3.65, "learning_rate": 1.1877603916056598e-05, "loss": 0.0916, "step": 3986 }, { "epoch": 3.65, "learning_rate": 1.1873958949875273e-05, "loss": 0.0965, "step": 3987 }, { "epoch": 3.66, "learning_rate": 1.1870313725644296e-05, "loss": 0.0804, "step": 3988 }, { "epoch": 3.66, "learning_rate": 1.186666824386562e-05, "loss": 0.0962, "step": 3989 }, { "epoch": 3.66, "learning_rate": 1.186302250504124e-05, "loss": 0.0952, "step": 3990 }, { "epoch": 3.66, "learning_rate": 1.185937650967318e-05, "loss": 0.0986, "step": 3991 }, { "epoch": 3.66, "learning_rate": 1.1855730258263511e-05, "loss": 0.0857, "step": 3992 }, { "epoch": 3.66, "learning_rate": 1.185208375131433e-05, "loss": 0.0919, "step": 3993 }, { "epoch": 3.66, "learning_rate": 1.1848436989327769e-05, "loss": 0.0954, "step": 3994 }, { "epoch": 3.66, "learning_rate": 1.1844789972806005e-05, "loss": 0.1012, "step": 3995 }, { "epoch": 3.66, "learning_rate": 1.1841142702251236e-05, "loss": 0.0942, "step": 3996 }, { "epoch": 3.66, "learning_rate": 1.1837495178165706e-05, "loss": 0.0878, "step": 3997 }, { "epoch": 3.66, "learning_rate": 1.1833847401051686e-05, "loss": 0.1001, "step": 3998 }, { "epoch": 3.67, "learning_rate": 1.1830199371411489e-05, "loss": 0.0942, "step": 3999 }, { "epoch": 3.67, "learning_rate": 1.1826551089747456e-05, "loss": 0.0931, "step": 4000 }, { "epoch": 3.67, "learning_rate": 1.182290255656197e-05, "loss": 0.0897, "step": 4001 }, { "epoch": 3.67, "learning_rate": 1.1819253772357442e-05, "loss": 0.0949, "step": 4002 }, { "epoch": 3.67, "learning_rate": 1.1815604737636321e-05, "loss": 0.097, "step": 4003 }, { "epoch": 3.67, "learning_rate": 1.1811955452901092e-05, "loss": 0.0999, "step": 4004 }, { "epoch": 3.67, "learning_rate": 1.1808305918654267e-05, "loss": 0.0887, "step": 4005 }, { "epoch": 3.67, "learning_rate": 1.1804656135398405e-05, "loss": 0.0939, "step": 4006 }, { "epoch": 3.67, "learning_rate": 1.1801006103636086e-05, "loss": 0.0976, "step": 4007 }, { "epoch": 3.67, "learning_rate": 1.1797355823869932e-05, "loss": 0.0942, "step": 4008 }, { "epoch": 3.67, "learning_rate": 1.1793705296602595e-05, "loss": 0.093, "step": 4009 }, { "epoch": 3.68, "learning_rate": 1.1790054522336767e-05, "loss": 0.0947, "step": 4010 }, { "epoch": 3.68, "learning_rate": 1.178640350157517e-05, "loss": 0.0965, "step": 4011 }, { "epoch": 3.68, "learning_rate": 1.1782752234820557e-05, "loss": 0.0885, "step": 4012 }, { "epoch": 3.68, "learning_rate": 1.1779100722575722e-05, "loss": 0.0868, "step": 4013 }, { "epoch": 3.68, "learning_rate": 1.1775448965343488e-05, "loss": 0.1052, "step": 4014 }, { "epoch": 3.68, "learning_rate": 1.1771796963626711e-05, "loss": 0.087, "step": 4015 }, { "epoch": 3.68, "learning_rate": 1.1768144717928282e-05, "loss": 0.0982, "step": 4016 }, { "epoch": 3.68, "learning_rate": 1.176449222875113e-05, "loss": 0.0933, "step": 4017 }, { "epoch": 3.68, "learning_rate": 1.1760839496598208e-05, "loss": 0.0923, "step": 4018 }, { "epoch": 3.68, "learning_rate": 1.1757186521972512e-05, "loss": 0.1006, "step": 4019 }, { "epoch": 3.68, "learning_rate": 1.1753533305377067e-05, "loss": 0.0977, "step": 4020 }, { "epoch": 3.69, "learning_rate": 1.174987984731493e-05, "loss": 0.0823, "step": 4021 }, { "epoch": 3.69, "learning_rate": 1.1746226148289193e-05, "loss": 0.0935, "step": 4022 }, { "epoch": 3.69, "learning_rate": 1.174257220880298e-05, "loss": 0.0963, "step": 4023 }, { "epoch": 3.69, "learning_rate": 1.1738918029359454e-05, "loss": 0.0843, "step": 4024 }, { "epoch": 3.69, "learning_rate": 1.1735263610461802e-05, "loss": 0.103, "step": 4025 }, { "epoch": 3.69, "learning_rate": 1.173160895261325e-05, "loss": 0.0968, "step": 4026 }, { "epoch": 3.69, "learning_rate": 1.1727954056317055e-05, "loss": 0.0959, "step": 4027 }, { "epoch": 3.69, "learning_rate": 1.1724298922076504e-05, "loss": 0.097, "step": 4028 }, { "epoch": 3.69, "learning_rate": 1.1720643550394928e-05, "loss": 0.0904, "step": 4029 }, { "epoch": 3.69, "learning_rate": 1.171698794177567e-05, "loss": 0.0977, "step": 4030 }, { "epoch": 3.69, "learning_rate": 1.171333209672213e-05, "loss": 0.0948, "step": 4031 }, { "epoch": 3.7, "learning_rate": 1.1709676015737721e-05, "loss": 0.0957, "step": 4032 }, { "epoch": 3.7, "learning_rate": 1.1706019699325898e-05, "loss": 0.0902, "step": 4033 }, { "epoch": 3.7, "learning_rate": 1.170236314799015e-05, "loss": 0.0961, "step": 4034 }, { "epoch": 3.7, "learning_rate": 1.1698706362233987e-05, "loss": 0.0898, "step": 4035 }, { "epoch": 3.7, "learning_rate": 1.1695049342560969e-05, "loss": 0.0872, "step": 4036 }, { "epoch": 3.7, "learning_rate": 1.169139208947467e-05, "loss": 0.0889, "step": 4037 }, { "epoch": 3.7, "learning_rate": 1.1687734603478711e-05, "loss": 0.0951, "step": 4038 }, { "epoch": 3.7, "learning_rate": 1.1684076885076732e-05, "loss": 0.0929, "step": 4039 }, { "epoch": 3.7, "learning_rate": 1.1680418934772418e-05, "loss": 0.0893, "step": 4040 }, { "epoch": 3.7, "learning_rate": 1.1676760753069474e-05, "loss": 0.0989, "step": 4041 }, { "epoch": 3.7, "learning_rate": 1.1673102340471645e-05, "loss": 0.1085, "step": 4042 }, { "epoch": 3.71, "learning_rate": 1.1669443697482705e-05, "loss": 0.0927, "step": 4043 }, { "epoch": 3.71, "learning_rate": 1.1665784824606461e-05, "loss": 0.0993, "step": 4044 }, { "epoch": 3.71, "learning_rate": 1.166212572234675e-05, "loss": 0.1043, "step": 4045 }, { "epoch": 3.71, "learning_rate": 1.165846639120744e-05, "loss": 0.0812, "step": 4046 }, { "epoch": 3.71, "learning_rate": 1.165480683169243e-05, "loss": 0.1015, "step": 4047 }, { "epoch": 3.71, "learning_rate": 1.1651147044305657e-05, "loss": 0.0998, "step": 4048 }, { "epoch": 3.71, "learning_rate": 1.1647487029551083e-05, "loss": 0.0962, "step": 4049 }, { "epoch": 3.71, "learning_rate": 1.1643826787932702e-05, "loss": 0.0959, "step": 4050 }, { "epoch": 3.71, "learning_rate": 1.1640166319954535e-05, "loss": 0.094, "step": 4051 }, { "epoch": 3.71, "learning_rate": 1.1636505626120649e-05, "loss": 0.0762, "step": 4052 }, { "epoch": 3.71, "learning_rate": 1.1632844706935124e-05, "loss": 0.0854, "step": 4053 }, { "epoch": 3.72, "learning_rate": 1.1629183562902087e-05, "loss": 0.0903, "step": 4054 }, { "epoch": 3.72, "learning_rate": 1.162552219452568e-05, "loss": 0.0921, "step": 4055 }, { "epoch": 3.72, "learning_rate": 1.1621860602310089e-05, "loss": 0.0925, "step": 4056 }, { "epoch": 3.72, "learning_rate": 1.1618198786759524e-05, "loss": 0.0904, "step": 4057 }, { "epoch": 3.72, "learning_rate": 1.161453674837823e-05, "loss": 0.088, "step": 4058 }, { "epoch": 3.72, "learning_rate": 1.1610874487670481e-05, "loss": 0.0961, "step": 4059 }, { "epoch": 3.72, "learning_rate": 1.1607212005140576e-05, "loss": 0.0939, "step": 4060 }, { "epoch": 3.72, "learning_rate": 1.1603549301292856e-05, "loss": 0.0927, "step": 4061 }, { "epoch": 3.72, "learning_rate": 1.159988637663168e-05, "loss": 0.1026, "step": 4062 }, { "epoch": 3.72, "learning_rate": 1.159622323166145e-05, "loss": 0.1014, "step": 4063 }, { "epoch": 3.73, "learning_rate": 1.1592559866886586e-05, "loss": 0.0819, "step": 4064 }, { "epoch": 3.73, "learning_rate": 1.1588896282811546e-05, "loss": 0.088, "step": 4065 }, { "epoch": 3.73, "learning_rate": 1.1585232479940816e-05, "loss": 0.0974, "step": 4066 }, { "epoch": 3.73, "learning_rate": 1.1581568458778916e-05, "loss": 0.0905, "step": 4067 }, { "epoch": 3.73, "learning_rate": 1.1577904219830387e-05, "loss": 0.0934, "step": 4068 }, { "epoch": 3.73, "learning_rate": 1.157423976359981e-05, "loss": 0.0875, "step": 4069 }, { "epoch": 3.73, "learning_rate": 1.157057509059179e-05, "loss": 0.0897, "step": 4070 }, { "epoch": 3.73, "learning_rate": 1.1566910201310961e-05, "loss": 0.0854, "step": 4071 }, { "epoch": 3.73, "learning_rate": 1.1563245096261994e-05, "loss": 0.0898, "step": 4072 }, { "epoch": 3.73, "learning_rate": 1.155957977594958e-05, "loss": 0.0947, "step": 4073 }, { "epoch": 3.73, "learning_rate": 1.1555914240878446e-05, "loss": 0.0854, "step": 4074 }, { "epoch": 3.74, "learning_rate": 1.1552248491553344e-05, "loss": 0.0925, "step": 4075 }, { "epoch": 3.74, "learning_rate": 1.1548582528479064e-05, "loss": 0.0996, "step": 4076 }, { "epoch": 3.74, "learning_rate": 1.1544916352160418e-05, "loss": 0.0854, "step": 4077 }, { "epoch": 3.74, "learning_rate": 1.1541249963102246e-05, "loss": 0.103, "step": 4078 }, { "epoch": 3.74, "learning_rate": 1.1537583361809425e-05, "loss": 0.0965, "step": 4079 }, { "epoch": 3.74, "learning_rate": 1.1533916548786856e-05, "loss": 0.0957, "step": 4080 }, { "epoch": 3.74, "learning_rate": 1.1530249524539467e-05, "loss": 0.0936, "step": 4081 }, { "epoch": 3.74, "learning_rate": 1.152658228957222e-05, "loss": 0.1175, "step": 4082 }, { "epoch": 3.74, "learning_rate": 1.1522914844390102e-05, "loss": 0.1044, "step": 4083 }, { "epoch": 3.74, "learning_rate": 1.1519247189498136e-05, "loss": 0.0926, "step": 4084 }, { "epoch": 3.74, "learning_rate": 1.1515579325401365e-05, "loss": 0.102, "step": 4085 }, { "epoch": 3.75, "learning_rate": 1.1511911252604867e-05, "loss": 0.084, "step": 4086 }, { "epoch": 3.75, "learning_rate": 1.1508242971613741e-05, "loss": 0.0913, "step": 4087 }, { "epoch": 3.75, "learning_rate": 1.1504574482933125e-05, "loss": 0.0946, "step": 4088 }, { "epoch": 3.75, "learning_rate": 1.1500905787068177e-05, "loss": 0.0993, "step": 4089 }, { "epoch": 3.75, "learning_rate": 1.1497236884524095e-05, "loss": 0.0917, "step": 4090 }, { "epoch": 3.75, "learning_rate": 1.149356777580609e-05, "loss": 0.0916, "step": 4091 }, { "epoch": 3.75, "learning_rate": 1.1489898461419408e-05, "loss": 0.0997, "step": 4092 }, { "epoch": 3.75, "learning_rate": 1.1486228941869335e-05, "loss": 0.0906, "step": 4093 }, { "epoch": 3.75, "learning_rate": 1.1482559217661162e-05, "loss": 0.0864, "step": 4094 }, { "epoch": 3.75, "learning_rate": 1.1478889289300228e-05, "loss": 0.0896, "step": 4095 }, { "epoch": 3.75, "learning_rate": 1.1475219157291892e-05, "loss": 0.0899, "step": 4096 }, { "epoch": 3.76, "learning_rate": 1.147154882214154e-05, "loss": 0.0963, "step": 4097 }, { "epoch": 3.76, "learning_rate": 1.146787828435459e-05, "loss": 0.1098, "step": 4098 }, { "epoch": 3.76, "learning_rate": 1.1464207544436483e-05, "loss": 0.0936, "step": 4099 }, { "epoch": 3.76, "learning_rate": 1.1460536602892695e-05, "loss": 0.0891, "step": 4100 }, { "epoch": 3.76, "learning_rate": 1.1456865460228721e-05, "loss": 0.0881, "step": 4101 }, { "epoch": 3.76, "learning_rate": 1.1453194116950092e-05, "loss": 0.0972, "step": 4102 }, { "epoch": 3.76, "learning_rate": 1.144952257356236e-05, "loss": 0.1015, "step": 4103 }, { "epoch": 3.76, "learning_rate": 1.144585083057111e-05, "loss": 0.0844, "step": 4104 }, { "epoch": 3.76, "learning_rate": 1.1442178888481947e-05, "loss": 0.0891, "step": 4105 }, { "epoch": 3.76, "learning_rate": 1.1438506747800515e-05, "loss": 0.0837, "step": 4106 }, { "epoch": 3.76, "learning_rate": 1.1434834409032472e-05, "loss": 0.0856, "step": 4107 }, { "epoch": 3.77, "learning_rate": 1.1431161872683512e-05, "loss": 0.0911, "step": 4108 }, { "epoch": 3.77, "learning_rate": 1.1427489139259356e-05, "loss": 0.0884, "step": 4109 }, { "epoch": 3.77, "learning_rate": 1.142381620926575e-05, "loss": 0.0881, "step": 4110 }, { "epoch": 3.77, "learning_rate": 1.1420143083208463e-05, "loss": 0.0926, "step": 4111 }, { "epoch": 3.77, "learning_rate": 1.1416469761593301e-05, "loss": 0.0962, "step": 4112 }, { "epoch": 3.77, "learning_rate": 1.141279624492609e-05, "loss": 0.0837, "step": 4113 }, { "epoch": 3.77, "learning_rate": 1.140912253371268e-05, "loss": 0.0854, "step": 4114 }, { "epoch": 3.77, "learning_rate": 1.1405448628458955e-05, "loss": 0.0929, "step": 4115 }, { "epoch": 3.77, "learning_rate": 1.1401774529670824e-05, "loss": 0.0872, "step": 4116 }, { "epoch": 3.77, "learning_rate": 1.1398100237854218e-05, "loss": 0.0809, "step": 4117 }, { "epoch": 3.77, "learning_rate": 1.13944257535151e-05, "loss": 0.0904, "step": 4118 }, { "epoch": 3.78, "learning_rate": 1.1390751077159454e-05, "loss": 0.0974, "step": 4119 }, { "epoch": 3.78, "learning_rate": 1.1387076209293299e-05, "loss": 0.0899, "step": 4120 }, { "epoch": 3.78, "learning_rate": 1.138340115042267e-05, "loss": 0.0891, "step": 4121 }, { "epoch": 3.78, "learning_rate": 1.1379725901053638e-05, "loss": 0.0807, "step": 4122 }, { "epoch": 3.78, "learning_rate": 1.137605046169229e-05, "loss": 0.0909, "step": 4123 }, { "epoch": 3.78, "learning_rate": 1.1372374832844751e-05, "loss": 0.1026, "step": 4124 }, { "epoch": 3.78, "learning_rate": 1.1368699015017161e-05, "loss": 0.0954, "step": 4125 }, { "epoch": 3.78, "learning_rate": 1.1365023008715693e-05, "loss": 0.0989, "step": 4126 }, { "epoch": 3.78, "learning_rate": 1.1361346814446546e-05, "loss": 0.091, "step": 4127 }, { "epoch": 3.78, "learning_rate": 1.1357670432715937e-05, "loss": 0.0922, "step": 4128 }, { "epoch": 3.78, "learning_rate": 1.1353993864030121e-05, "loss": 0.096, "step": 4129 }, { "epoch": 3.79, "learning_rate": 1.1350317108895367e-05, "loss": 0.0989, "step": 4130 }, { "epoch": 3.79, "learning_rate": 1.134664016781798e-05, "loss": 0.1009, "step": 4131 }, { "epoch": 3.79, "learning_rate": 1.134296304130428e-05, "loss": 0.0947, "step": 4132 }, { "epoch": 3.79, "learning_rate": 1.1339285729860622e-05, "loss": 0.0824, "step": 4133 }, { "epoch": 3.79, "learning_rate": 1.1335608233993384e-05, "loss": 0.0826, "step": 4134 }, { "epoch": 3.79, "learning_rate": 1.1331930554208966e-05, "loss": 0.0973, "step": 4135 }, { "epoch": 3.79, "learning_rate": 1.1328252691013793e-05, "loss": 0.0952, "step": 4136 }, { "epoch": 3.79, "learning_rate": 1.1324574644914319e-05, "loss": 0.0901, "step": 4137 }, { "epoch": 3.79, "learning_rate": 1.1320896416417026e-05, "loss": 0.0949, "step": 4138 }, { "epoch": 3.79, "learning_rate": 1.131721800602841e-05, "loss": 0.0926, "step": 4139 }, { "epoch": 3.79, "learning_rate": 1.1313539414255001e-05, "loss": 0.0882, "step": 4140 }, { "epoch": 3.8, "learning_rate": 1.1309860641603357e-05, "loss": 0.0858, "step": 4141 }, { "epoch": 3.8, "learning_rate": 1.1306181688580049e-05, "loss": 0.0824, "step": 4142 }, { "epoch": 3.8, "learning_rate": 1.1302502555691684e-05, "loss": 0.0931, "step": 4143 }, { "epoch": 3.8, "learning_rate": 1.1298823243444888e-05, "loss": 0.0938, "step": 4144 }, { "epoch": 3.8, "learning_rate": 1.1295143752346314e-05, "loss": 0.0912, "step": 4145 }, { "epoch": 3.8, "learning_rate": 1.1291464082902635e-05, "loss": 0.1007, "step": 4146 }, { "epoch": 3.8, "learning_rate": 1.1287784235620554e-05, "loss": 0.0906, "step": 4147 }, { "epoch": 3.8, "learning_rate": 1.1284104211006802e-05, "loss": 0.0913, "step": 4148 }, { "epoch": 3.8, "learning_rate": 1.1280424009568123e-05, "loss": 0.0944, "step": 4149 }, { "epoch": 3.8, "learning_rate": 1.1276743631811295e-05, "loss": 0.0876, "step": 4150 }, { "epoch": 3.8, "learning_rate": 1.1273063078243114e-05, "loss": 0.1047, "step": 4151 }, { "epoch": 3.81, "learning_rate": 1.1269382349370403e-05, "loss": 0.0936, "step": 4152 }, { "epoch": 3.81, "learning_rate": 1.1265701445700015e-05, "loss": 0.0757, "step": 4153 }, { "epoch": 3.81, "learning_rate": 1.1262020367738813e-05, "loss": 0.0965, "step": 4154 }, { "epoch": 3.81, "learning_rate": 1.1258339115993697e-05, "loss": 0.087, "step": 4155 }, { "epoch": 3.81, "learning_rate": 1.1254657690971587e-05, "loss": 0.0919, "step": 4156 }, { "epoch": 3.81, "learning_rate": 1.1250976093179424e-05, "loss": 0.0898, "step": 4157 }, { "epoch": 3.81, "learning_rate": 1.1247294323124174e-05, "loss": 0.0877, "step": 4158 }, { "epoch": 3.81, "learning_rate": 1.1243612381312831e-05, "loss": 0.0849, "step": 4159 }, { "epoch": 3.81, "learning_rate": 1.1239930268252406e-05, "loss": 0.0948, "step": 4160 }, { "epoch": 3.81, "learning_rate": 1.123624798444994e-05, "loss": 0.0894, "step": 4161 }, { "epoch": 3.81, "learning_rate": 1.123256553041249e-05, "loss": 0.0953, "step": 4162 }, { "epoch": 3.82, "learning_rate": 1.1228882906647142e-05, "loss": 0.0985, "step": 4163 }, { "epoch": 3.82, "learning_rate": 1.1225200113661007e-05, "loss": 0.0857, "step": 4164 }, { "epoch": 3.82, "learning_rate": 1.1221517151961211e-05, "loss": 0.0846, "step": 4165 }, { "epoch": 3.82, "learning_rate": 1.121783402205492e-05, "loss": 0.0885, "step": 4166 }, { "epoch": 3.82, "learning_rate": 1.1214150724449296e-05, "loss": 0.0883, "step": 4167 }, { "epoch": 3.82, "learning_rate": 1.1210467259651552e-05, "loss": 0.0844, "step": 4168 }, { "epoch": 3.82, "learning_rate": 1.1206783628168905e-05, "loss": 0.1058, "step": 4169 }, { "epoch": 3.82, "learning_rate": 1.1203099830508609e-05, "loss": 0.088, "step": 4170 }, { "epoch": 3.82, "learning_rate": 1.1199415867177926e-05, "loss": 0.0863, "step": 4171 }, { "epoch": 3.82, "learning_rate": 1.119573173868415e-05, "loss": 0.082, "step": 4172 }, { "epoch": 3.82, "learning_rate": 1.1192047445534601e-05, "loss": 0.092, "step": 4173 }, { "epoch": 3.83, "learning_rate": 1.1188362988236615e-05, "loss": 0.0986, "step": 4174 }, { "epoch": 3.83, "learning_rate": 1.118467836729755e-05, "loss": 0.0942, "step": 4175 }, { "epoch": 3.83, "learning_rate": 1.1180993583224793e-05, "loss": 0.0904, "step": 4176 }, { "epoch": 3.83, "learning_rate": 1.1177308636525745e-05, "loss": 0.081, "step": 4177 }, { "epoch": 3.83, "learning_rate": 1.117362352770784e-05, "loss": 0.0828, "step": 4178 }, { "epoch": 3.83, "learning_rate": 1.1169938257278523e-05, "loss": 0.0911, "step": 4179 }, { "epoch": 3.83, "learning_rate": 1.116625282574527e-05, "loss": 0.0881, "step": 4180 }, { "epoch": 3.83, "learning_rate": 1.1162567233615572e-05, "loss": 0.093, "step": 4181 }, { "epoch": 3.83, "learning_rate": 1.1158881481396951e-05, "loss": 0.0783, "step": 4182 }, { "epoch": 3.83, "learning_rate": 1.1155195569596939e-05, "loss": 0.0886, "step": 4183 }, { "epoch": 3.84, "learning_rate": 1.1151509498723105e-05, "loss": 0.0984, "step": 4184 }, { "epoch": 3.84, "learning_rate": 1.1147823269283027e-05, "loss": 0.0953, "step": 4185 }, { "epoch": 3.84, "learning_rate": 1.1144136881784311e-05, "loss": 0.0833, "step": 4186 }, { "epoch": 3.84, "learning_rate": 1.1140450336734581e-05, "loss": 0.0906, "step": 4187 }, { "epoch": 3.84, "learning_rate": 1.1136763634641488e-05, "loss": 0.1046, "step": 4188 }, { "epoch": 3.84, "learning_rate": 1.1133076776012702e-05, "loss": 0.0882, "step": 4189 }, { "epoch": 3.84, "learning_rate": 1.1129389761355909e-05, "loss": 0.0815, "step": 4190 }, { "epoch": 3.84, "learning_rate": 1.1125702591178829e-05, "loss": 0.0862, "step": 4191 }, { "epoch": 3.84, "learning_rate": 1.112201526598919e-05, "loss": 0.0897, "step": 4192 }, { "epoch": 3.84, "learning_rate": 1.1118327786294754e-05, "loss": 0.094, "step": 4193 }, { "epoch": 3.84, "learning_rate": 1.1114640152603292e-05, "loss": 0.0901, "step": 4194 }, { "epoch": 3.85, "learning_rate": 1.1110952365422605e-05, "loss": 0.1061, "step": 4195 }, { "epoch": 3.85, "learning_rate": 1.1107264425260509e-05, "loss": 0.0879, "step": 4196 }, { "epoch": 3.85, "learning_rate": 1.1103576332624849e-05, "loss": 0.0963, "step": 4197 }, { "epoch": 3.85, "learning_rate": 1.109988808802348e-05, "loss": 0.0943, "step": 4198 }, { "epoch": 3.85, "learning_rate": 1.1096199691964289e-05, "loss": 0.0934, "step": 4199 }, { "epoch": 3.85, "learning_rate": 1.109251114495518e-05, "loss": 0.0957, "step": 4200 }, { "epoch": 3.85, "learning_rate": 1.1088822447504075e-05, "loss": 0.0865, "step": 4201 }, { "epoch": 3.85, "learning_rate": 1.1085133600118918e-05, "loss": 0.0861, "step": 4202 }, { "epoch": 3.85, "learning_rate": 1.1081444603307672e-05, "loss": 0.0905, "step": 4203 }, { "epoch": 3.85, "learning_rate": 1.1077755457578326e-05, "loss": 0.0922, "step": 4204 }, { "epoch": 3.85, "learning_rate": 1.1074066163438883e-05, "loss": 0.0782, "step": 4205 }, { "epoch": 3.86, "learning_rate": 1.1070376721397374e-05, "loss": 0.0895, "step": 4206 }, { "epoch": 3.86, "learning_rate": 1.1066687131961844e-05, "loss": 0.0872, "step": 4207 }, { "epoch": 3.86, "learning_rate": 1.1062997395640361e-05, "loss": 0.078, "step": 4208 }, { "epoch": 3.86, "learning_rate": 1.1059307512941012e-05, "loss": 0.0833, "step": 4209 }, { "epoch": 3.86, "learning_rate": 1.1055617484371899e-05, "loss": 0.0944, "step": 4210 }, { "epoch": 3.86, "learning_rate": 1.105192731044116e-05, "loss": 0.0934, "step": 4211 }, { "epoch": 3.86, "learning_rate": 1.1048236991656935e-05, "loss": 0.0912, "step": 4212 }, { "epoch": 3.86, "learning_rate": 1.1044546528527395e-05, "loss": 0.0786, "step": 4213 }, { "epoch": 3.86, "learning_rate": 1.104085592156073e-05, "loss": 0.0996, "step": 4214 }, { "epoch": 3.86, "learning_rate": 1.103716517126514e-05, "loss": 0.0817, "step": 4215 }, { "epoch": 3.86, "learning_rate": 1.1033474278148863e-05, "loss": 0.0818, "step": 4216 }, { "epoch": 3.87, "learning_rate": 1.1029783242720136e-05, "loss": 0.0904, "step": 4217 }, { "epoch": 3.87, "learning_rate": 1.1026092065487228e-05, "loss": 0.0823, "step": 4218 }, { "epoch": 3.87, "learning_rate": 1.1022400746958428e-05, "loss": 0.1015, "step": 4219 }, { "epoch": 3.87, "learning_rate": 1.1018709287642038e-05, "loss": 0.0888, "step": 4220 }, { "epoch": 3.87, "learning_rate": 1.1015017688046386e-05, "loss": 0.0964, "step": 4221 }, { "epoch": 3.87, "learning_rate": 1.1011325948679812e-05, "loss": 0.0799, "step": 4222 }, { "epoch": 3.87, "learning_rate": 1.1007634070050683e-05, "loss": 0.0865, "step": 4223 }, { "epoch": 3.87, "learning_rate": 1.100394205266738e-05, "loss": 0.0886, "step": 4224 }, { "epoch": 3.87, "learning_rate": 1.1000249897038307e-05, "loss": 0.0957, "step": 4225 }, { "epoch": 3.87, "learning_rate": 1.099655760367188e-05, "loss": 0.0913, "step": 4226 }, { "epoch": 3.87, "learning_rate": 1.0992865173076543e-05, "loss": 0.0923, "step": 4227 }, { "epoch": 3.88, "learning_rate": 1.0989172605760752e-05, "loss": 0.099, "step": 4228 }, { "epoch": 3.88, "learning_rate": 1.0985479902232984e-05, "loss": 0.0856, "step": 4229 }, { "epoch": 3.88, "learning_rate": 1.0981787063001738e-05, "loss": 0.1006, "step": 4230 }, { "epoch": 3.88, "learning_rate": 1.0978094088575529e-05, "loss": 0.0942, "step": 4231 }, { "epoch": 3.88, "learning_rate": 1.0974400979462887e-05, "loss": 0.087, "step": 4232 }, { "epoch": 3.88, "learning_rate": 1.0970707736172363e-05, "loss": 0.0887, "step": 4233 }, { "epoch": 3.88, "learning_rate": 1.0967014359212533e-05, "loss": 0.1018, "step": 4234 }, { "epoch": 3.88, "learning_rate": 1.0963320849091983e-05, "loss": 0.1018, "step": 4235 }, { "epoch": 3.88, "learning_rate": 1.095962720631932e-05, "loss": 0.091, "step": 4236 }, { "epoch": 3.88, "learning_rate": 1.0955933431403167e-05, "loss": 0.0919, "step": 4237 }, { "epoch": 3.88, "learning_rate": 1.095223952485217e-05, "loss": 0.0864, "step": 4238 }, { "epoch": 3.89, "learning_rate": 1.0948545487174996e-05, "loss": 0.088, "step": 4239 }, { "epoch": 3.89, "learning_rate": 1.0944851318880314e-05, "loss": 0.0824, "step": 4240 }, { "epoch": 3.89, "learning_rate": 1.0941157020476829e-05, "loss": 0.0934, "step": 4241 }, { "epoch": 3.89, "learning_rate": 1.0937462592473257e-05, "loss": 0.0889, "step": 4242 }, { "epoch": 3.89, "learning_rate": 1.0933768035378331e-05, "loss": 0.0957, "step": 4243 }, { "epoch": 3.89, "learning_rate": 1.0930073349700798e-05, "loss": 0.0966, "step": 4244 }, { "epoch": 3.89, "learning_rate": 1.0926378535949434e-05, "loss": 0.09, "step": 4245 }, { "epoch": 3.89, "learning_rate": 1.092268359463302e-05, "loss": 0.0861, "step": 4246 }, { "epoch": 3.89, "learning_rate": 1.0918988526260363e-05, "loss": 0.0878, "step": 4247 }, { "epoch": 3.89, "learning_rate": 1.0915293331340286e-05, "loss": 0.0977, "step": 4248 }, { "epoch": 3.89, "learning_rate": 1.0911598010381626e-05, "loss": 0.0948, "step": 4249 }, { "epoch": 3.9, "learning_rate": 1.090790256389324e-05, "loss": 0.0844, "step": 4250 }, { "epoch": 3.9, "learning_rate": 1.0904206992383999e-05, "loss": 0.0878, "step": 4251 }, { "epoch": 3.9, "learning_rate": 1.0900511296362802e-05, "loss": 0.0963, "step": 4252 }, { "epoch": 3.9, "learning_rate": 1.0896815476338547e-05, "loss": 0.094, "step": 4253 }, { "epoch": 3.9, "learning_rate": 1.0893119532820164e-05, "loss": 0.0882, "step": 4254 }, { "epoch": 3.9, "learning_rate": 1.0889423466316602e-05, "loss": 0.0932, "step": 4255 }, { "epoch": 3.9, "learning_rate": 1.0885727277336809e-05, "loss": 0.0855, "step": 4256 }, { "epoch": 3.9, "learning_rate": 1.0882030966389766e-05, "loss": 0.0991, "step": 4257 }, { "epoch": 3.9, "learning_rate": 1.0878334533984468e-05, "loss": 0.0963, "step": 4258 }, { "epoch": 3.9, "learning_rate": 1.0874637980629923e-05, "loss": 0.0912, "step": 4259 }, { "epoch": 3.9, "learning_rate": 1.0870941306835154e-05, "loss": 0.0973, "step": 4260 }, { "epoch": 3.91, "learning_rate": 1.0867244513109207e-05, "loss": 0.0912, "step": 4261 }, { "epoch": 3.91, "learning_rate": 1.0863547599961143e-05, "loss": 0.0825, "step": 4262 }, { "epoch": 3.91, "learning_rate": 1.0859850567900035e-05, "loss": 0.0825, "step": 4263 }, { "epoch": 3.91, "learning_rate": 1.0856153417434976e-05, "loss": 0.0878, "step": 4264 }, { "epoch": 3.91, "learning_rate": 1.0852456149075072e-05, "loss": 0.091, "step": 4265 }, { "epoch": 3.91, "learning_rate": 1.0848758763329455e-05, "loss": 0.0969, "step": 4266 }, { "epoch": 3.91, "learning_rate": 1.0845061260707257e-05, "loss": 0.0935, "step": 4267 }, { "epoch": 3.91, "learning_rate": 1.084136364171764e-05, "loss": 0.0884, "step": 4268 }, { "epoch": 3.91, "learning_rate": 1.083766590686978e-05, "loss": 0.1052, "step": 4269 }, { "epoch": 3.91, "learning_rate": 1.0833968056672854e-05, "loss": 0.0933, "step": 4270 }, { "epoch": 3.91, "learning_rate": 1.0830270091636081e-05, "loss": 0.0849, "step": 4271 }, { "epoch": 3.92, "learning_rate": 1.0826572012268675e-05, "loss": 0.0928, "step": 4272 }, { "epoch": 3.92, "learning_rate": 1.0822873819079873e-05, "loss": 0.0826, "step": 4273 }, { "epoch": 3.92, "learning_rate": 1.0819175512578925e-05, "loss": 0.0915, "step": 4274 }, { "epoch": 3.92, "learning_rate": 1.0815477093275104e-05, "loss": 0.0925, "step": 4275 }, { "epoch": 3.92, "learning_rate": 1.0811778561677687e-05, "loss": 0.0828, "step": 4276 }, { "epoch": 3.92, "learning_rate": 1.0808079918295975e-05, "loss": 0.0791, "step": 4277 }, { "epoch": 3.92, "learning_rate": 1.0804381163639285e-05, "loss": 0.0908, "step": 4278 }, { "epoch": 3.92, "learning_rate": 1.0800682298216942e-05, "loss": 0.0829, "step": 4279 }, { "epoch": 3.92, "learning_rate": 1.0796983322538295e-05, "loss": 0.1, "step": 4280 }, { "epoch": 3.92, "learning_rate": 1.07932842371127e-05, "loss": 0.0963, "step": 4281 }, { "epoch": 3.92, "learning_rate": 1.0789585042449534e-05, "loss": 0.0835, "step": 4282 }, { "epoch": 3.93, "learning_rate": 1.0785885739058183e-05, "loss": 0.0885, "step": 4283 }, { "epoch": 3.93, "learning_rate": 1.0782186327448058e-05, "loss": 0.073, "step": 4284 }, { "epoch": 3.93, "learning_rate": 1.0778486808128579e-05, "loss": 0.0841, "step": 4285 }, { "epoch": 3.93, "learning_rate": 1.0774787181609175e-05, "loss": 0.0883, "step": 4286 }, { "epoch": 3.93, "learning_rate": 1.0771087448399298e-05, "loss": 0.0855, "step": 4287 }, { "epoch": 3.93, "learning_rate": 1.0767387609008413e-05, "loss": 0.0929, "step": 4288 }, { "epoch": 3.93, "learning_rate": 1.0763687663946e-05, "loss": 0.0973, "step": 4289 }, { "epoch": 3.93, "learning_rate": 1.0759987613721547e-05, "loss": 0.096, "step": 4290 }, { "epoch": 3.93, "learning_rate": 1.075628745884457e-05, "loss": 0.0813, "step": 4291 }, { "epoch": 3.93, "learning_rate": 1.0752587199824582e-05, "loss": 0.0838, "step": 4292 }, { "epoch": 3.93, "learning_rate": 1.0748886837171124e-05, "loss": 0.0826, "step": 4293 }, { "epoch": 3.94, "learning_rate": 1.0745186371393752e-05, "loss": 0.0894, "step": 4294 }, { "epoch": 3.94, "learning_rate": 1.0741485803002021e-05, "loss": 0.0908, "step": 4295 }, { "epoch": 3.94, "learning_rate": 1.0737785132505518e-05, "loss": 0.0907, "step": 4296 }, { "epoch": 3.94, "learning_rate": 1.0734084360413833e-05, "loss": 0.0944, "step": 4297 }, { "epoch": 3.94, "learning_rate": 1.0730383487236573e-05, "loss": 0.1066, "step": 4298 }, { "epoch": 3.94, "learning_rate": 1.0726682513483359e-05, "loss": 0.0729, "step": 4299 }, { "epoch": 3.94, "learning_rate": 1.0722981439663829e-05, "loss": 0.0861, "step": 4300 }, { "epoch": 3.94, "learning_rate": 1.0719280266287625e-05, "loss": 0.083, "step": 4301 }, { "epoch": 3.94, "learning_rate": 1.0715578993864417e-05, "loss": 0.089, "step": 4302 }, { "epoch": 3.94, "learning_rate": 1.0711877622903878e-05, "loss": 0.0936, "step": 4303 }, { "epoch": 3.95, "learning_rate": 1.0708176153915695e-05, "loss": 0.075, "step": 4304 }, { "epoch": 3.95, "learning_rate": 1.0704474587409576e-05, "loss": 0.0821, "step": 4305 }, { "epoch": 3.95, "learning_rate": 1.0700772923895235e-05, "loss": 0.0893, "step": 4306 }, { "epoch": 3.95, "learning_rate": 1.06970711638824e-05, "loss": 0.0927, "step": 4307 }, { "epoch": 3.95, "learning_rate": 1.0693369307880817e-05, "loss": 0.1028, "step": 4308 }, { "epoch": 3.95, "learning_rate": 1.0689667356400244e-05, "loss": 0.0835, "step": 4309 }, { "epoch": 3.95, "learning_rate": 1.0685965309950448e-05, "loss": 0.0906, "step": 4310 }, { "epoch": 3.95, "learning_rate": 1.068226316904121e-05, "loss": 0.0835, "step": 4311 }, { "epoch": 3.95, "learning_rate": 1.0678560934182331e-05, "loss": 0.0868, "step": 4312 }, { "epoch": 3.95, "learning_rate": 1.0674858605883616e-05, "loss": 0.0903, "step": 4313 }, { "epoch": 3.95, "learning_rate": 1.0671156184654887e-05, "loss": 0.0799, "step": 4314 }, { "epoch": 3.96, "learning_rate": 1.0667453671005976e-05, "loss": 0.087, "step": 4315 }, { "epoch": 3.96, "learning_rate": 1.0663751065446735e-05, "loss": 0.0862, "step": 4316 }, { "epoch": 3.96, "learning_rate": 1.0660048368487019e-05, "loss": 0.0877, "step": 4317 }, { "epoch": 3.96, "learning_rate": 1.0656345580636702e-05, "loss": 0.0869, "step": 4318 }, { "epoch": 3.96, "learning_rate": 1.0652642702405672e-05, "loss": 0.0961, "step": 4319 }, { "epoch": 3.96, "learning_rate": 1.064893973430382e-05, "loss": 0.0796, "step": 4320 }, { "epoch": 3.96, "learning_rate": 1.0645236676841061e-05, "loss": 0.0843, "step": 4321 }, { "epoch": 3.96, "learning_rate": 1.0641533530527313e-05, "loss": 0.0858, "step": 4322 }, { "epoch": 3.96, "learning_rate": 1.0637830295872514e-05, "loss": 0.0788, "step": 4323 }, { "epoch": 3.96, "learning_rate": 1.0634126973386607e-05, "loss": 0.094, "step": 4324 }, { "epoch": 3.96, "learning_rate": 1.063042356357955e-05, "loss": 0.1016, "step": 4325 }, { "epoch": 3.97, "learning_rate": 1.0626720066961318e-05, "loss": 0.081, "step": 4326 }, { "epoch": 3.97, "learning_rate": 1.0623016484041886e-05, "loss": 0.0886, "step": 4327 }, { "epoch": 3.97, "learning_rate": 1.0619312815331258e-05, "loss": 0.0844, "step": 4328 }, { "epoch": 3.97, "learning_rate": 1.0615609061339431e-05, "loss": 0.0943, "step": 4329 }, { "epoch": 3.97, "learning_rate": 1.0611905222576426e-05, "loss": 0.0833, "step": 4330 }, { "epoch": 3.97, "learning_rate": 1.0608201299552272e-05, "loss": 0.0808, "step": 4331 }, { "epoch": 3.97, "learning_rate": 1.0604497292777016e-05, "loss": 0.0898, "step": 4332 }, { "epoch": 3.97, "learning_rate": 1.0600793202760702e-05, "loss": 0.09, "step": 4333 }, { "epoch": 3.97, "learning_rate": 1.0597089030013397e-05, "loss": 0.0928, "step": 4334 }, { "epoch": 3.97, "learning_rate": 1.0593384775045181e-05, "loss": 0.0958, "step": 4335 }, { "epoch": 3.97, "learning_rate": 1.0589680438366135e-05, "loss": 0.0918, "step": 4336 }, { "epoch": 3.98, "learning_rate": 1.0585976020486358e-05, "loss": 0.0881, "step": 4337 }, { "epoch": 3.98, "learning_rate": 1.0582271521915962e-05, "loss": 0.0831, "step": 4338 }, { "epoch": 3.98, "learning_rate": 1.0578566943165067e-05, "loss": 0.09, "step": 4339 }, { "epoch": 3.98, "learning_rate": 1.0574862284743798e-05, "loss": 0.0818, "step": 4340 }, { "epoch": 3.98, "learning_rate": 1.057115754716231e-05, "loss": 0.0894, "step": 4341 }, { "epoch": 3.98, "learning_rate": 1.0567452730930743e-05, "loss": 0.0833, "step": 4342 }, { "epoch": 3.98, "learning_rate": 1.0563747836559267e-05, "loss": 0.0937, "step": 4343 }, { "epoch": 3.98, "learning_rate": 1.0560042864558058e-05, "loss": 0.0771, "step": 4344 }, { "epoch": 3.98, "learning_rate": 1.0556337815437299e-05, "loss": 0.0931, "step": 4345 }, { "epoch": 3.98, "learning_rate": 1.0552632689707188e-05, "loss": 0.0883, "step": 4346 }, { "epoch": 3.98, "learning_rate": 1.0548927487877932e-05, "loss": 0.0788, "step": 4347 }, { "epoch": 3.99, "learning_rate": 1.0545222210459744e-05, "loss": 0.0944, "step": 4348 }, { "epoch": 3.99, "learning_rate": 1.0541516857962856e-05, "loss": 0.0877, "step": 4349 }, { "epoch": 3.99, "learning_rate": 1.0537811430897506e-05, "loss": 0.0886, "step": 4350 }, { "epoch": 3.99, "learning_rate": 1.0534105929773941e-05, "loss": 0.0901, "step": 4351 }, { "epoch": 3.99, "learning_rate": 1.0530400355102419e-05, "loss": 0.0749, "step": 4352 }, { "epoch": 3.99, "learning_rate": 1.0526694707393209e-05, "loss": 0.0903, "step": 4353 }, { "epoch": 3.99, "learning_rate": 1.0522988987156586e-05, "loss": 0.0846, "step": 4354 }, { "epoch": 3.99, "learning_rate": 1.0519283194902847e-05, "loss": 0.0825, "step": 4355 }, { "epoch": 3.99, "learning_rate": 1.0515577331142281e-05, "loss": 0.0821, "step": 4356 }, { "epoch": 3.99, "learning_rate": 1.0511871396385204e-05, "loss": 0.0791, "step": 4357 }, { "epoch": 3.99, "learning_rate": 1.050816539114193e-05, "loss": 0.0855, "step": 4358 }, { "epoch": 4.0, "learning_rate": 1.0504459315922788e-05, "loss": 0.0827, "step": 4359 }, { "epoch": 4.0, "learning_rate": 1.0500753171238116e-05, "loss": 0.0834, "step": 4360 }, { "epoch": 4.0, "learning_rate": 1.0497046957598257e-05, "loss": 0.0978, "step": 4361 }, { "epoch": 4.0, "learning_rate": 1.0493340675513574e-05, "loss": 0.0795, "step": 4362 }, { "epoch": 4.0, "learning_rate": 1.048963432549443e-05, "loss": 0.0919, "step": 4363 }, { "epoch": 4.0, "learning_rate": 1.04859279080512e-05, "loss": 0.0834, "step": 4364 }, { "epoch": 4.0, "learning_rate": 1.0482221423694263e-05, "loss": 0.0813, "step": 4365 }, { "epoch": 4.0, "learning_rate": 1.0478514872934022e-05, "loss": 0.0836, "step": 4366 }, { "epoch": 4.0, "learning_rate": 1.0474808256280877e-05, "loss": 0.0766, "step": 4367 }, { "epoch": 4.0, "learning_rate": 1.047110157424524e-05, "loss": 0.0707, "step": 4368 }, { "epoch": 4.0, "learning_rate": 1.046739482733753e-05, "loss": 0.0687, "step": 4369 }, { "epoch": 4.01, "learning_rate": 1.0463688016068176e-05, "loss": 0.0777, "step": 4370 }, { "epoch": 4.01, "learning_rate": 1.0459981140947624e-05, "loss": 0.0761, "step": 4371 }, { "epoch": 4.01, "learning_rate": 1.0456274202486313e-05, "loss": 0.0664, "step": 4372 }, { "epoch": 4.01, "learning_rate": 1.0452567201194703e-05, "loss": 0.0668, "step": 4373 }, { "epoch": 4.01, "learning_rate": 1.0448860137583258e-05, "loss": 0.0639, "step": 4374 }, { "epoch": 4.01, "learning_rate": 1.0445153012162457e-05, "loss": 0.0754, "step": 4375 }, { "epoch": 4.01, "learning_rate": 1.0441445825442773e-05, "loss": 0.0751, "step": 4376 }, { "epoch": 4.01, "learning_rate": 1.0437738577934702e-05, "loss": 0.0652, "step": 4377 }, { "epoch": 4.01, "learning_rate": 1.0434031270148743e-05, "loss": 0.0701, "step": 4378 }, { "epoch": 4.01, "learning_rate": 1.0430323902595404e-05, "loss": 0.0706, "step": 4379 }, { "epoch": 4.01, "learning_rate": 1.0426616475785196e-05, "loss": 0.0693, "step": 4380 }, { "epoch": 4.02, "learning_rate": 1.0422908990228644e-05, "loss": 0.0756, "step": 4381 }, { "epoch": 4.02, "learning_rate": 1.0419201446436282e-05, "loss": 0.0769, "step": 4382 }, { "epoch": 4.02, "learning_rate": 1.041549384491865e-05, "loss": 0.0777, "step": 4383 }, { "epoch": 4.02, "learning_rate": 1.0411786186186293e-05, "loss": 0.0781, "step": 4384 }, { "epoch": 4.02, "learning_rate": 1.0408078470749768e-05, "loss": 0.0748, "step": 4385 }, { "epoch": 4.02, "learning_rate": 1.0404370699119637e-05, "loss": 0.0817, "step": 4386 }, { "epoch": 4.02, "learning_rate": 1.0400662871806474e-05, "loss": 0.0695, "step": 4387 }, { "epoch": 4.02, "learning_rate": 1.0396954989320852e-05, "loss": 0.0694, "step": 4388 }, { "epoch": 4.02, "learning_rate": 1.0393247052173363e-05, "loss": 0.0702, "step": 4389 }, { "epoch": 4.02, "learning_rate": 1.0389539060874599e-05, "loss": 0.0749, "step": 4390 }, { "epoch": 4.02, "learning_rate": 1.038583101593516e-05, "loss": 0.0684, "step": 4391 }, { "epoch": 4.03, "learning_rate": 1.0382122917865658e-05, "loss": 0.0796, "step": 4392 }, { "epoch": 4.03, "learning_rate": 1.0378414767176706e-05, "loss": 0.076, "step": 4393 }, { "epoch": 4.03, "learning_rate": 1.0374706564378927e-05, "loss": 0.0721, "step": 4394 }, { "epoch": 4.03, "learning_rate": 1.0370998309982953e-05, "loss": 0.0789, "step": 4395 }, { "epoch": 4.03, "learning_rate": 1.036729000449942e-05, "loss": 0.069, "step": 4396 }, { "epoch": 4.03, "learning_rate": 1.0363581648438975e-05, "loss": 0.084, "step": 4397 }, { "epoch": 4.03, "learning_rate": 1.0359873242312269e-05, "loss": 0.0793, "step": 4398 }, { "epoch": 4.03, "learning_rate": 1.0356164786629956e-05, "loss": 0.0685, "step": 4399 }, { "epoch": 4.03, "learning_rate": 1.0352456281902708e-05, "loss": 0.071, "step": 4400 }, { "epoch": 4.03, "learning_rate": 1.0348747728641194e-05, "loss": 0.071, "step": 4401 }, { "epoch": 4.03, "learning_rate": 1.0345039127356091e-05, "loss": 0.0817, "step": 4402 }, { "epoch": 4.04, "learning_rate": 1.0341330478558087e-05, "loss": 0.0744, "step": 4403 }, { "epoch": 4.04, "learning_rate": 1.0337621782757871e-05, "loss": 0.0735, "step": 4404 }, { "epoch": 4.04, "learning_rate": 1.0333913040466145e-05, "loss": 0.0645, "step": 4405 }, { "epoch": 4.04, "learning_rate": 1.0330204252193612e-05, "loss": 0.0825, "step": 4406 }, { "epoch": 4.04, "learning_rate": 1.032649541845098e-05, "loss": 0.0698, "step": 4407 }, { "epoch": 4.04, "learning_rate": 1.0322786539748973e-05, "loss": 0.0778, "step": 4408 }, { "epoch": 4.04, "learning_rate": 1.0319077616598306e-05, "loss": 0.0866, "step": 4409 }, { "epoch": 4.04, "learning_rate": 1.0315368649509716e-05, "loss": 0.0714, "step": 4410 }, { "epoch": 4.04, "learning_rate": 1.0311659638993933e-05, "loss": 0.0759, "step": 4411 }, { "epoch": 4.04, "learning_rate": 1.0307950585561705e-05, "loss": 0.0789, "step": 4412 }, { "epoch": 4.04, "learning_rate": 1.0304241489723775e-05, "loss": 0.0616, "step": 4413 }, { "epoch": 4.05, "learning_rate": 1.0300532351990899e-05, "loss": 0.0756, "step": 4414 }, { "epoch": 4.05, "learning_rate": 1.0296823172873831e-05, "loss": 0.0688, "step": 4415 }, { "epoch": 4.05, "learning_rate": 1.029311395288334e-05, "loss": 0.0808, "step": 4416 }, { "epoch": 4.05, "learning_rate": 1.0289404692530197e-05, "loss": 0.0714, "step": 4417 }, { "epoch": 4.05, "learning_rate": 1.0285695392325178e-05, "loss": 0.064, "step": 4418 }, { "epoch": 4.05, "learning_rate": 1.0281986052779063e-05, "loss": 0.078, "step": 4419 }, { "epoch": 4.05, "learning_rate": 1.0278276674402639e-05, "loss": 0.0594, "step": 4420 }, { "epoch": 4.05, "learning_rate": 1.02745672577067e-05, "loss": 0.0858, "step": 4421 }, { "epoch": 4.05, "learning_rate": 1.0270857803202039e-05, "loss": 0.0749, "step": 4422 }, { "epoch": 4.05, "learning_rate": 1.0267148311399464e-05, "loss": 0.0751, "step": 4423 }, { "epoch": 4.05, "learning_rate": 1.0263438782809784e-05, "loss": 0.0712, "step": 4424 }, { "epoch": 4.06, "learning_rate": 1.0259729217943805e-05, "loss": 0.0798, "step": 4425 }, { "epoch": 4.06, "learning_rate": 1.0256019617312354e-05, "loss": 0.0735, "step": 4426 }, { "epoch": 4.06, "learning_rate": 1.0252309981426245e-05, "loss": 0.0709, "step": 4427 }, { "epoch": 4.06, "learning_rate": 1.0248600310796313e-05, "loss": 0.0861, "step": 4428 }, { "epoch": 4.06, "learning_rate": 1.0244890605933384e-05, "loss": 0.0714, "step": 4429 }, { "epoch": 4.06, "learning_rate": 1.0241180867348302e-05, "loss": 0.0689, "step": 4430 }, { "epoch": 4.06, "learning_rate": 1.0237471095551903e-05, "loss": 0.0849, "step": 4431 }, { "epoch": 4.06, "learning_rate": 1.0233761291055036e-05, "loss": 0.0695, "step": 4432 }, { "epoch": 4.06, "learning_rate": 1.0230051454368552e-05, "loss": 0.0719, "step": 4433 }, { "epoch": 4.06, "learning_rate": 1.0226341586003308e-05, "loss": 0.0745, "step": 4434 }, { "epoch": 4.07, "learning_rate": 1.0222631686470162e-05, "loss": 0.0762, "step": 4435 }, { "epoch": 4.07, "learning_rate": 1.0218921756279976e-05, "loss": 0.0708, "step": 4436 }, { "epoch": 4.07, "learning_rate": 1.0215211795943625e-05, "loss": 0.0744, "step": 4437 }, { "epoch": 4.07, "learning_rate": 1.0211501805971973e-05, "loss": 0.0661, "step": 4438 }, { "epoch": 4.07, "learning_rate": 1.02077917868759e-05, "loss": 0.0701, "step": 4439 }, { "epoch": 4.07, "learning_rate": 1.0204081739166287e-05, "loss": 0.076, "step": 4440 }, { "epoch": 4.07, "learning_rate": 1.0200371663354023e-05, "loss": 0.0733, "step": 4441 }, { "epoch": 4.07, "learning_rate": 1.0196661559949988e-05, "loss": 0.0698, "step": 4442 }, { "epoch": 4.07, "learning_rate": 1.0192951429465079e-05, "loss": 0.0809, "step": 4443 }, { "epoch": 4.07, "learning_rate": 1.0189241272410191e-05, "loss": 0.077, "step": 4444 }, { "epoch": 4.07, "learning_rate": 1.0185531089296223e-05, "loss": 0.0705, "step": 4445 }, { "epoch": 4.08, "learning_rate": 1.0181820880634081e-05, "loss": 0.0754, "step": 4446 }, { "epoch": 4.08, "learning_rate": 1.0178110646934667e-05, "loss": 0.0854, "step": 4447 }, { "epoch": 4.08, "learning_rate": 1.0174400388708892e-05, "loss": 0.0763, "step": 4448 }, { "epoch": 4.08, "learning_rate": 1.0170690106467672e-05, "loss": 0.0777, "step": 4449 }, { "epoch": 4.08, "learning_rate": 1.0166979800721923e-05, "loss": 0.0814, "step": 4450 }, { "epoch": 4.08, "learning_rate": 1.0163269471982567e-05, "loss": 0.0737, "step": 4451 }, { "epoch": 4.08, "learning_rate": 1.015955912076052e-05, "loss": 0.0753, "step": 4452 }, { "epoch": 4.08, "learning_rate": 1.0155848747566719e-05, "loss": 0.0794, "step": 4453 }, { "epoch": 4.08, "learning_rate": 1.0152138352912082e-05, "loss": 0.0895, "step": 4454 }, { "epoch": 4.08, "learning_rate": 1.0148427937307548e-05, "loss": 0.0711, "step": 4455 }, { "epoch": 4.08, "learning_rate": 1.0144717501264052e-05, "loss": 0.0793, "step": 4456 }, { "epoch": 4.09, "learning_rate": 1.0141007045292528e-05, "loss": 0.0835, "step": 4457 }, { "epoch": 4.09, "learning_rate": 1.013729656990392e-05, "loss": 0.0784, "step": 4458 }, { "epoch": 4.09, "learning_rate": 1.0133586075609172e-05, "loss": 0.0666, "step": 4459 }, { "epoch": 4.09, "learning_rate": 1.012987556291923e-05, "loss": 0.0831, "step": 4460 }, { "epoch": 4.09, "learning_rate": 1.012616503234504e-05, "loss": 0.0751, "step": 4461 }, { "epoch": 4.09, "learning_rate": 1.012245448439755e-05, "loss": 0.0648, "step": 4462 }, { "epoch": 4.09, "learning_rate": 1.011874391958772e-05, "loss": 0.0773, "step": 4463 }, { "epoch": 4.09, "learning_rate": 1.0115033338426502e-05, "loss": 0.0714, "step": 4464 }, { "epoch": 4.09, "learning_rate": 1.011132274142486e-05, "loss": 0.0718, "step": 4465 }, { "epoch": 4.09, "learning_rate": 1.0107612129093746e-05, "loss": 0.0804, "step": 4466 }, { "epoch": 4.09, "learning_rate": 1.0103901501944126e-05, "loss": 0.0754, "step": 4467 }, { "epoch": 4.1, "learning_rate": 1.0100190860486964e-05, "loss": 0.0763, "step": 4468 }, { "epoch": 4.1, "learning_rate": 1.0096480205233228e-05, "loss": 0.0705, "step": 4469 }, { "epoch": 4.1, "learning_rate": 1.009276953669388e-05, "loss": 0.0704, "step": 4470 }, { "epoch": 4.1, "learning_rate": 1.0089058855379894e-05, "loss": 0.0863, "step": 4471 }, { "epoch": 4.1, "learning_rate": 1.0085348161802246e-05, "loss": 0.0821, "step": 4472 }, { "epoch": 4.1, "learning_rate": 1.0081637456471903e-05, "loss": 0.0848, "step": 4473 }, { "epoch": 4.1, "learning_rate": 1.0077926739899842e-05, "loss": 0.0781, "step": 4474 }, { "epoch": 4.1, "learning_rate": 1.007421601259704e-05, "loss": 0.0702, "step": 4475 }, { "epoch": 4.1, "learning_rate": 1.0070505275074476e-05, "loss": 0.0782, "step": 4476 }, { "epoch": 4.1, "learning_rate": 1.0066794527843125e-05, "loss": 0.0723, "step": 4477 }, { "epoch": 4.1, "learning_rate": 1.0063083771413975e-05, "loss": 0.0677, "step": 4478 }, { "epoch": 4.11, "learning_rate": 1.0059373006298002e-05, "loss": 0.0756, "step": 4479 }, { "epoch": 4.11, "learning_rate": 1.005566223300619e-05, "loss": 0.0827, "step": 4480 }, { "epoch": 4.11, "learning_rate": 1.005195145204953e-05, "loss": 0.0754, "step": 4481 }, { "epoch": 4.11, "learning_rate": 1.0048240663938997e-05, "loss": 0.077, "step": 4482 }, { "epoch": 4.11, "learning_rate": 1.0044529869185584e-05, "loss": 0.0806, "step": 4483 }, { "epoch": 4.11, "learning_rate": 1.004081906830028e-05, "loss": 0.0735, "step": 4484 }, { "epoch": 4.11, "learning_rate": 1.0037108261794067e-05, "loss": 0.0859, "step": 4485 }, { "epoch": 4.11, "learning_rate": 1.0033397450177938e-05, "loss": 0.0683, "step": 4486 }, { "epoch": 4.11, "learning_rate": 1.002968663396288e-05, "loss": 0.079, "step": 4487 }, { "epoch": 4.11, "learning_rate": 1.002597581365989e-05, "loss": 0.0723, "step": 4488 }, { "epoch": 4.11, "learning_rate": 1.0022264989779951e-05, "loss": 0.0729, "step": 4489 }, { "epoch": 4.12, "learning_rate": 1.001855416283406e-05, "loss": 0.0734, "step": 4490 }, { "epoch": 4.12, "learning_rate": 1.0014843333333205e-05, "loss": 0.0711, "step": 4491 }, { "epoch": 4.12, "learning_rate": 1.0011132501788378e-05, "loss": 0.0657, "step": 4492 }, { "epoch": 4.12, "learning_rate": 1.0007421668710573e-05, "loss": 0.079, "step": 4493 }, { "epoch": 4.12, "learning_rate": 1.0003710834610785e-05, "loss": 0.0695, "step": 4494 }, { "epoch": 4.12, "learning_rate": 1e-05, "loss": 0.0803, "step": 4495 }, { "epoch": 4.12, "learning_rate": 9.99628916538922e-06, "loss": 0.0657, "step": 4496 }, { "epoch": 4.12, "learning_rate": 9.992578331289427e-06, "loss": 0.0814, "step": 4497 }, { "epoch": 4.12, "learning_rate": 9.988867498211623e-06, "loss": 0.0816, "step": 4498 }, { "epoch": 4.12, "learning_rate": 9.985156666666799e-06, "loss": 0.0776, "step": 4499 }, { "epoch": 4.12, "learning_rate": 9.981445837165946e-06, "loss": 0.0794, "step": 4500 }, { "epoch": 4.13, "learning_rate": 9.977735010220049e-06, "loss": 0.0916, "step": 4501 }, { "epoch": 4.13, "learning_rate": 9.974024186340112e-06, "loss": 0.065, "step": 4502 }, { "epoch": 4.13, "learning_rate": 9.970313366037122e-06, "loss": 0.0822, "step": 4503 }, { "epoch": 4.13, "learning_rate": 9.966602549822064e-06, "loss": 0.0774, "step": 4504 }, { "epoch": 4.13, "learning_rate": 9.962891738205935e-06, "loss": 0.0741, "step": 4505 }, { "epoch": 4.13, "learning_rate": 9.959180931699723e-06, "loss": 0.0814, "step": 4506 }, { "epoch": 4.13, "learning_rate": 9.95547013081442e-06, "loss": 0.0734, "step": 4507 }, { "epoch": 4.13, "learning_rate": 9.951759336061006e-06, "loss": 0.0742, "step": 4508 }, { "epoch": 4.13, "learning_rate": 9.948048547950474e-06, "loss": 0.068, "step": 4509 }, { "epoch": 4.13, "learning_rate": 9.944337766993813e-06, "loss": 0.0809, "step": 4510 }, { "epoch": 4.13, "learning_rate": 9.940626993702003e-06, "loss": 0.0767, "step": 4511 }, { "epoch": 4.14, "learning_rate": 9.936916228586027e-06, "loss": 0.0741, "step": 4512 }, { "epoch": 4.14, "learning_rate": 9.933205472156877e-06, "loss": 0.0789, "step": 4513 }, { "epoch": 4.14, "learning_rate": 9.92949472492553e-06, "loss": 0.0723, "step": 4514 }, { "epoch": 4.14, "learning_rate": 9.925783987402962e-06, "loss": 0.0755, "step": 4515 }, { "epoch": 4.14, "learning_rate": 9.922073260100161e-06, "loss": 0.0804, "step": 4516 }, { "epoch": 4.14, "learning_rate": 9.918362543528102e-06, "loss": 0.0749, "step": 4517 }, { "epoch": 4.14, "learning_rate": 9.91465183819776e-06, "loss": 0.0817, "step": 4518 }, { "epoch": 4.14, "learning_rate": 9.910941144620106e-06, "loss": 0.0725, "step": 4519 }, { "epoch": 4.14, "learning_rate": 9.907230463306123e-06, "loss": 0.0686, "step": 4520 }, { "epoch": 4.14, "learning_rate": 9.903519794766779e-06, "loss": 0.0718, "step": 4521 }, { "epoch": 4.14, "learning_rate": 9.899809139513038e-06, "loss": 0.0777, "step": 4522 }, { "epoch": 4.15, "learning_rate": 9.896098498055876e-06, "loss": 0.0873, "step": 4523 }, { "epoch": 4.15, "learning_rate": 9.892387870906257e-06, "loss": 0.0871, "step": 4524 }, { "epoch": 4.15, "learning_rate": 9.888677258575141e-06, "loss": 0.0786, "step": 4525 }, { "epoch": 4.15, "learning_rate": 9.884966661573498e-06, "loss": 0.0795, "step": 4526 }, { "epoch": 4.15, "learning_rate": 9.881256080412283e-06, "loss": 0.0696, "step": 4527 }, { "epoch": 4.15, "learning_rate": 9.877545515602453e-06, "loss": 0.0796, "step": 4528 }, { "epoch": 4.15, "learning_rate": 9.873834967654965e-06, "loss": 0.0759, "step": 4529 }, { "epoch": 4.15, "learning_rate": 9.870124437080774e-06, "loss": 0.0731, "step": 4530 }, { "epoch": 4.15, "learning_rate": 9.866413924390831e-06, "loss": 0.0724, "step": 4531 }, { "epoch": 4.15, "learning_rate": 9.862703430096083e-06, "loss": 0.0743, "step": 4532 }, { "epoch": 4.15, "learning_rate": 9.858992954707474e-06, "loss": 0.0681, "step": 4533 }, { "epoch": 4.16, "learning_rate": 9.855282498735952e-06, "loss": 0.078, "step": 4534 }, { "epoch": 4.16, "learning_rate": 9.851572062692457e-06, "loss": 0.0816, "step": 4535 }, { "epoch": 4.16, "learning_rate": 9.847861647087919e-06, "loss": 0.0884, "step": 4536 }, { "epoch": 4.16, "learning_rate": 9.844151252433285e-06, "loss": 0.0787, "step": 4537 }, { "epoch": 4.16, "learning_rate": 9.840440879239482e-06, "loss": 0.0876, "step": 4538 }, { "epoch": 4.16, "learning_rate": 9.836730528017438e-06, "loss": 0.0738, "step": 4539 }, { "epoch": 4.16, "learning_rate": 9.833020199278076e-06, "loss": 0.0796, "step": 4540 }, { "epoch": 4.16, "learning_rate": 9.829309893532329e-06, "loss": 0.0745, "step": 4541 }, { "epoch": 4.16, "learning_rate": 9.825599611291112e-06, "loss": 0.0836, "step": 4542 }, { "epoch": 4.16, "learning_rate": 9.82188935306534e-06, "loss": 0.0672, "step": 4543 }, { "epoch": 4.16, "learning_rate": 9.818179119365922e-06, "loss": 0.0833, "step": 4544 }, { "epoch": 4.17, "learning_rate": 9.81446891070378e-06, "loss": 0.0824, "step": 4545 }, { "epoch": 4.17, "learning_rate": 9.810758727589814e-06, "loss": 0.0809, "step": 4546 }, { "epoch": 4.17, "learning_rate": 9.807048570534923e-06, "loss": 0.0651, "step": 4547 }, { "epoch": 4.17, "learning_rate": 9.803338440050015e-06, "loss": 0.0724, "step": 4548 }, { "epoch": 4.17, "learning_rate": 9.799628336645979e-06, "loss": 0.0736, "step": 4549 }, { "epoch": 4.17, "learning_rate": 9.795918260833714e-06, "loss": 0.0701, "step": 4550 }, { "epoch": 4.17, "learning_rate": 9.792208213124102e-06, "loss": 0.0811, "step": 4551 }, { "epoch": 4.17, "learning_rate": 9.78849819402803e-06, "loss": 0.0785, "step": 4552 }, { "epoch": 4.17, "learning_rate": 9.784788204056382e-06, "loss": 0.0781, "step": 4553 }, { "epoch": 4.17, "learning_rate": 9.781078243720026e-06, "loss": 0.0747, "step": 4554 }, { "epoch": 4.18, "learning_rate": 9.777368313529841e-06, "loss": 0.068, "step": 4555 }, { "epoch": 4.18, "learning_rate": 9.773658413996697e-06, "loss": 0.0761, "step": 4556 }, { "epoch": 4.18, "learning_rate": 9.769948545631453e-06, "loss": 0.071, "step": 4557 }, { "epoch": 4.18, "learning_rate": 9.766238708944964e-06, "loss": 0.0741, "step": 4558 }, { "epoch": 4.18, "learning_rate": 9.762528904448102e-06, "loss": 0.0772, "step": 4559 }, { "epoch": 4.18, "learning_rate": 9.758819132651703e-06, "loss": 0.0774, "step": 4560 }, { "epoch": 4.18, "learning_rate": 9.755109394066616e-06, "loss": 0.0814, "step": 4561 }, { "epoch": 4.18, "learning_rate": 9.75139968920369e-06, "loss": 0.0782, "step": 4562 }, { "epoch": 4.18, "learning_rate": 9.747690018573758e-06, "loss": 0.0691, "step": 4563 }, { "epoch": 4.18, "learning_rate": 9.743980382687651e-06, "loss": 0.0695, "step": 4564 }, { "epoch": 4.18, "learning_rate": 9.740270782056195e-06, "loss": 0.074, "step": 4565 }, { "epoch": 4.19, "learning_rate": 9.73656121719022e-06, "loss": 0.0815, "step": 4566 }, { "epoch": 4.19, "learning_rate": 9.73285168860054e-06, "loss": 0.0883, "step": 4567 }, { "epoch": 4.19, "learning_rate": 9.729142196797961e-06, "loss": 0.068, "step": 4568 }, { "epoch": 4.19, "learning_rate": 9.725432742293304e-06, "loss": 0.0781, "step": 4569 }, { "epoch": 4.19, "learning_rate": 9.721723325597366e-06, "loss": 0.0643, "step": 4570 }, { "epoch": 4.19, "learning_rate": 9.71801394722094e-06, "loss": 0.0701, "step": 4571 }, { "epoch": 4.19, "learning_rate": 9.714304607674825e-06, "loss": 0.0728, "step": 4572 }, { "epoch": 4.19, "learning_rate": 9.710595307469805e-06, "loss": 0.0632, "step": 4573 }, { "epoch": 4.19, "learning_rate": 9.706886047116662e-06, "loss": 0.0692, "step": 4574 }, { "epoch": 4.19, "learning_rate": 9.703176827126174e-06, "loss": 0.0827, "step": 4575 }, { "epoch": 4.19, "learning_rate": 9.699467648009104e-06, "loss": 0.0709, "step": 4576 }, { "epoch": 4.2, "learning_rate": 9.695758510276227e-06, "loss": 0.0787, "step": 4577 }, { "epoch": 4.2, "learning_rate": 9.692049414438298e-06, "loss": 0.0787, "step": 4578 }, { "epoch": 4.2, "learning_rate": 9.688340361006067e-06, "loss": 0.0821, "step": 4579 }, { "epoch": 4.2, "learning_rate": 9.684631350490287e-06, "loss": 0.0711, "step": 4580 }, { "epoch": 4.2, "learning_rate": 9.680922383401697e-06, "loss": 0.0761, "step": 4581 }, { "epoch": 4.2, "learning_rate": 9.677213460251034e-06, "loss": 0.0639, "step": 4582 }, { "epoch": 4.2, "learning_rate": 9.67350458154902e-06, "loss": 0.0666, "step": 4583 }, { "epoch": 4.2, "learning_rate": 9.669795747806392e-06, "loss": 0.0775, "step": 4584 }, { "epoch": 4.2, "learning_rate": 9.66608695953386e-06, "loss": 0.0737, "step": 4585 }, { "epoch": 4.2, "learning_rate": 9.662378217242129e-06, "loss": 0.0716, "step": 4586 }, { "epoch": 4.2, "learning_rate": 9.658669521441915e-06, "loss": 0.0756, "step": 4587 }, { "epoch": 4.21, "learning_rate": 9.654960872643912e-06, "loss": 0.0726, "step": 4588 }, { "epoch": 4.21, "learning_rate": 9.651252271358811e-06, "loss": 0.0759, "step": 4589 }, { "epoch": 4.21, "learning_rate": 9.647543718097294e-06, "loss": 0.0699, "step": 4590 }, { "epoch": 4.21, "learning_rate": 9.643835213370047e-06, "loss": 0.0755, "step": 4591 }, { "epoch": 4.21, "learning_rate": 9.640126757687735e-06, "loss": 0.066, "step": 4592 }, { "epoch": 4.21, "learning_rate": 9.636418351561027e-06, "loss": 0.0698, "step": 4593 }, { "epoch": 4.21, "learning_rate": 9.632709995500584e-06, "loss": 0.069, "step": 4594 }, { "epoch": 4.21, "learning_rate": 9.62900169001705e-06, "loss": 0.0732, "step": 4595 }, { "epoch": 4.21, "learning_rate": 9.625293435621078e-06, "loss": 0.0708, "step": 4596 }, { "epoch": 4.21, "learning_rate": 9.6215852328233e-06, "loss": 0.0726, "step": 4597 }, { "epoch": 4.21, "learning_rate": 9.617877082134344e-06, "loss": 0.09, "step": 4598 }, { "epoch": 4.22, "learning_rate": 9.614168984064844e-06, "loss": 0.0673, "step": 4599 }, { "epoch": 4.22, "learning_rate": 9.610460939125406e-06, "loss": 0.082, "step": 4600 }, { "epoch": 4.22, "learning_rate": 9.606752947826639e-06, "loss": 0.0761, "step": 4601 }, { "epoch": 4.22, "learning_rate": 9.60304501067915e-06, "loss": 0.0852, "step": 4602 }, { "epoch": 4.22, "learning_rate": 9.599337128193532e-06, "loss": 0.0786, "step": 4603 }, { "epoch": 4.22, "learning_rate": 9.595629300880365e-06, "loss": 0.0734, "step": 4604 }, { "epoch": 4.22, "learning_rate": 9.591921529250236e-06, "loss": 0.0695, "step": 4605 }, { "epoch": 4.22, "learning_rate": 9.588213813813712e-06, "loss": 0.0691, "step": 4606 }, { "epoch": 4.22, "learning_rate": 9.584506155081356e-06, "loss": 0.0845, "step": 4607 }, { "epoch": 4.22, "learning_rate": 9.580798553563718e-06, "loss": 0.0716, "step": 4608 }, { "epoch": 4.22, "learning_rate": 9.577091009771359e-06, "loss": 0.0749, "step": 4609 }, { "epoch": 4.23, "learning_rate": 9.573383524214809e-06, "loss": 0.077, "step": 4610 }, { "epoch": 4.23, "learning_rate": 9.569676097404598e-06, "loss": 0.0795, "step": 4611 }, { "epoch": 4.23, "learning_rate": 9.565968729851259e-06, "loss": 0.0731, "step": 4612 }, { "epoch": 4.23, "learning_rate": 9.562261422065301e-06, "loss": 0.0819, "step": 4613 }, { "epoch": 4.23, "learning_rate": 9.55855417455723e-06, "loss": 0.0724, "step": 4614 }, { "epoch": 4.23, "learning_rate": 9.554846987837548e-06, "loss": 0.0654, "step": 4615 }, { "epoch": 4.23, "learning_rate": 9.551139862416743e-06, "loss": 0.083, "step": 4616 }, { "epoch": 4.23, "learning_rate": 9.5474327988053e-06, "loss": 0.0812, "step": 4617 }, { "epoch": 4.23, "learning_rate": 9.54372579751369e-06, "loss": 0.0687, "step": 4618 }, { "epoch": 4.23, "learning_rate": 9.54001885905238e-06, "loss": 0.0716, "step": 4619 }, { "epoch": 4.23, "learning_rate": 9.536311983931826e-06, "loss": 0.072, "step": 4620 }, { "epoch": 4.24, "learning_rate": 9.532605172662475e-06, "loss": 0.0711, "step": 4621 }, { "epoch": 4.24, "learning_rate": 9.528898425754762e-06, "loss": 0.0858, "step": 4622 }, { "epoch": 4.24, "learning_rate": 9.525191743719124e-06, "loss": 0.0756, "step": 4623 }, { "epoch": 4.24, "learning_rate": 9.52148512706598e-06, "loss": 0.0741, "step": 4624 }, { "epoch": 4.24, "learning_rate": 9.517778576305736e-06, "loss": 0.069, "step": 4625 }, { "epoch": 4.24, "learning_rate": 9.514072091948805e-06, "loss": 0.0724, "step": 4626 }, { "epoch": 4.24, "learning_rate": 9.510365674505576e-06, "loss": 0.08, "step": 4627 }, { "epoch": 4.24, "learning_rate": 9.506659324486431e-06, "loss": 0.0837, "step": 4628 }, { "epoch": 4.24, "learning_rate": 9.502953042401744e-06, "loss": 0.0722, "step": 4629 }, { "epoch": 4.24, "learning_rate": 9.499246828761887e-06, "loss": 0.0756, "step": 4630 }, { "epoch": 4.24, "learning_rate": 9.495540684077215e-06, "loss": 0.0771, "step": 4631 }, { "epoch": 4.25, "learning_rate": 9.491834608858075e-06, "loss": 0.0739, "step": 4632 }, { "epoch": 4.25, "learning_rate": 9.488128603614797e-06, "loss": 0.072, "step": 4633 }, { "epoch": 4.25, "learning_rate": 9.48442266885772e-06, "loss": 0.0693, "step": 4634 }, { "epoch": 4.25, "learning_rate": 9.480716805097158e-06, "loss": 0.0695, "step": 4635 }, { "epoch": 4.25, "learning_rate": 9.477011012843414e-06, "loss": 0.0778, "step": 4636 }, { "epoch": 4.25, "learning_rate": 9.473305292606796e-06, "loss": 0.0792, "step": 4637 }, { "epoch": 4.25, "learning_rate": 9.469599644897585e-06, "loss": 0.0823, "step": 4638 }, { "epoch": 4.25, "learning_rate": 9.465894070226064e-06, "loss": 0.0722, "step": 4639 }, { "epoch": 4.25, "learning_rate": 9.462188569102497e-06, "loss": 0.0666, "step": 4640 }, { "epoch": 4.25, "learning_rate": 9.458483142037145e-06, "loss": 0.0798, "step": 4641 }, { "epoch": 4.25, "learning_rate": 9.454777789540259e-06, "loss": 0.0739, "step": 4642 }, { "epoch": 4.26, "learning_rate": 9.45107251212207e-06, "loss": 0.0764, "step": 4643 }, { "epoch": 4.26, "learning_rate": 9.447367310292814e-06, "loss": 0.0834, "step": 4644 }, { "epoch": 4.26, "learning_rate": 9.443662184562706e-06, "loss": 0.0755, "step": 4645 }, { "epoch": 4.26, "learning_rate": 9.439957135441947e-06, "loss": 0.0683, "step": 4646 }, { "epoch": 4.26, "learning_rate": 9.436252163440735e-06, "loss": 0.0641, "step": 4647 }, { "epoch": 4.26, "learning_rate": 9.43254726906926e-06, "loss": 0.0707, "step": 4648 }, { "epoch": 4.26, "learning_rate": 9.428842452837697e-06, "loss": 0.0702, "step": 4649 }, { "epoch": 4.26, "learning_rate": 9.425137715256202e-06, "loss": 0.0716, "step": 4650 }, { "epoch": 4.26, "learning_rate": 9.421433056834936e-06, "loss": 0.0719, "step": 4651 }, { "epoch": 4.26, "learning_rate": 9.41772847808404e-06, "loss": 0.0642, "step": 4652 }, { "epoch": 4.26, "learning_rate": 9.414023979513645e-06, "loss": 0.0724, "step": 4653 }, { "epoch": 4.27, "learning_rate": 9.410319561633867e-06, "loss": 0.0693, "step": 4654 }, { "epoch": 4.27, "learning_rate": 9.406615224954822e-06, "loss": 0.0656, "step": 4655 }, { "epoch": 4.27, "learning_rate": 9.402910969986605e-06, "loss": 0.0718, "step": 4656 }, { "epoch": 4.27, "learning_rate": 9.399206797239298e-06, "loss": 0.0728, "step": 4657 }, { "epoch": 4.27, "learning_rate": 9.395502707222987e-06, "loss": 0.0662, "step": 4658 }, { "epoch": 4.27, "learning_rate": 9.39179870044773e-06, "loss": 0.0744, "step": 4659 }, { "epoch": 4.27, "learning_rate": 9.388094777423577e-06, "loss": 0.0783, "step": 4660 }, { "epoch": 4.27, "learning_rate": 9.384390938660572e-06, "loss": 0.0795, "step": 4661 }, { "epoch": 4.27, "learning_rate": 9.380687184668745e-06, "loss": 0.0676, "step": 4662 }, { "epoch": 4.27, "learning_rate": 9.376983515958116e-06, "loss": 0.0615, "step": 4663 }, { "epoch": 4.27, "learning_rate": 9.373279933038689e-06, "loss": 0.0709, "step": 4664 }, { "epoch": 4.28, "learning_rate": 9.369576436420452e-06, "loss": 0.078, "step": 4665 }, { "epoch": 4.28, "learning_rate": 9.365873026613396e-06, "loss": 0.0778, "step": 4666 }, { "epoch": 4.28, "learning_rate": 9.362169704127491e-06, "loss": 0.0787, "step": 4667 }, { "epoch": 4.28, "learning_rate": 9.358466469472687e-06, "loss": 0.0758, "step": 4668 }, { "epoch": 4.28, "learning_rate": 9.354763323158942e-06, "loss": 0.0778, "step": 4669 }, { "epoch": 4.28, "learning_rate": 9.351060265696183e-06, "loss": 0.0755, "step": 4670 }, { "epoch": 4.28, "learning_rate": 9.347357297594333e-06, "loss": 0.0714, "step": 4671 }, { "epoch": 4.28, "learning_rate": 9.343654419363299e-06, "loss": 0.0688, "step": 4672 }, { "epoch": 4.28, "learning_rate": 9.339951631512984e-06, "loss": 0.0755, "step": 4673 }, { "epoch": 4.28, "learning_rate": 9.33624893455327e-06, "loss": 0.0811, "step": 4674 }, { "epoch": 4.29, "learning_rate": 9.332546328994024e-06, "loss": 0.0713, "step": 4675 }, { "epoch": 4.29, "learning_rate": 9.328843815345117e-06, "loss": 0.0711, "step": 4676 }, { "epoch": 4.29, "learning_rate": 9.325141394116389e-06, "loss": 0.0671, "step": 4677 }, { "epoch": 4.29, "learning_rate": 9.321439065817674e-06, "loss": 0.0825, "step": 4678 }, { "epoch": 4.29, "learning_rate": 9.31773683095879e-06, "loss": 0.0716, "step": 4679 }, { "epoch": 4.29, "learning_rate": 9.314034690049555e-06, "loss": 0.0777, "step": 4680 }, { "epoch": 4.29, "learning_rate": 9.310332643599757e-06, "loss": 0.0817, "step": 4681 }, { "epoch": 4.29, "learning_rate": 9.306630692119183e-06, "loss": 0.0731, "step": 4682 }, { "epoch": 4.29, "learning_rate": 9.302928836117603e-06, "loss": 0.0761, "step": 4683 }, { "epoch": 4.29, "learning_rate": 9.299227076104768e-06, "loss": 0.0666, "step": 4684 }, { "epoch": 4.29, "learning_rate": 9.295525412590429e-06, "loss": 0.0675, "step": 4685 }, { "epoch": 4.3, "learning_rate": 9.291823846084305e-06, "loss": 0.0709, "step": 4686 }, { "epoch": 4.3, "learning_rate": 9.288122377096124e-06, "loss": 0.0708, "step": 4687 }, { "epoch": 4.3, "learning_rate": 9.284421006135586e-06, "loss": 0.0872, "step": 4688 }, { "epoch": 4.3, "learning_rate": 9.28071973371238e-06, "loss": 0.0686, "step": 4689 }, { "epoch": 4.3, "learning_rate": 9.277018560336175e-06, "loss": 0.0705, "step": 4690 }, { "epoch": 4.3, "learning_rate": 9.273317486516644e-06, "loss": 0.0691, "step": 4691 }, { "epoch": 4.3, "learning_rate": 9.269616512763432e-06, "loss": 0.0701, "step": 4692 }, { "epoch": 4.3, "learning_rate": 9.265915639586169e-06, "loss": 0.0659, "step": 4693 }, { "epoch": 4.3, "learning_rate": 9.262214867494484e-06, "loss": 0.0678, "step": 4694 }, { "epoch": 4.3, "learning_rate": 9.25851419699798e-06, "loss": 0.0696, "step": 4695 }, { "epoch": 4.3, "learning_rate": 9.254813628606254e-06, "loss": 0.0712, "step": 4696 }, { "epoch": 4.31, "learning_rate": 9.251113162828875e-06, "loss": 0.068, "step": 4697 }, { "epoch": 4.31, "learning_rate": 9.247412800175422e-06, "loss": 0.0711, "step": 4698 }, { "epoch": 4.31, "learning_rate": 9.243712541155437e-06, "loss": 0.0769, "step": 4699 }, { "epoch": 4.31, "learning_rate": 9.240012386278455e-06, "loss": 0.0657, "step": 4700 }, { "epoch": 4.31, "learning_rate": 9.236312336054004e-06, "loss": 0.077, "step": 4701 }, { "epoch": 4.31, "learning_rate": 9.23261239099159e-06, "loss": 0.0661, "step": 4702 }, { "epoch": 4.31, "learning_rate": 9.228912551600705e-06, "loss": 0.0796, "step": 4703 }, { "epoch": 4.31, "learning_rate": 9.225212818390829e-06, "loss": 0.0655, "step": 4704 }, { "epoch": 4.31, "learning_rate": 9.221513191871425e-06, "loss": 0.065, "step": 4705 }, { "epoch": 4.31, "learning_rate": 9.217813672551943e-06, "loss": 0.0774, "step": 4706 }, { "epoch": 4.31, "learning_rate": 9.214114260941819e-06, "loss": 0.0663, "step": 4707 }, { "epoch": 4.32, "learning_rate": 9.210414957550468e-06, "loss": 0.0656, "step": 4708 }, { "epoch": 4.32, "learning_rate": 9.206715762887304e-06, "loss": 0.0676, "step": 4709 }, { "epoch": 4.32, "learning_rate": 9.20301667746171e-06, "loss": 0.0779, "step": 4710 }, { "epoch": 4.32, "learning_rate": 9.19931770178306e-06, "loss": 0.0681, "step": 4711 }, { "epoch": 4.32, "learning_rate": 9.195618836360718e-06, "loss": 0.0634, "step": 4712 }, { "epoch": 4.32, "learning_rate": 9.191920081704028e-06, "loss": 0.078, "step": 4713 }, { "epoch": 4.32, "learning_rate": 9.188221438322315e-06, "loss": 0.0715, "step": 4714 }, { "epoch": 4.32, "learning_rate": 9.184522906724899e-06, "loss": 0.0745, "step": 4715 }, { "epoch": 4.32, "learning_rate": 9.180824487421077e-06, "loss": 0.07, "step": 4716 }, { "epoch": 4.32, "learning_rate": 9.177126180920132e-06, "loss": 0.0763, "step": 4717 }, { "epoch": 4.32, "learning_rate": 9.173427987731327e-06, "loss": 0.0655, "step": 4718 }, { "epoch": 4.33, "learning_rate": 9.16972990836392e-06, "loss": 0.072, "step": 4719 }, { "epoch": 4.33, "learning_rate": 9.166031943327147e-06, "loss": 0.0788, "step": 4720 }, { "epoch": 4.33, "learning_rate": 9.162334093130227e-06, "loss": 0.0743, "step": 4721 }, { "epoch": 4.33, "learning_rate": 9.15863635828236e-06, "loss": 0.0751, "step": 4722 }, { "epoch": 4.33, "learning_rate": 9.154938739292745e-06, "loss": 0.0769, "step": 4723 }, { "epoch": 4.33, "learning_rate": 9.151241236670548e-06, "loss": 0.0731, "step": 4724 }, { "epoch": 4.33, "learning_rate": 9.147543850924927e-06, "loss": 0.0765, "step": 4725 }, { "epoch": 4.33, "learning_rate": 9.143846582565029e-06, "loss": 0.074, "step": 4726 }, { "epoch": 4.33, "learning_rate": 9.140149432099968e-06, "loss": 0.0779, "step": 4727 }, { "epoch": 4.33, "learning_rate": 9.136452400038862e-06, "loss": 0.0731, "step": 4728 }, { "epoch": 4.33, "learning_rate": 9.132755486890795e-06, "loss": 0.0783, "step": 4729 }, { "epoch": 4.34, "learning_rate": 9.129058693164849e-06, "loss": 0.0744, "step": 4730 }, { "epoch": 4.34, "learning_rate": 9.125362019370082e-06, "loss": 0.0663, "step": 4731 }, { "epoch": 4.34, "learning_rate": 9.121665466015532e-06, "loss": 0.076, "step": 4732 }, { "epoch": 4.34, "learning_rate": 9.117969033610235e-06, "loss": 0.0732, "step": 4733 }, { "epoch": 4.34, "learning_rate": 9.114272722663195e-06, "loss": 0.0836, "step": 4734 }, { "epoch": 4.34, "learning_rate": 9.110576533683405e-06, "loss": 0.0759, "step": 4735 }, { "epoch": 4.34, "learning_rate": 9.106880467179836e-06, "loss": 0.0827, "step": 4736 }, { "epoch": 4.34, "learning_rate": 9.103184523661456e-06, "loss": 0.0667, "step": 4737 }, { "epoch": 4.34, "learning_rate": 9.099488703637205e-06, "loss": 0.0749, "step": 4738 }, { "epoch": 4.34, "learning_rate": 9.095793007616001e-06, "loss": 0.0724, "step": 4739 }, { "epoch": 4.34, "learning_rate": 9.092097436106764e-06, "loss": 0.0716, "step": 4740 }, { "epoch": 4.35, "learning_rate": 9.088401989618378e-06, "loss": 0.0696, "step": 4741 }, { "epoch": 4.35, "learning_rate": 9.084706668659719e-06, "loss": 0.0807, "step": 4742 }, { "epoch": 4.35, "learning_rate": 9.081011473739637e-06, "loss": 0.0691, "step": 4743 }, { "epoch": 4.35, "learning_rate": 9.07731640536698e-06, "loss": 0.0618, "step": 4744 }, { "epoch": 4.35, "learning_rate": 9.073621464050571e-06, "loss": 0.072, "step": 4745 }, { "epoch": 4.35, "learning_rate": 9.069926650299202e-06, "loss": 0.081, "step": 4746 }, { "epoch": 4.35, "learning_rate": 9.066231964621672e-06, "loss": 0.0816, "step": 4747 }, { "epoch": 4.35, "learning_rate": 9.062537407526744e-06, "loss": 0.0797, "step": 4748 }, { "epoch": 4.35, "learning_rate": 9.058842979523173e-06, "loss": 0.0599, "step": 4749 }, { "epoch": 4.35, "learning_rate": 9.055148681119688e-06, "loss": 0.0688, "step": 4750 }, { "epoch": 4.35, "learning_rate": 9.051454512825009e-06, "loss": 0.0669, "step": 4751 }, { "epoch": 4.36, "learning_rate": 9.047760475147832e-06, "loss": 0.0752, "step": 4752 }, { "epoch": 4.36, "learning_rate": 9.044066568596838e-06, "loss": 0.0658, "step": 4753 }, { "epoch": 4.36, "learning_rate": 9.040372793680684e-06, "loss": 0.0738, "step": 4754 }, { "epoch": 4.36, "learning_rate": 9.03667915090802e-06, "loss": 0.0692, "step": 4755 }, { "epoch": 4.36, "learning_rate": 9.03298564078747e-06, "loss": 0.0694, "step": 4756 }, { "epoch": 4.36, "learning_rate": 9.029292263827637e-06, "loss": 0.0751, "step": 4757 }, { "epoch": 4.36, "learning_rate": 9.025599020537116e-06, "loss": 0.0807, "step": 4758 }, { "epoch": 4.36, "learning_rate": 9.021905911424476e-06, "loss": 0.0744, "step": 4759 }, { "epoch": 4.36, "learning_rate": 9.018212936998265e-06, "loss": 0.077, "step": 4760 }, { "epoch": 4.36, "learning_rate": 9.014520097767016e-06, "loss": 0.0679, "step": 4761 }, { "epoch": 4.36, "learning_rate": 9.01082739423925e-06, "loss": 0.0831, "step": 4762 }, { "epoch": 4.37, "learning_rate": 9.00713482692346e-06, "loss": 0.0704, "step": 4763 }, { "epoch": 4.37, "learning_rate": 9.00344239632812e-06, "loss": 0.073, "step": 4764 }, { "epoch": 4.37, "learning_rate": 8.999750102961694e-06, "loss": 0.0809, "step": 4765 }, { "epoch": 4.37, "learning_rate": 8.996057947332623e-06, "loss": 0.0718, "step": 4766 }, { "epoch": 4.37, "learning_rate": 8.992365929949319e-06, "loss": 0.0656, "step": 4767 }, { "epoch": 4.37, "learning_rate": 8.98867405132019e-06, "loss": 0.0715, "step": 4768 }, { "epoch": 4.37, "learning_rate": 8.984982311953619e-06, "loss": 0.0653, "step": 4769 }, { "epoch": 4.37, "learning_rate": 8.981290712357964e-06, "loss": 0.0711, "step": 4770 }, { "epoch": 4.37, "learning_rate": 8.977599253041575e-06, "loss": 0.0806, "step": 4771 }, { "epoch": 4.37, "learning_rate": 8.973907934512775e-06, "loss": 0.075, "step": 4772 }, { "epoch": 4.37, "learning_rate": 8.970216757279869e-06, "loss": 0.0885, "step": 4773 }, { "epoch": 4.38, "learning_rate": 8.966525721851142e-06, "loss": 0.0767, "step": 4774 }, { "epoch": 4.38, "learning_rate": 8.962834828734858e-06, "loss": 0.0732, "step": 4775 }, { "epoch": 4.38, "learning_rate": 8.959144078439272e-06, "loss": 0.0756, "step": 4776 }, { "epoch": 4.38, "learning_rate": 8.955453471472607e-06, "loss": 0.0681, "step": 4777 }, { "epoch": 4.38, "learning_rate": 8.95176300834307e-06, "loss": 0.0809, "step": 4778 }, { "epoch": 4.38, "learning_rate": 8.948072689558843e-06, "loss": 0.0735, "step": 4779 }, { "epoch": 4.38, "learning_rate": 8.944382515628104e-06, "loss": 0.0742, "step": 4780 }, { "epoch": 4.38, "learning_rate": 8.940692487058995e-06, "loss": 0.0728, "step": 4781 }, { "epoch": 4.38, "learning_rate": 8.93700260435964e-06, "loss": 0.0793, "step": 4782 }, { "epoch": 4.38, "learning_rate": 8.933312868038157e-06, "loss": 0.0757, "step": 4783 }, { "epoch": 4.38, "learning_rate": 8.929623278602627e-06, "loss": 0.0739, "step": 4784 }, { "epoch": 4.39, "learning_rate": 8.92593383656112e-06, "loss": 0.0699, "step": 4785 }, { "epoch": 4.39, "learning_rate": 8.922244542421676e-06, "loss": 0.0639, "step": 4786 }, { "epoch": 4.39, "learning_rate": 8.918555396692333e-06, "loss": 0.0705, "step": 4787 }, { "epoch": 4.39, "learning_rate": 8.914866399881088e-06, "loss": 0.0675, "step": 4788 }, { "epoch": 4.39, "learning_rate": 8.911177552495926e-06, "loss": 0.0714, "step": 4789 }, { "epoch": 4.39, "learning_rate": 8.907488855044821e-06, "loss": 0.0684, "step": 4790 }, { "epoch": 4.39, "learning_rate": 8.903800308035713e-06, "loss": 0.0867, "step": 4791 }, { "epoch": 4.39, "learning_rate": 8.900111911976523e-06, "loss": 0.0717, "step": 4792 }, { "epoch": 4.39, "learning_rate": 8.896423667375156e-06, "loss": 0.0765, "step": 4793 }, { "epoch": 4.39, "learning_rate": 8.892735574739495e-06, "loss": 0.0696, "step": 4794 }, { "epoch": 4.4, "learning_rate": 8.8890476345774e-06, "loss": 0.0687, "step": 4795 }, { "epoch": 4.4, "learning_rate": 8.885359847396711e-06, "loss": 0.0719, "step": 4796 }, { "epoch": 4.4, "learning_rate": 8.881672213705249e-06, "loss": 0.0661, "step": 4797 }, { "epoch": 4.4, "learning_rate": 8.877984734010813e-06, "loss": 0.0829, "step": 4798 }, { "epoch": 4.4, "learning_rate": 8.874297408821176e-06, "loss": 0.0706, "step": 4799 }, { "epoch": 4.4, "learning_rate": 8.870610238644091e-06, "loss": 0.0605, "step": 4800 }, { "epoch": 4.4, "learning_rate": 8.866923223987303e-06, "loss": 0.0683, "step": 4801 }, { "epoch": 4.4, "learning_rate": 8.863236365358516e-06, "loss": 0.0819, "step": 4802 }, { "epoch": 4.4, "learning_rate": 8.85954966326542e-06, "loss": 0.0705, "step": 4803 }, { "epoch": 4.4, "learning_rate": 8.855863118215692e-06, "loss": 0.0758, "step": 4804 }, { "epoch": 4.4, "learning_rate": 8.852176730716976e-06, "loss": 0.0623, "step": 4805 }, { "epoch": 4.41, "learning_rate": 8.848490501276898e-06, "loss": 0.0764, "step": 4806 }, { "epoch": 4.41, "learning_rate": 8.84480443040306e-06, "loss": 0.0635, "step": 4807 }, { "epoch": 4.41, "learning_rate": 8.841118518603052e-06, "loss": 0.0669, "step": 4808 }, { "epoch": 4.41, "learning_rate": 8.83743276638443e-06, "loss": 0.0765, "step": 4809 }, { "epoch": 4.41, "learning_rate": 8.833747174254737e-06, "loss": 0.08, "step": 4810 }, { "epoch": 4.41, "learning_rate": 8.830061742721479e-06, "loss": 0.079, "step": 4811 }, { "epoch": 4.41, "learning_rate": 8.826376472292165e-06, "loss": 0.0757, "step": 4812 }, { "epoch": 4.41, "learning_rate": 8.822691363474257e-06, "loss": 0.0746, "step": 4813 }, { "epoch": 4.41, "learning_rate": 8.81900641677521e-06, "loss": 0.0676, "step": 4814 }, { "epoch": 4.41, "learning_rate": 8.815321632702452e-06, "loss": 0.0754, "step": 4815 }, { "epoch": 4.41, "learning_rate": 8.811637011763389e-06, "loss": 0.0772, "step": 4816 }, { "epoch": 4.42, "learning_rate": 8.807952554465404e-06, "loss": 0.0624, "step": 4817 }, { "epoch": 4.42, "learning_rate": 8.80426826131585e-06, "loss": 0.0638, "step": 4818 }, { "epoch": 4.42, "learning_rate": 8.800584132822079e-06, "loss": 0.078, "step": 4819 }, { "epoch": 4.42, "learning_rate": 8.796900169491397e-06, "loss": 0.0626, "step": 4820 }, { "epoch": 4.42, "learning_rate": 8.793216371831095e-06, "loss": 0.0765, "step": 4821 }, { "epoch": 4.42, "learning_rate": 8.789532740348452e-06, "loss": 0.0741, "step": 4822 }, { "epoch": 4.42, "learning_rate": 8.785849275550708e-06, "loss": 0.0724, "step": 4823 }, { "epoch": 4.42, "learning_rate": 8.782165977945087e-06, "loss": 0.0729, "step": 4824 }, { "epoch": 4.42, "learning_rate": 8.778482848038787e-06, "loss": 0.0711, "step": 4825 }, { "epoch": 4.42, "learning_rate": 8.774799886338996e-06, "loss": 0.079, "step": 4826 }, { "epoch": 4.42, "learning_rate": 8.771117093352861e-06, "loss": 0.0738, "step": 4827 }, { "epoch": 4.43, "learning_rate": 8.767434469587512e-06, "loss": 0.0621, "step": 4828 }, { "epoch": 4.43, "learning_rate": 8.763752015550064e-06, "loss": 0.0744, "step": 4829 }, { "epoch": 4.43, "learning_rate": 8.760069731747597e-06, "loss": 0.0647, "step": 4830 }, { "epoch": 4.43, "learning_rate": 8.756387618687174e-06, "loss": 0.0714, "step": 4831 }, { "epoch": 4.43, "learning_rate": 8.752705676875826e-06, "loss": 0.0712, "step": 4832 }, { "epoch": 4.43, "learning_rate": 8.749023906820578e-06, "loss": 0.0628, "step": 4833 }, { "epoch": 4.43, "learning_rate": 8.745342309028418e-06, "loss": 0.0757, "step": 4834 }, { "epoch": 4.43, "learning_rate": 8.741660884006303e-06, "loss": 0.0725, "step": 4835 }, { "epoch": 4.43, "learning_rate": 8.737979632261188e-06, "loss": 0.0701, "step": 4836 }, { "epoch": 4.43, "learning_rate": 8.734298554299989e-06, "loss": 0.0775, "step": 4837 }, { "epoch": 4.43, "learning_rate": 8.730617650629599e-06, "loss": 0.068, "step": 4838 }, { "epoch": 4.44, "learning_rate": 8.72693692175689e-06, "loss": 0.0737, "step": 4839 }, { "epoch": 4.44, "learning_rate": 8.723256368188708e-06, "loss": 0.0681, "step": 4840 }, { "epoch": 4.44, "learning_rate": 8.719575990431882e-06, "loss": 0.0694, "step": 4841 }, { "epoch": 4.44, "learning_rate": 8.715895788993203e-06, "loss": 0.0755, "step": 4842 }, { "epoch": 4.44, "learning_rate": 8.712215764379446e-06, "loss": 0.0685, "step": 4843 }, { "epoch": 4.44, "learning_rate": 8.70853591709737e-06, "loss": 0.061, "step": 4844 }, { "epoch": 4.44, "learning_rate": 8.704856247653692e-06, "loss": 0.0716, "step": 4845 }, { "epoch": 4.44, "learning_rate": 8.701176756555114e-06, "loss": 0.0714, "step": 4846 }, { "epoch": 4.44, "learning_rate": 8.697497444308318e-06, "loss": 0.0795, "step": 4847 }, { "epoch": 4.44, "learning_rate": 8.693818311419954e-06, "loss": 0.0576, "step": 4848 }, { "epoch": 4.44, "learning_rate": 8.690139358396648e-06, "loss": 0.0682, "step": 4849 }, { "epoch": 4.45, "learning_rate": 8.686460585744999e-06, "loss": 0.0665, "step": 4850 }, { "epoch": 4.45, "learning_rate": 8.682781993971594e-06, "loss": 0.0681, "step": 4851 }, { "epoch": 4.45, "learning_rate": 8.67910358358298e-06, "loss": 0.0791, "step": 4852 }, { "epoch": 4.45, "learning_rate": 8.675425355085681e-06, "loss": 0.0712, "step": 4853 }, { "epoch": 4.45, "learning_rate": 8.671747308986209e-06, "loss": 0.0674, "step": 4854 }, { "epoch": 4.45, "learning_rate": 8.668069445791039e-06, "loss": 0.074, "step": 4855 }, { "epoch": 4.45, "learning_rate": 8.664391766006617e-06, "loss": 0.0707, "step": 4856 }, { "epoch": 4.45, "learning_rate": 8.660714270139378e-06, "loss": 0.0648, "step": 4857 }, { "epoch": 4.45, "learning_rate": 8.657036958695722e-06, "loss": 0.0719, "step": 4858 }, { "epoch": 4.45, "learning_rate": 8.653359832182024e-06, "loss": 0.0825, "step": 4859 }, { "epoch": 4.45, "learning_rate": 8.649682891104634e-06, "loss": 0.0694, "step": 4860 }, { "epoch": 4.46, "learning_rate": 8.64600613596988e-06, "loss": 0.0658, "step": 4861 }, { "epoch": 4.46, "learning_rate": 8.642329567284065e-06, "loss": 0.0662, "step": 4862 }, { "epoch": 4.46, "learning_rate": 8.638653185553459e-06, "loss": 0.0713, "step": 4863 }, { "epoch": 4.46, "learning_rate": 8.634976991284307e-06, "loss": 0.0786, "step": 4864 }, { "epoch": 4.46, "learning_rate": 8.631300984982842e-06, "loss": 0.076, "step": 4865 }, { "epoch": 4.46, "learning_rate": 8.627625167155254e-06, "loss": 0.0723, "step": 4866 }, { "epoch": 4.46, "learning_rate": 8.623949538307715e-06, "loss": 0.062, "step": 4867 }, { "epoch": 4.46, "learning_rate": 8.620274098946365e-06, "loss": 0.0607, "step": 4868 }, { "epoch": 4.46, "learning_rate": 8.616598849577334e-06, "loss": 0.0806, "step": 4869 }, { "epoch": 4.46, "learning_rate": 8.612923790706706e-06, "loss": 0.0749, "step": 4870 }, { "epoch": 4.46, "learning_rate": 8.609248922840548e-06, "loss": 0.0625, "step": 4871 }, { "epoch": 4.47, "learning_rate": 8.605574246484904e-06, "loss": 0.092, "step": 4872 }, { "epoch": 4.47, "learning_rate": 8.601899762145787e-06, "loss": 0.0762, "step": 4873 }, { "epoch": 4.47, "learning_rate": 8.598225470329181e-06, "loss": 0.068, "step": 4874 }, { "epoch": 4.47, "learning_rate": 8.594551371541046e-06, "loss": 0.0788, "step": 4875 }, { "epoch": 4.47, "learning_rate": 8.590877466287322e-06, "loss": 0.0658, "step": 4876 }, { "epoch": 4.47, "learning_rate": 8.587203755073917e-06, "loss": 0.0704, "step": 4877 }, { "epoch": 4.47, "learning_rate": 8.5835302384067e-06, "loss": 0.0659, "step": 4878 }, { "epoch": 4.47, "learning_rate": 8.579856916791539e-06, "loss": 0.0676, "step": 4879 }, { "epoch": 4.47, "learning_rate": 8.576183790734253e-06, "loss": 0.0721, "step": 4880 }, { "epoch": 4.47, "learning_rate": 8.572510860740648e-06, "loss": 0.0796, "step": 4881 }, { "epoch": 4.47, "learning_rate": 8.568838127316491e-06, "loss": 0.0732, "step": 4882 }, { "epoch": 4.48, "learning_rate": 8.565165590967532e-06, "loss": 0.076, "step": 4883 }, { "epoch": 4.48, "learning_rate": 8.561493252199491e-06, "loss": 0.0711, "step": 4884 }, { "epoch": 4.48, "learning_rate": 8.557821111518056e-06, "loss": 0.0752, "step": 4885 }, { "epoch": 4.48, "learning_rate": 8.554149169428894e-06, "loss": 0.069, "step": 4886 }, { "epoch": 4.48, "learning_rate": 8.550477426437642e-06, "loss": 0.0708, "step": 4887 }, { "epoch": 4.48, "learning_rate": 8.546805883049913e-06, "loss": 0.0785, "step": 4888 }, { "epoch": 4.48, "learning_rate": 8.543134539771279e-06, "loss": 0.0749, "step": 4889 }, { "epoch": 4.48, "learning_rate": 8.539463397107308e-06, "loss": 0.0774, "step": 4890 }, { "epoch": 4.48, "learning_rate": 8.53579245556352e-06, "loss": 0.0637, "step": 4891 }, { "epoch": 4.48, "learning_rate": 8.532121715645411e-06, "loss": 0.0782, "step": 4892 }, { "epoch": 4.48, "learning_rate": 8.528451177858463e-06, "loss": 0.0791, "step": 4893 }, { "epoch": 4.49, "learning_rate": 8.524780842708111e-06, "loss": 0.068, "step": 4894 }, { "epoch": 4.49, "learning_rate": 8.521110710699775e-06, "loss": 0.073, "step": 4895 }, { "epoch": 4.49, "learning_rate": 8.51744078233884e-06, "loss": 0.0694, "step": 4896 }, { "epoch": 4.49, "learning_rate": 8.513771058130669e-06, "loss": 0.0806, "step": 4897 }, { "epoch": 4.49, "learning_rate": 8.510101538580593e-06, "loss": 0.0787, "step": 4898 }, { "epoch": 4.49, "learning_rate": 8.506432224193915e-06, "loss": 0.0684, "step": 4899 }, { "epoch": 4.49, "learning_rate": 8.502763115475908e-06, "loss": 0.0729, "step": 4900 }, { "epoch": 4.49, "learning_rate": 8.499094212931825e-06, "loss": 0.0626, "step": 4901 }, { "epoch": 4.49, "learning_rate": 8.495425517066878e-06, "loss": 0.0739, "step": 4902 }, { "epoch": 4.49, "learning_rate": 8.491757028386262e-06, "loss": 0.0758, "step": 4903 }, { "epoch": 4.49, "learning_rate": 8.488088747395137e-06, "loss": 0.0806, "step": 4904 }, { "epoch": 4.5, "learning_rate": 8.484420674598638e-06, "loss": 0.0716, "step": 4905 }, { "epoch": 4.5, "learning_rate": 8.480752810501867e-06, "loss": 0.08, "step": 4906 }, { "epoch": 4.5, "learning_rate": 8.477085155609898e-06, "loss": 0.073, "step": 4907 }, { "epoch": 4.5, "learning_rate": 8.473417710427782e-06, "loss": 0.0609, "step": 4908 }, { "epoch": 4.5, "learning_rate": 8.469750475460537e-06, "loss": 0.0807, "step": 4909 }, { "epoch": 4.5, "learning_rate": 8.466083451213145e-06, "loss": 0.0666, "step": 4910 }, { "epoch": 4.5, "learning_rate": 8.462416638190577e-06, "loss": 0.0726, "step": 4911 }, { "epoch": 4.5, "learning_rate": 8.458750036897756e-06, "loss": 0.0713, "step": 4912 }, { "epoch": 4.5, "learning_rate": 8.455083647839587e-06, "loss": 0.0768, "step": 4913 }, { "epoch": 4.5, "learning_rate": 8.451417471520938e-06, "loss": 0.0764, "step": 4914 }, { "epoch": 4.51, "learning_rate": 8.447751508446658e-06, "loss": 0.0651, "step": 4915 }, { "epoch": 4.51, "learning_rate": 8.44408575912156e-06, "loss": 0.0823, "step": 4916 }, { "epoch": 4.51, "learning_rate": 8.440420224050422e-06, "loss": 0.0707, "step": 4917 }, { "epoch": 4.51, "learning_rate": 8.43675490373801e-06, "loss": 0.0714, "step": 4918 }, { "epoch": 4.51, "learning_rate": 8.433089798689042e-06, "loss": 0.072, "step": 4919 }, { "epoch": 4.51, "learning_rate": 8.429424909408215e-06, "loss": 0.0831, "step": 4920 }, { "epoch": 4.51, "learning_rate": 8.425760236400191e-06, "loss": 0.0623, "step": 4921 }, { "epoch": 4.51, "learning_rate": 8.422095780169615e-06, "loss": 0.0694, "step": 4922 }, { "epoch": 4.51, "learning_rate": 8.418431541221087e-06, "loss": 0.0727, "step": 4923 }, { "epoch": 4.51, "learning_rate": 8.414767520059184e-06, "loss": 0.0746, "step": 4924 }, { "epoch": 4.51, "learning_rate": 8.411103717188458e-06, "loss": 0.0692, "step": 4925 }, { "epoch": 4.52, "learning_rate": 8.407440133113418e-06, "loss": 0.0699, "step": 4926 }, { "epoch": 4.52, "learning_rate": 8.403776768338555e-06, "loss": 0.0648, "step": 4927 }, { "epoch": 4.52, "learning_rate": 8.400113623368322e-06, "loss": 0.0729, "step": 4928 }, { "epoch": 4.52, "learning_rate": 8.396450698707148e-06, "loss": 0.0724, "step": 4929 }, { "epoch": 4.52, "learning_rate": 8.392787994859427e-06, "loss": 0.0568, "step": 4930 }, { "epoch": 4.52, "learning_rate": 8.389125512329524e-06, "loss": 0.0658, "step": 4931 }, { "epoch": 4.52, "learning_rate": 8.38546325162177e-06, "loss": 0.067, "step": 4932 }, { "epoch": 4.52, "learning_rate": 8.381801213240478e-06, "loss": 0.0749, "step": 4933 }, { "epoch": 4.52, "learning_rate": 8.378139397689914e-06, "loss": 0.0652, "step": 4934 }, { "epoch": 4.52, "learning_rate": 8.374477805474322e-06, "loss": 0.0705, "step": 4935 }, { "epoch": 4.52, "learning_rate": 8.370816437097917e-06, "loss": 0.0661, "step": 4936 }, { "epoch": 4.53, "learning_rate": 8.367155293064877e-06, "loss": 0.0731, "step": 4937 }, { "epoch": 4.53, "learning_rate": 8.363494373879354e-06, "loss": 0.0856, "step": 4938 }, { "epoch": 4.53, "learning_rate": 8.359833680045463e-06, "loss": 0.0698, "step": 4939 }, { "epoch": 4.53, "learning_rate": 8.356173212067303e-06, "loss": 0.0574, "step": 4940 }, { "epoch": 4.53, "learning_rate": 8.35251297044892e-06, "loss": 0.0741, "step": 4941 }, { "epoch": 4.53, "learning_rate": 8.348852955694343e-06, "loss": 0.0817, "step": 4942 }, { "epoch": 4.53, "learning_rate": 8.345193168307571e-06, "loss": 0.0728, "step": 4943 }, { "epoch": 4.53, "learning_rate": 8.341533608792564e-06, "loss": 0.0637, "step": 4944 }, { "epoch": 4.53, "learning_rate": 8.337874277653254e-06, "loss": 0.0698, "step": 4945 }, { "epoch": 4.53, "learning_rate": 8.33421517539354e-06, "loss": 0.0758, "step": 4946 }, { "epoch": 4.53, "learning_rate": 8.330556302517298e-06, "loss": 0.0643, "step": 4947 }, { "epoch": 4.54, "learning_rate": 8.326897659528359e-06, "loss": 0.0702, "step": 4948 }, { "epoch": 4.54, "learning_rate": 8.32323924693053e-06, "loss": 0.0594, "step": 4949 }, { "epoch": 4.54, "learning_rate": 8.319581065227585e-06, "loss": 0.0783, "step": 4950 }, { "epoch": 4.54, "learning_rate": 8.315923114923271e-06, "loss": 0.0747, "step": 4951 }, { "epoch": 4.54, "learning_rate": 8.312265396521296e-06, "loss": 0.0655, "step": 4952 }, { "epoch": 4.54, "learning_rate": 8.30860791052533e-06, "loss": 0.0675, "step": 4953 }, { "epoch": 4.54, "learning_rate": 8.304950657439034e-06, "loss": 0.0693, "step": 4954 }, { "epoch": 4.54, "learning_rate": 8.301293637766014e-06, "loss": 0.0692, "step": 4955 }, { "epoch": 4.54, "learning_rate": 8.297636852009855e-06, "loss": 0.0694, "step": 4956 }, { "epoch": 4.54, "learning_rate": 8.293980300674103e-06, "loss": 0.0839, "step": 4957 }, { "epoch": 4.54, "learning_rate": 8.290323984262282e-06, "loss": 0.0664, "step": 4958 }, { "epoch": 4.55, "learning_rate": 8.286667903277874e-06, "loss": 0.072, "step": 4959 }, { "epoch": 4.55, "learning_rate": 8.28301205822433e-06, "loss": 0.0866, "step": 4960 }, { "epoch": 4.55, "learning_rate": 8.279356449605075e-06, "loss": 0.0647, "step": 4961 }, { "epoch": 4.55, "learning_rate": 8.275701077923498e-06, "loss": 0.0751, "step": 4962 }, { "epoch": 4.55, "learning_rate": 8.27204594368295e-06, "loss": 0.0737, "step": 4963 }, { "epoch": 4.55, "learning_rate": 8.26839104738675e-06, "loss": 0.0631, "step": 4964 }, { "epoch": 4.55, "learning_rate": 8.2647363895382e-06, "loss": 0.0644, "step": 4965 }, { "epoch": 4.55, "learning_rate": 8.26108197064055e-06, "loss": 0.0746, "step": 4966 }, { "epoch": 4.55, "learning_rate": 8.25742779119702e-06, "loss": 0.0851, "step": 4967 }, { "epoch": 4.55, "learning_rate": 8.253773851710812e-06, "loss": 0.0666, "step": 4968 }, { "epoch": 4.55, "learning_rate": 8.250120152685074e-06, "loss": 0.0652, "step": 4969 }, { "epoch": 4.56, "learning_rate": 8.246466694622938e-06, "loss": 0.0711, "step": 4970 }, { "epoch": 4.56, "learning_rate": 8.242813478027491e-06, "loss": 0.0688, "step": 4971 }, { "epoch": 4.56, "learning_rate": 8.239160503401794e-06, "loss": 0.0691, "step": 4972 }, { "epoch": 4.56, "learning_rate": 8.235507771248874e-06, "loss": 0.0778, "step": 4973 }, { "epoch": 4.56, "learning_rate": 8.231855282071718e-06, "loss": 0.0809, "step": 4974 }, { "epoch": 4.56, "learning_rate": 8.22820303637329e-06, "loss": 0.0725, "step": 4975 }, { "epoch": 4.56, "learning_rate": 8.224551034656514e-06, "loss": 0.0612, "step": 4976 }, { "epoch": 4.56, "learning_rate": 8.220899277424282e-06, "loss": 0.0721, "step": 4977 }, { "epoch": 4.56, "learning_rate": 8.217247765179442e-06, "loss": 0.0774, "step": 4978 }, { "epoch": 4.56, "learning_rate": 8.213596498424831e-06, "loss": 0.0724, "step": 4979 }, { "epoch": 4.56, "learning_rate": 8.209945477663237e-06, "loss": 0.0646, "step": 4980 }, { "epoch": 4.57, "learning_rate": 8.206294703397405e-06, "loss": 0.0796, "step": 4981 }, { "epoch": 4.57, "learning_rate": 8.202644176130071e-06, "loss": 0.0753, "step": 4982 }, { "epoch": 4.57, "learning_rate": 8.198993896363919e-06, "loss": 0.0674, "step": 4983 }, { "epoch": 4.57, "learning_rate": 8.1953438646016e-06, "loss": 0.0739, "step": 4984 }, { "epoch": 4.57, "learning_rate": 8.191694081345734e-06, "loss": 0.0791, "step": 4985 }, { "epoch": 4.57, "learning_rate": 8.188044547098912e-06, "loss": 0.0657, "step": 4986 }, { "epoch": 4.57, "learning_rate": 8.184395262363682e-06, "loss": 0.0705, "step": 4987 }, { "epoch": 4.57, "learning_rate": 8.180746227642561e-06, "loss": 0.0729, "step": 4988 }, { "epoch": 4.57, "learning_rate": 8.177097443438033e-06, "loss": 0.079, "step": 4989 }, { "epoch": 4.57, "learning_rate": 8.173448910252549e-06, "loss": 0.0705, "step": 4990 }, { "epoch": 4.57, "learning_rate": 8.169800628588516e-06, "loss": 0.0628, "step": 4991 }, { "epoch": 4.58, "learning_rate": 8.166152598948317e-06, "loss": 0.0698, "step": 4992 }, { "epoch": 4.58, "learning_rate": 8.162504821834296e-06, "loss": 0.0647, "step": 4993 }, { "epoch": 4.58, "learning_rate": 8.158857297748767e-06, "loss": 0.0711, "step": 4994 }, { "epoch": 4.58, "learning_rate": 8.155210027194e-06, "loss": 0.0748, "step": 4995 }, { "epoch": 4.58, "learning_rate": 8.15156301067223e-06, "loss": 0.0761, "step": 4996 }, { "epoch": 4.58, "learning_rate": 8.147916248685672e-06, "loss": 0.0673, "step": 4997 }, { "epoch": 4.58, "learning_rate": 8.144269741736492e-06, "loss": 0.0803, "step": 4998 }, { "epoch": 4.58, "learning_rate": 8.14062349032682e-06, "loss": 0.0763, "step": 4999 }, { "epoch": 4.58, "learning_rate": 8.136977494958765e-06, "loss": 0.0685, "step": 5000 }, { "epoch": 4.58, "learning_rate": 8.133331756134384e-06, "loss": 0.082, "step": 5001 }, { "epoch": 4.58, "learning_rate": 8.129686274355707e-06, "loss": 0.0693, "step": 5002 }, { "epoch": 4.59, "learning_rate": 8.126041050124727e-06, "loss": 0.0693, "step": 5003 }, { "epoch": 4.59, "learning_rate": 8.122396083943407e-06, "loss": 0.0786, "step": 5004 }, { "epoch": 4.59, "learning_rate": 8.118751376313666e-06, "loss": 0.075, "step": 5005 }, { "epoch": 4.59, "learning_rate": 8.115106927737385e-06, "loss": 0.0628, "step": 5006 }, { "epoch": 4.59, "learning_rate": 8.111462738716427e-06, "loss": 0.0786, "step": 5007 }, { "epoch": 4.59, "learning_rate": 8.107818809752603e-06, "loss": 0.0761, "step": 5008 }, { "epoch": 4.59, "learning_rate": 8.10417514134769e-06, "loss": 0.0655, "step": 5009 }, { "epoch": 4.59, "learning_rate": 8.100531734003429e-06, "loss": 0.0713, "step": 5010 }, { "epoch": 4.59, "learning_rate": 8.096888588221536e-06, "loss": 0.0713, "step": 5011 }, { "epoch": 4.59, "learning_rate": 8.093245704503678e-06, "loss": 0.0735, "step": 5012 }, { "epoch": 4.59, "learning_rate": 8.089603083351494e-06, "loss": 0.0793, "step": 5013 }, { "epoch": 4.6, "learning_rate": 8.085960725266581e-06, "loss": 0.0742, "step": 5014 }, { "epoch": 4.6, "learning_rate": 8.082318630750502e-06, "loss": 0.0628, "step": 5015 }, { "epoch": 4.6, "learning_rate": 8.078676800304788e-06, "loss": 0.068, "step": 5016 }, { "epoch": 4.6, "learning_rate": 8.07503523443092e-06, "loss": 0.0612, "step": 5017 }, { "epoch": 4.6, "learning_rate": 8.071393933630366e-06, "loss": 0.0727, "step": 5018 }, { "epoch": 4.6, "learning_rate": 8.067752898404538e-06, "loss": 0.0729, "step": 5019 }, { "epoch": 4.6, "learning_rate": 8.064112129254814e-06, "loss": 0.0691, "step": 5020 }, { "epoch": 4.6, "learning_rate": 8.060471626682538e-06, "loss": 0.0751, "step": 5021 }, { "epoch": 4.6, "learning_rate": 8.056831391189024e-06, "loss": 0.065, "step": 5022 }, { "epoch": 4.6, "learning_rate": 8.053191423275542e-06, "loss": 0.0745, "step": 5023 }, { "epoch": 4.6, "learning_rate": 8.049551723443318e-06, "loss": 0.0794, "step": 5024 }, { "epoch": 4.61, "learning_rate": 8.045912292193561e-06, "loss": 0.0655, "step": 5025 }, { "epoch": 4.61, "learning_rate": 8.042273130027427e-06, "loss": 0.0685, "step": 5026 }, { "epoch": 4.61, "learning_rate": 8.038634237446037e-06, "loss": 0.0675, "step": 5027 }, { "epoch": 4.61, "learning_rate": 8.034995614950473e-06, "loss": 0.061, "step": 5028 }, { "epoch": 4.61, "learning_rate": 8.031357263041796e-06, "loss": 0.0771, "step": 5029 }, { "epoch": 4.61, "learning_rate": 8.02771918222101e-06, "loss": 0.0645, "step": 5030 }, { "epoch": 4.61, "learning_rate": 8.024081372989088e-06, "loss": 0.0757, "step": 5031 }, { "epoch": 4.61, "learning_rate": 8.020443835846972e-06, "loss": 0.0803, "step": 5032 }, { "epoch": 4.61, "learning_rate": 8.016806571295561e-06, "loss": 0.0589, "step": 5033 }, { "epoch": 4.61, "learning_rate": 8.013169579835713e-06, "loss": 0.0817, "step": 5034 }, { "epoch": 4.62, "learning_rate": 8.009532861968255e-06, "loss": 0.0702, "step": 5035 }, { "epoch": 4.62, "learning_rate": 8.005896418193972e-06, "loss": 0.0798, "step": 5036 }, { "epoch": 4.62, "learning_rate": 8.00226024901362e-06, "loss": 0.0686, "step": 5037 }, { "epoch": 4.62, "learning_rate": 7.9986243549279e-06, "loss": 0.0719, "step": 5038 }, { "epoch": 4.62, "learning_rate": 7.99498873643749e-06, "loss": 0.0712, "step": 5039 }, { "epoch": 4.62, "learning_rate": 7.991353394043028e-06, "loss": 0.0727, "step": 5040 }, { "epoch": 4.62, "learning_rate": 7.987718328245109e-06, "loss": 0.0657, "step": 5041 }, { "epoch": 4.62, "learning_rate": 7.984083539544287e-06, "loss": 0.0792, "step": 5042 }, { "epoch": 4.62, "learning_rate": 7.980449028441093e-06, "loss": 0.0554, "step": 5043 }, { "epoch": 4.62, "learning_rate": 7.976814795436004e-06, "loss": 0.0676, "step": 5044 }, { "epoch": 4.62, "learning_rate": 7.973180841029466e-06, "loss": 0.0699, "step": 5045 }, { "epoch": 4.63, "learning_rate": 7.96954716572188e-06, "loss": 0.0757, "step": 5046 }, { "epoch": 4.63, "learning_rate": 7.965913770013625e-06, "loss": 0.0705, "step": 5047 }, { "epoch": 4.63, "learning_rate": 7.962280654405021e-06, "loss": 0.0637, "step": 5048 }, { "epoch": 4.63, "learning_rate": 7.958647819396358e-06, "loss": 0.0789, "step": 5049 }, { "epoch": 4.63, "learning_rate": 7.955015265487895e-06, "loss": 0.0777, "step": 5050 }, { "epoch": 4.63, "learning_rate": 7.951382993179843e-06, "loss": 0.0664, "step": 5051 }, { "epoch": 4.63, "learning_rate": 7.947751002972375e-06, "loss": 0.0694, "step": 5052 }, { "epoch": 4.63, "learning_rate": 7.944119295365624e-06, "loss": 0.0609, "step": 5053 }, { "epoch": 4.63, "learning_rate": 7.940487870859693e-06, "loss": 0.0678, "step": 5054 }, { "epoch": 4.63, "learning_rate": 7.936856729954635e-06, "loss": 0.0741, "step": 5055 }, { "epoch": 4.63, "learning_rate": 7.93322587315047e-06, "loss": 0.0697, "step": 5056 }, { "epoch": 4.64, "learning_rate": 7.929595300947182e-06, "loss": 0.0712, "step": 5057 }, { "epoch": 4.64, "learning_rate": 7.925965013844704e-06, "loss": 0.0622, "step": 5058 }, { "epoch": 4.64, "learning_rate": 7.922335012342943e-06, "loss": 0.0718, "step": 5059 }, { "epoch": 4.64, "learning_rate": 7.918705296941758e-06, "loss": 0.0684, "step": 5060 }, { "epoch": 4.64, "learning_rate": 7.91507586814097e-06, "loss": 0.0794, "step": 5061 }, { "epoch": 4.64, "learning_rate": 7.911446726440368e-06, "loss": 0.0671, "step": 5062 }, { "epoch": 4.64, "learning_rate": 7.907817872339687e-06, "loss": 0.0796, "step": 5063 }, { "epoch": 4.64, "learning_rate": 7.904189306338642e-06, "loss": 0.0653, "step": 5064 }, { "epoch": 4.64, "learning_rate": 7.900561028936887e-06, "loss": 0.0716, "step": 5065 }, { "epoch": 4.64, "learning_rate": 7.896933040634053e-06, "loss": 0.0694, "step": 5066 }, { "epoch": 4.64, "learning_rate": 7.893305341929716e-06, "loss": 0.0667, "step": 5067 }, { "epoch": 4.65, "learning_rate": 7.889677933323431e-06, "loss": 0.0705, "step": 5068 }, { "epoch": 4.65, "learning_rate": 7.886050815314701e-06, "loss": 0.0691, "step": 5069 }, { "epoch": 4.65, "learning_rate": 7.882423988402982e-06, "loss": 0.0765, "step": 5070 }, { "epoch": 4.65, "learning_rate": 7.878797453087712e-06, "loss": 0.0648, "step": 5071 }, { "epoch": 4.65, "learning_rate": 7.875171209868265e-06, "loss": 0.0849, "step": 5072 }, { "epoch": 4.65, "learning_rate": 7.871545259243992e-06, "loss": 0.0711, "step": 5073 }, { "epoch": 4.65, "learning_rate": 7.867919601714188e-06, "loss": 0.0783, "step": 5074 }, { "epoch": 4.65, "learning_rate": 7.864294237778127e-06, "loss": 0.0633, "step": 5075 }, { "epoch": 4.65, "learning_rate": 7.860669167935028e-06, "loss": 0.0748, "step": 5076 }, { "epoch": 4.65, "learning_rate": 7.857044392684073e-06, "loss": 0.074, "step": 5077 }, { "epoch": 4.65, "learning_rate": 7.853419912524405e-06, "loss": 0.0598, "step": 5078 }, { "epoch": 4.66, "learning_rate": 7.849795727955125e-06, "loss": 0.075, "step": 5079 }, { "epoch": 4.66, "learning_rate": 7.846171839475296e-06, "loss": 0.0671, "step": 5080 }, { "epoch": 4.66, "learning_rate": 7.842548247583934e-06, "loss": 0.0782, "step": 5081 }, { "epoch": 4.66, "learning_rate": 7.838924952780023e-06, "loss": 0.0574, "step": 5082 }, { "epoch": 4.66, "learning_rate": 7.8353019555625e-06, "loss": 0.0662, "step": 5083 }, { "epoch": 4.66, "learning_rate": 7.831679256430262e-06, "loss": 0.0611, "step": 5084 }, { "epoch": 4.66, "learning_rate": 7.82805685588216e-06, "loss": 0.0723, "step": 5085 }, { "epoch": 4.66, "learning_rate": 7.824434754417019e-06, "loss": 0.0702, "step": 5086 }, { "epoch": 4.66, "learning_rate": 7.820812952533607e-06, "loss": 0.0622, "step": 5087 }, { "epoch": 4.66, "learning_rate": 7.817191450730655e-06, "loss": 0.0755, "step": 5088 }, { "epoch": 4.66, "learning_rate": 7.81357024950686e-06, "loss": 0.0752, "step": 5089 }, { "epoch": 4.67, "learning_rate": 7.809949349360872e-06, "loss": 0.0664, "step": 5090 }, { "epoch": 4.67, "learning_rate": 7.806328750791294e-06, "loss": 0.078, "step": 5091 }, { "epoch": 4.67, "learning_rate": 7.802708454296693e-06, "loss": 0.0632, "step": 5092 }, { "epoch": 4.67, "learning_rate": 7.799088460375604e-06, "loss": 0.0751, "step": 5093 }, { "epoch": 4.67, "learning_rate": 7.795468769526501e-06, "loss": 0.0608, "step": 5094 }, { "epoch": 4.67, "learning_rate": 7.791849382247825e-06, "loss": 0.0746, "step": 5095 }, { "epoch": 4.67, "learning_rate": 7.788230299037987e-06, "loss": 0.0727, "step": 5096 }, { "epoch": 4.67, "learning_rate": 7.784611520395338e-06, "loss": 0.0731, "step": 5097 }, { "epoch": 4.67, "learning_rate": 7.780993046818194e-06, "loss": 0.0664, "step": 5098 }, { "epoch": 4.67, "learning_rate": 7.777374878804831e-06, "loss": 0.0582, "step": 5099 }, { "epoch": 4.67, "learning_rate": 7.773757016853483e-06, "loss": 0.0639, "step": 5100 }, { "epoch": 4.68, "learning_rate": 7.770139461462335e-06, "loss": 0.0633, "step": 5101 }, { "epoch": 4.68, "learning_rate": 7.766522213129544e-06, "loss": 0.0659, "step": 5102 }, { "epoch": 4.68, "learning_rate": 7.762905272353205e-06, "loss": 0.0591, "step": 5103 }, { "epoch": 4.68, "learning_rate": 7.759288639631388e-06, "loss": 0.0651, "step": 5104 }, { "epoch": 4.68, "learning_rate": 7.755672315462115e-06, "loss": 0.0664, "step": 5105 }, { "epoch": 4.68, "learning_rate": 7.752056300343358e-06, "loss": 0.0709, "step": 5106 }, { "epoch": 4.68, "learning_rate": 7.74844059477306e-06, "loss": 0.07, "step": 5107 }, { "epoch": 4.68, "learning_rate": 7.744825199249113e-06, "loss": 0.0726, "step": 5108 }, { "epoch": 4.68, "learning_rate": 7.741210114269365e-06, "loss": 0.0632, "step": 5109 }, { "epoch": 4.68, "learning_rate": 7.73759534033162e-06, "loss": 0.0615, "step": 5110 }, { "epoch": 4.68, "learning_rate": 7.733980877933652e-06, "loss": 0.0664, "step": 5111 }, { "epoch": 4.69, "learning_rate": 7.73036672757318e-06, "loss": 0.0789, "step": 5112 }, { "epoch": 4.69, "learning_rate": 7.726752889747876e-06, "loss": 0.0707, "step": 5113 }, { "epoch": 4.69, "learning_rate": 7.723139364955388e-06, "loss": 0.0619, "step": 5114 }, { "epoch": 4.69, "learning_rate": 7.719526153693302e-06, "loss": 0.0686, "step": 5115 }, { "epoch": 4.69, "learning_rate": 7.715913256459168e-06, "loss": 0.0715, "step": 5116 }, { "epoch": 4.69, "learning_rate": 7.712300673750492e-06, "loss": 0.0711, "step": 5117 }, { "epoch": 4.69, "learning_rate": 7.70868840606474e-06, "loss": 0.0745, "step": 5118 }, { "epoch": 4.69, "learning_rate": 7.705076453899331e-06, "loss": 0.0574, "step": 5119 }, { "epoch": 4.69, "learning_rate": 7.701464817751639e-06, "loss": 0.0649, "step": 5120 }, { "epoch": 4.69, "learning_rate": 7.697853498119001e-06, "loss": 0.0705, "step": 5121 }, { "epoch": 4.69, "learning_rate": 7.694242495498706e-06, "loss": 0.0686, "step": 5122 }, { "epoch": 4.7, "learning_rate": 7.690631810387993e-06, "loss": 0.0691, "step": 5123 }, { "epoch": 4.7, "learning_rate": 7.687021443284071e-06, "loss": 0.0652, "step": 5124 }, { "epoch": 4.7, "learning_rate": 7.683411394684096e-06, "loss": 0.0734, "step": 5125 }, { "epoch": 4.7, "learning_rate": 7.679801665085183e-06, "loss": 0.065, "step": 5126 }, { "epoch": 4.7, "learning_rate": 7.676192254984401e-06, "loss": 0.0666, "step": 5127 }, { "epoch": 4.7, "learning_rate": 7.672583164878775e-06, "loss": 0.0651, "step": 5128 }, { "epoch": 4.7, "learning_rate": 7.668974395265293e-06, "loss": 0.0657, "step": 5129 }, { "epoch": 4.7, "learning_rate": 7.66536594664089e-06, "loss": 0.074, "step": 5130 }, { "epoch": 4.7, "learning_rate": 7.661757819502454e-06, "loss": 0.0683, "step": 5131 }, { "epoch": 4.7, "learning_rate": 7.658150014346846e-06, "loss": 0.0776, "step": 5132 }, { "epoch": 4.7, "learning_rate": 7.654542531670866e-06, "loss": 0.0706, "step": 5133 }, { "epoch": 4.71, "learning_rate": 7.650935371971274e-06, "loss": 0.0698, "step": 5134 }, { "epoch": 4.71, "learning_rate": 7.647328535744782e-06, "loss": 0.0709, "step": 5135 }, { "epoch": 4.71, "learning_rate": 7.643722023488072e-06, "loss": 0.0647, "step": 5136 }, { "epoch": 4.71, "learning_rate": 7.640115835697765e-06, "loss": 0.0624, "step": 5137 }, { "epoch": 4.71, "learning_rate": 7.636509972870442e-06, "loss": 0.0681, "step": 5138 }, { "epoch": 4.71, "learning_rate": 7.63290443550265e-06, "loss": 0.0681, "step": 5139 }, { "epoch": 4.71, "learning_rate": 7.629299224090874e-06, "loss": 0.0671, "step": 5140 }, { "epoch": 4.71, "learning_rate": 7.625694339131564e-06, "loss": 0.0694, "step": 5141 }, { "epoch": 4.71, "learning_rate": 7.62208978112112e-06, "loss": 0.0699, "step": 5142 }, { "epoch": 4.71, "learning_rate": 7.618485550555909e-06, "loss": 0.0642, "step": 5143 }, { "epoch": 4.71, "learning_rate": 7.614881647932235e-06, "loss": 0.0665, "step": 5144 }, { "epoch": 4.72, "learning_rate": 7.6112780737463705e-06, "loss": 0.0645, "step": 5145 }, { "epoch": 4.72, "learning_rate": 7.60767482849454e-06, "loss": 0.0761, "step": 5146 }, { "epoch": 4.72, "learning_rate": 7.604071912672916e-06, "loss": 0.0655, "step": 5147 }, { "epoch": 4.72, "learning_rate": 7.6004693267776375e-06, "loss": 0.0593, "step": 5148 }, { "epoch": 4.72, "learning_rate": 7.596867071304781e-06, "loss": 0.0716, "step": 5149 }, { "epoch": 4.72, "learning_rate": 7.593265146750397e-06, "loss": 0.0612, "step": 5150 }, { "epoch": 4.72, "learning_rate": 7.589663553610479e-06, "loss": 0.0781, "step": 5151 }, { "epoch": 4.72, "learning_rate": 7.586062292380971e-06, "loss": 0.0692, "step": 5152 }, { "epoch": 4.72, "learning_rate": 7.582461363557786e-06, "loss": 0.0745, "step": 5153 }, { "epoch": 4.72, "learning_rate": 7.578860767636779e-06, "loss": 0.0733, "step": 5154 }, { "epoch": 4.73, "learning_rate": 7.575260505113762e-06, "loss": 0.0759, "step": 5155 }, { "epoch": 4.73, "learning_rate": 7.571660576484496e-06, "loss": 0.0654, "step": 5156 }, { "epoch": 4.73, "learning_rate": 7.5680609822447135e-06, "loss": 0.0647, "step": 5157 }, { "epoch": 4.73, "learning_rate": 7.564461722890082e-06, "loss": 0.0841, "step": 5158 }, { "epoch": 4.73, "learning_rate": 7.560862798916229e-06, "loss": 0.0681, "step": 5159 }, { "epoch": 4.73, "learning_rate": 7.557264210818741e-06, "loss": 0.0607, "step": 5160 }, { "epoch": 4.73, "learning_rate": 7.553665959093154e-06, "loss": 0.0778, "step": 5161 }, { "epoch": 4.73, "learning_rate": 7.550068044234955e-06, "loss": 0.0693, "step": 5162 }, { "epoch": 4.73, "learning_rate": 7.546470466739586e-06, "loss": 0.0771, "step": 5163 }, { "epoch": 4.73, "learning_rate": 7.542873227102449e-06, "loss": 0.0636, "step": 5164 }, { "epoch": 4.73, "learning_rate": 7.539276325818894e-06, "loss": 0.0802, "step": 5165 }, { "epoch": 4.74, "learning_rate": 7.53567976338422e-06, "loss": 0.0626, "step": 5166 }, { "epoch": 4.74, "learning_rate": 7.532083540293688e-06, "loss": 0.0705, "step": 5167 }, { "epoch": 4.74, "learning_rate": 7.528487657042507e-06, "loss": 0.0695, "step": 5168 }, { "epoch": 4.74, "learning_rate": 7.524892114125843e-06, "loss": 0.0685, "step": 5169 }, { "epoch": 4.74, "learning_rate": 7.52129691203881e-06, "loss": 0.0692, "step": 5170 }, { "epoch": 4.74, "learning_rate": 7.517702051276478e-06, "loss": 0.0735, "step": 5171 }, { "epoch": 4.74, "learning_rate": 7.514107532333875e-06, "loss": 0.0675, "step": 5172 }, { "epoch": 4.74, "learning_rate": 7.510513355705973e-06, "loss": 0.0635, "step": 5173 }, { "epoch": 4.74, "learning_rate": 7.506919521887695e-06, "loss": 0.0758, "step": 5174 }, { "epoch": 4.74, "learning_rate": 7.503326031373932e-06, "loss": 0.0693, "step": 5175 }, { "epoch": 4.74, "learning_rate": 7.499732884659517e-06, "loss": 0.0748, "step": 5176 }, { "epoch": 4.75, "learning_rate": 7.496140082239227e-06, "loss": 0.0727, "step": 5177 }, { "epoch": 4.75, "learning_rate": 7.492547624607813e-06, "loss": 0.0737, "step": 5178 }, { "epoch": 4.75, "learning_rate": 7.488955512259963e-06, "loss": 0.075, "step": 5179 }, { "epoch": 4.75, "learning_rate": 7.485363745690321e-06, "loss": 0.0647, "step": 5180 }, { "epoch": 4.75, "learning_rate": 7.48177232539348e-06, "loss": 0.0707, "step": 5181 }, { "epoch": 4.75, "learning_rate": 7.478181251863999e-06, "loss": 0.0728, "step": 5182 }, { "epoch": 4.75, "learning_rate": 7.47459052559637e-06, "loss": 0.0811, "step": 5183 }, { "epoch": 4.75, "learning_rate": 7.471000147085048e-06, "loss": 0.0785, "step": 5184 }, { "epoch": 4.75, "learning_rate": 7.467410116824443e-06, "loss": 0.0671, "step": 5185 }, { "epoch": 4.75, "learning_rate": 7.4638204353089105e-06, "loss": 0.0796, "step": 5186 }, { "epoch": 4.75, "learning_rate": 7.460231103032759e-06, "loss": 0.0727, "step": 5187 }, { "epoch": 4.76, "learning_rate": 7.45664212049025e-06, "loss": 0.0674, "step": 5188 }, { "epoch": 4.76, "learning_rate": 7.4530534881756015e-06, "loss": 0.0785, "step": 5189 }, { "epoch": 4.76, "learning_rate": 7.449465206582973e-06, "loss": 0.0644, "step": 5190 }, { "epoch": 4.76, "learning_rate": 7.445877276206485e-06, "loss": 0.0673, "step": 5191 }, { "epoch": 4.76, "learning_rate": 7.4422896975402004e-06, "loss": 0.0673, "step": 5192 }, { "epoch": 4.76, "learning_rate": 7.438702471078148e-06, "loss": 0.0718, "step": 5193 }, { "epoch": 4.76, "learning_rate": 7.435115597314295e-06, "loss": 0.0749, "step": 5194 }, { "epoch": 4.76, "learning_rate": 7.431529076742561e-06, "loss": 0.069, "step": 5195 }, { "epoch": 4.76, "learning_rate": 7.427942909856828e-06, "loss": 0.071, "step": 5196 }, { "epoch": 4.76, "learning_rate": 7.424357097150917e-06, "loss": 0.0651, "step": 5197 }, { "epoch": 4.76, "learning_rate": 7.420771639118607e-06, "loss": 0.0559, "step": 5198 }, { "epoch": 4.77, "learning_rate": 7.41718653625362e-06, "loss": 0.061, "step": 5199 }, { "epoch": 4.77, "learning_rate": 7.413601789049644e-06, "loss": 0.071, "step": 5200 }, { "epoch": 4.77, "learning_rate": 7.410017398000304e-06, "loss": 0.0635, "step": 5201 }, { "epoch": 4.77, "learning_rate": 7.406433363599179e-06, "loss": 0.0802, "step": 5202 }, { "epoch": 4.77, "learning_rate": 7.402849686339811e-06, "loss": 0.0832, "step": 5203 }, { "epoch": 4.77, "learning_rate": 7.399266366715674e-06, "loss": 0.0766, "step": 5204 }, { "epoch": 4.77, "learning_rate": 7.395683405220204e-06, "loss": 0.0695, "step": 5205 }, { "epoch": 4.77, "learning_rate": 7.392100802346782e-06, "loss": 0.0689, "step": 5206 }, { "epoch": 4.77, "learning_rate": 7.38851855858875e-06, "loss": 0.0647, "step": 5207 }, { "epoch": 4.77, "learning_rate": 7.384936674439389e-06, "loss": 0.0865, "step": 5208 }, { "epoch": 4.77, "learning_rate": 7.3813551503919335e-06, "loss": 0.0657, "step": 5209 }, { "epoch": 4.78, "learning_rate": 7.377773986939574e-06, "loss": 0.0699, "step": 5210 }, { "epoch": 4.78, "learning_rate": 7.374193184575444e-06, "loss": 0.0628, "step": 5211 }, { "epoch": 4.78, "learning_rate": 7.370612743792635e-06, "loss": 0.0642, "step": 5212 }, { "epoch": 4.78, "learning_rate": 7.367032665084179e-06, "loss": 0.0644, "step": 5213 }, { "epoch": 4.78, "learning_rate": 7.363452948943065e-06, "loss": 0.0711, "step": 5214 }, { "epoch": 4.78, "learning_rate": 7.359873595862233e-06, "loss": 0.0734, "step": 5215 }, { "epoch": 4.78, "learning_rate": 7.3562946063345665e-06, "loss": 0.061, "step": 5216 }, { "epoch": 4.78, "learning_rate": 7.3527159808529044e-06, "loss": 0.0581, "step": 5217 }, { "epoch": 4.78, "learning_rate": 7.349137719910037e-06, "loss": 0.0676, "step": 5218 }, { "epoch": 4.78, "learning_rate": 7.345559823998698e-06, "loss": 0.0556, "step": 5219 }, { "epoch": 4.78, "learning_rate": 7.3419822936115716e-06, "loss": 0.0699, "step": 5220 }, { "epoch": 4.79, "learning_rate": 7.3384051292412996e-06, "loss": 0.0676, "step": 5221 }, { "epoch": 4.79, "learning_rate": 7.334828331380468e-06, "loss": 0.0702, "step": 5222 }, { "epoch": 4.79, "learning_rate": 7.33125190052161e-06, "loss": 0.0636, "step": 5223 }, { "epoch": 4.79, "learning_rate": 7.3276758371572064e-06, "loss": 0.0729, "step": 5224 }, { "epoch": 4.79, "learning_rate": 7.3241001417797e-06, "loss": 0.0638, "step": 5225 }, { "epoch": 4.79, "learning_rate": 7.320524814881471e-06, "loss": 0.0605, "step": 5226 }, { "epoch": 4.79, "learning_rate": 7.316949856954847e-06, "loss": 0.054, "step": 5227 }, { "epoch": 4.79, "learning_rate": 7.3133752684921225e-06, "loss": 0.0611, "step": 5228 }, { "epoch": 4.79, "learning_rate": 7.309801049985518e-06, "loss": 0.0587, "step": 5229 }, { "epoch": 4.79, "learning_rate": 7.3062272019272185e-06, "loss": 0.0648, "step": 5230 }, { "epoch": 4.79, "learning_rate": 7.302653724809351e-06, "loss": 0.0733, "step": 5231 }, { "epoch": 4.8, "learning_rate": 7.299080619123998e-06, "loss": 0.0671, "step": 5232 }, { "epoch": 4.8, "learning_rate": 7.295507885363182e-06, "loss": 0.0617, "step": 5233 }, { "epoch": 4.8, "learning_rate": 7.2919355240188805e-06, "loss": 0.0728, "step": 5234 }, { "epoch": 4.8, "learning_rate": 7.288363535583019e-06, "loss": 0.0665, "step": 5235 }, { "epoch": 4.8, "learning_rate": 7.284791920547472e-06, "loss": 0.0581, "step": 5236 }, { "epoch": 4.8, "learning_rate": 7.281220679404059e-06, "loss": 0.0684, "step": 5237 }, { "epoch": 4.8, "learning_rate": 7.277649812644549e-06, "loss": 0.0727, "step": 5238 }, { "epoch": 4.8, "learning_rate": 7.274079320760666e-06, "loss": 0.0635, "step": 5239 }, { "epoch": 4.8, "learning_rate": 7.270509204244074e-06, "loss": 0.0604, "step": 5240 }, { "epoch": 4.8, "learning_rate": 7.266939463586385e-06, "loss": 0.0563, "step": 5241 }, { "epoch": 4.8, "learning_rate": 7.263370099279173e-06, "loss": 0.0669, "step": 5242 }, { "epoch": 4.81, "learning_rate": 7.25980111181394e-06, "loss": 0.0694, "step": 5243 }, { "epoch": 4.81, "learning_rate": 7.256232501682151e-06, "loss": 0.076, "step": 5244 }, { "epoch": 4.81, "learning_rate": 7.252664269375209e-06, "loss": 0.063, "step": 5245 }, { "epoch": 4.81, "learning_rate": 7.2490964153844776e-06, "loss": 0.0755, "step": 5246 }, { "epoch": 4.81, "learning_rate": 7.245528940201257e-06, "loss": 0.0682, "step": 5247 }, { "epoch": 4.81, "learning_rate": 7.241961844316794e-06, "loss": 0.0632, "step": 5248 }, { "epoch": 4.81, "learning_rate": 7.238395128222299e-06, "loss": 0.0677, "step": 5249 }, { "epoch": 4.81, "learning_rate": 7.234828792408912e-06, "loss": 0.0695, "step": 5250 }, { "epoch": 4.81, "learning_rate": 7.2312628373677295e-06, "loss": 0.0725, "step": 5251 }, { "epoch": 4.81, "learning_rate": 7.22769726358979e-06, "loss": 0.069, "step": 5252 }, { "epoch": 4.81, "learning_rate": 7.2241320715660924e-06, "loss": 0.0681, "step": 5253 }, { "epoch": 4.82, "learning_rate": 7.220567261787567e-06, "loss": 0.0619, "step": 5254 }, { "epoch": 4.82, "learning_rate": 7.2170028347451014e-06, "loss": 0.0626, "step": 5255 }, { "epoch": 4.82, "learning_rate": 7.213438790929525e-06, "loss": 0.0731, "step": 5256 }, { "epoch": 4.82, "learning_rate": 7.209875130831621e-06, "loss": 0.0632, "step": 5257 }, { "epoch": 4.82, "learning_rate": 7.206311854942115e-06, "loss": 0.0745, "step": 5258 }, { "epoch": 4.82, "learning_rate": 7.202748963751678e-06, "loss": 0.0632, "step": 5259 }, { "epoch": 4.82, "learning_rate": 7.199186457750931e-06, "loss": 0.0774, "step": 5260 }, { "epoch": 4.82, "learning_rate": 7.195624337430445e-06, "loss": 0.0765, "step": 5261 }, { "epoch": 4.82, "learning_rate": 7.192062603280731e-06, "loss": 0.0685, "step": 5262 }, { "epoch": 4.82, "learning_rate": 7.188501255792251e-06, "loss": 0.0621, "step": 5263 }, { "epoch": 4.82, "learning_rate": 7.184940295455414e-06, "loss": 0.0586, "step": 5264 }, { "epoch": 4.83, "learning_rate": 7.181379722760575e-06, "loss": 0.0636, "step": 5265 }, { "epoch": 4.83, "learning_rate": 7.177819538198031e-06, "loss": 0.0699, "step": 5266 }, { "epoch": 4.83, "learning_rate": 7.174259742258036e-06, "loss": 0.0626, "step": 5267 }, { "epoch": 4.83, "learning_rate": 7.170700335430782e-06, "loss": 0.0616, "step": 5268 }, { "epoch": 4.83, "learning_rate": 7.167141318206412e-06, "loss": 0.068, "step": 5269 }, { "epoch": 4.83, "learning_rate": 7.163582691075004e-06, "loss": 0.0662, "step": 5270 }, { "epoch": 4.83, "learning_rate": 7.160024454526602e-06, "loss": 0.0635, "step": 5271 }, { "epoch": 4.83, "learning_rate": 7.156466609051182e-06, "loss": 0.0691, "step": 5272 }, { "epoch": 4.83, "learning_rate": 7.152909155138666e-06, "loss": 0.0739, "step": 5273 }, { "epoch": 4.83, "learning_rate": 7.149352093278934e-06, "loss": 0.0705, "step": 5274 }, { "epoch": 4.84, "learning_rate": 7.145795423961798e-06, "loss": 0.0654, "step": 5275 }, { "epoch": 4.84, "learning_rate": 7.142239147677021e-06, "loss": 0.0718, "step": 5276 }, { "epoch": 4.84, "learning_rate": 7.138683264914315e-06, "loss": 0.0651, "step": 5277 }, { "epoch": 4.84, "learning_rate": 7.135127776163336e-06, "loss": 0.064, "step": 5278 }, { "epoch": 4.84, "learning_rate": 7.131572681913684e-06, "loss": 0.0612, "step": 5279 }, { "epoch": 4.84, "learning_rate": 7.128017982654908e-06, "loss": 0.0741, "step": 5280 }, { "epoch": 4.84, "learning_rate": 7.124463678876494e-06, "loss": 0.0725, "step": 5281 }, { "epoch": 4.84, "learning_rate": 7.1209097710678895e-06, "loss": 0.0593, "step": 5282 }, { "epoch": 4.84, "learning_rate": 7.117356259718475e-06, "loss": 0.0724, "step": 5283 }, { "epoch": 4.84, "learning_rate": 7.113803145317572e-06, "loss": 0.07, "step": 5284 }, { "epoch": 4.84, "learning_rate": 7.110250428354466e-06, "loss": 0.0614, "step": 5285 }, { "epoch": 4.85, "learning_rate": 7.106698109318371e-06, "loss": 0.0694, "step": 5286 }, { "epoch": 4.85, "learning_rate": 7.103146188698453e-06, "loss": 0.0657, "step": 5287 }, { "epoch": 4.85, "learning_rate": 7.099594666983816e-06, "loss": 0.0698, "step": 5288 }, { "epoch": 4.85, "learning_rate": 7.096043544663525e-06, "loss": 0.0676, "step": 5289 }, { "epoch": 4.85, "learning_rate": 7.0924928222265736e-06, "loss": 0.0601, "step": 5290 }, { "epoch": 4.85, "learning_rate": 7.088942500161904e-06, "loss": 0.0635, "step": 5291 }, { "epoch": 4.85, "learning_rate": 7.085392578958415e-06, "loss": 0.0627, "step": 5292 }, { "epoch": 4.85, "learning_rate": 7.081843059104937e-06, "loss": 0.0704, "step": 5293 }, { "epoch": 4.85, "learning_rate": 7.078293941090248e-06, "loss": 0.073, "step": 5294 }, { "epoch": 4.85, "learning_rate": 7.074745225403068e-06, "loss": 0.0736, "step": 5295 }, { "epoch": 4.85, "learning_rate": 7.071196912532075e-06, "loss": 0.0647, "step": 5296 }, { "epoch": 4.86, "learning_rate": 7.067649002965877e-06, "loss": 0.0681, "step": 5297 }, { "epoch": 4.86, "learning_rate": 7.0641014971930275e-06, "loss": 0.0665, "step": 5298 }, { "epoch": 4.86, "learning_rate": 7.0605543957020375e-06, "loss": 0.0789, "step": 5299 }, { "epoch": 4.86, "learning_rate": 7.057007698981346e-06, "loss": 0.0667, "step": 5300 }, { "epoch": 4.86, "learning_rate": 7.0534614075193495e-06, "loss": 0.0583, "step": 5301 }, { "epoch": 4.86, "learning_rate": 7.049915521804377e-06, "loss": 0.0668, "step": 5302 }, { "epoch": 4.86, "learning_rate": 7.04637004232471e-06, "loss": 0.0658, "step": 5303 }, { "epoch": 4.86, "learning_rate": 7.0428249695685725e-06, "loss": 0.0655, "step": 5304 }, { "epoch": 4.86, "learning_rate": 7.039280304024126e-06, "loss": 0.0615, "step": 5305 }, { "epoch": 4.86, "learning_rate": 7.035736046179491e-06, "loss": 0.0664, "step": 5306 }, { "epoch": 4.86, "learning_rate": 7.032192196522716e-06, "loss": 0.0544, "step": 5307 }, { "epoch": 4.87, "learning_rate": 7.0286487555418e-06, "loss": 0.0735, "step": 5308 }, { "epoch": 4.87, "learning_rate": 7.025105723724684e-06, "loss": 0.0701, "step": 5309 }, { "epoch": 4.87, "learning_rate": 7.021563101559257e-06, "loss": 0.066, "step": 5310 }, { "epoch": 4.87, "learning_rate": 7.018020889533348e-06, "loss": 0.0644, "step": 5311 }, { "epoch": 4.87, "learning_rate": 7.014479088134729e-06, "loss": 0.0638, "step": 5312 }, { "epoch": 4.87, "learning_rate": 7.010937697851112e-06, "loss": 0.0658, "step": 5313 }, { "epoch": 4.87, "learning_rate": 7.007396719170165e-06, "loss": 0.0704, "step": 5314 }, { "epoch": 4.87, "learning_rate": 7.003856152579489e-06, "loss": 0.0636, "step": 5315 }, { "epoch": 4.87, "learning_rate": 7.000315998566622e-06, "loss": 0.0656, "step": 5316 }, { "epoch": 4.87, "learning_rate": 6.996776257619068e-06, "loss": 0.0806, "step": 5317 }, { "epoch": 4.87, "learning_rate": 6.993236930224251e-06, "loss": 0.062, "step": 5318 }, { "epoch": 4.88, "learning_rate": 6.989698016869546e-06, "loss": 0.068, "step": 5319 }, { "epoch": 4.88, "learning_rate": 6.986159518042273e-06, "loss": 0.0768, "step": 5320 }, { "epoch": 4.88, "learning_rate": 6.982621434229697e-06, "loss": 0.0622, "step": 5321 }, { "epoch": 4.88, "learning_rate": 6.979083765919019e-06, "loss": 0.067, "step": 5322 }, { "epoch": 4.88, "learning_rate": 6.975546513597387e-06, "loss": 0.066, "step": 5323 }, { "epoch": 4.88, "learning_rate": 6.9720096777518906e-06, "loss": 0.0597, "step": 5324 }, { "epoch": 4.88, "learning_rate": 6.968473258869566e-06, "loss": 0.0606, "step": 5325 }, { "epoch": 4.88, "learning_rate": 6.964937257437385e-06, "loss": 0.0704, "step": 5326 }, { "epoch": 4.88, "learning_rate": 6.961401673942263e-06, "loss": 0.0648, "step": 5327 }, { "epoch": 4.88, "learning_rate": 6.957866508871068e-06, "loss": 0.0583, "step": 5328 }, { "epoch": 4.88, "learning_rate": 6.954331762710597e-06, "loss": 0.0624, "step": 5329 }, { "epoch": 4.89, "learning_rate": 6.950797435947594e-06, "loss": 0.0828, "step": 5330 }, { "epoch": 4.89, "learning_rate": 6.94726352906875e-06, "loss": 0.069, "step": 5331 }, { "epoch": 4.89, "learning_rate": 6.943730042560692e-06, "loss": 0.0679, "step": 5332 }, { "epoch": 4.89, "learning_rate": 6.940196976909994e-06, "loss": 0.0652, "step": 5333 }, { "epoch": 4.89, "learning_rate": 6.936664332603163e-06, "loss": 0.0644, "step": 5334 }, { "epoch": 4.89, "learning_rate": 6.933132110126663e-06, "loss": 0.0673, "step": 5335 }, { "epoch": 4.89, "learning_rate": 6.929600309966889e-06, "loss": 0.0599, "step": 5336 }, { "epoch": 4.89, "learning_rate": 6.926068932610171e-06, "loss": 0.0664, "step": 5337 }, { "epoch": 4.89, "learning_rate": 6.922537978542805e-06, "loss": 0.0813, "step": 5338 }, { "epoch": 4.89, "learning_rate": 6.919007448251005e-06, "loss": 0.0715, "step": 5339 }, { "epoch": 4.89, "learning_rate": 6.915477342220938e-06, "loss": 0.0648, "step": 5340 }, { "epoch": 4.9, "learning_rate": 6.911947660938705e-06, "loss": 0.0639, "step": 5341 }, { "epoch": 4.9, "learning_rate": 6.9084184048903604e-06, "loss": 0.059, "step": 5342 }, { "epoch": 4.9, "learning_rate": 6.904889574561888e-06, "loss": 0.0744, "step": 5343 }, { "epoch": 4.9, "learning_rate": 6.901361170439224e-06, "loss": 0.0687, "step": 5344 }, { "epoch": 4.9, "learning_rate": 6.897833193008232e-06, "loss": 0.0679, "step": 5345 }, { "epoch": 4.9, "learning_rate": 6.89430564275473e-06, "loss": 0.0676, "step": 5346 }, { "epoch": 4.9, "learning_rate": 6.890778520164474e-06, "loss": 0.0644, "step": 5347 }, { "epoch": 4.9, "learning_rate": 6.887251825723153e-06, "loss": 0.0645, "step": 5348 }, { "epoch": 4.9, "learning_rate": 6.883725559916409e-06, "loss": 0.0608, "step": 5349 }, { "epoch": 4.9, "learning_rate": 6.880199723229817e-06, "loss": 0.0608, "step": 5350 }, { "epoch": 4.9, "learning_rate": 6.876674316148895e-06, "loss": 0.0677, "step": 5351 }, { "epoch": 4.91, "learning_rate": 6.873149339159097e-06, "loss": 0.0798, "step": 5352 }, { "epoch": 4.91, "learning_rate": 6.869624792745832e-06, "loss": 0.0722, "step": 5353 }, { "epoch": 4.91, "learning_rate": 6.866100677394436e-06, "loss": 0.0698, "step": 5354 }, { "epoch": 4.91, "learning_rate": 6.862576993590186e-06, "loss": 0.0642, "step": 5355 }, { "epoch": 4.91, "learning_rate": 6.859053741818311e-06, "loss": 0.0631, "step": 5356 }, { "epoch": 4.91, "learning_rate": 6.85553092256397e-06, "loss": 0.0713, "step": 5357 }, { "epoch": 4.91, "learning_rate": 6.852008536312265e-06, "loss": 0.0729, "step": 5358 }, { "epoch": 4.91, "learning_rate": 6.848486583548235e-06, "loss": 0.0691, "step": 5359 }, { "epoch": 4.91, "learning_rate": 6.844965064756871e-06, "loss": 0.0737, "step": 5360 }, { "epoch": 4.91, "learning_rate": 6.841443980423092e-06, "loss": 0.0651, "step": 5361 }, { "epoch": 4.91, "learning_rate": 6.837923331031761e-06, "loss": 0.074, "step": 5362 }, { "epoch": 4.92, "learning_rate": 6.834403117067685e-06, "loss": 0.0732, "step": 5363 }, { "epoch": 4.92, "learning_rate": 6.830883339015608e-06, "loss": 0.0644, "step": 5364 }, { "epoch": 4.92, "learning_rate": 6.827363997360208e-06, "loss": 0.065, "step": 5365 }, { "epoch": 4.92, "learning_rate": 6.823845092586114e-06, "loss": 0.0694, "step": 5366 }, { "epoch": 4.92, "learning_rate": 6.820326625177886e-06, "loss": 0.062, "step": 5367 }, { "epoch": 4.92, "learning_rate": 6.8168085956200345e-06, "loss": 0.0618, "step": 5368 }, { "epoch": 4.92, "learning_rate": 6.8132910043969955e-06, "loss": 0.0713, "step": 5369 }, { "epoch": 4.92, "learning_rate": 6.80977385199315e-06, "loss": 0.0632, "step": 5370 }, { "epoch": 4.92, "learning_rate": 6.8062571388928275e-06, "loss": 0.0645, "step": 5371 }, { "epoch": 4.92, "learning_rate": 6.802740865580287e-06, "loss": 0.0607, "step": 5372 }, { "epoch": 4.92, "learning_rate": 6.799225032539724e-06, "loss": 0.0687, "step": 5373 }, { "epoch": 4.93, "learning_rate": 6.795709640255287e-06, "loss": 0.0638, "step": 5374 }, { "epoch": 4.93, "learning_rate": 6.792194689211053e-06, "loss": 0.0631, "step": 5375 }, { "epoch": 4.93, "learning_rate": 6.788680179891041e-06, "loss": 0.0652, "step": 5376 }, { "epoch": 4.93, "learning_rate": 6.785166112779205e-06, "loss": 0.0711, "step": 5377 }, { "epoch": 4.93, "learning_rate": 6.7816524883594505e-06, "loss": 0.0621, "step": 5378 }, { "epoch": 4.93, "learning_rate": 6.77813930711561e-06, "loss": 0.0621, "step": 5379 }, { "epoch": 4.93, "learning_rate": 6.774626569531453e-06, "loss": 0.0748, "step": 5380 }, { "epoch": 4.93, "learning_rate": 6.7711142760907045e-06, "loss": 0.0673, "step": 5381 }, { "epoch": 4.93, "learning_rate": 6.7676024272770135e-06, "loss": 0.0648, "step": 5382 }, { "epoch": 4.93, "learning_rate": 6.7640910235739685e-06, "loss": 0.0651, "step": 5383 }, { "epoch": 4.93, "learning_rate": 6.760580065465101e-06, "loss": 0.0636, "step": 5384 }, { "epoch": 4.94, "learning_rate": 6.757069553433885e-06, "loss": 0.062, "step": 5385 }, { "epoch": 4.94, "learning_rate": 6.753559487963724e-06, "loss": 0.0634, "step": 5386 }, { "epoch": 4.94, "learning_rate": 6.7500498695379635e-06, "loss": 0.0677, "step": 5387 }, { "epoch": 4.94, "learning_rate": 6.746540698639893e-06, "loss": 0.0645, "step": 5388 }, { "epoch": 4.94, "learning_rate": 6.743031975752732e-06, "loss": 0.0603, "step": 5389 }, { "epoch": 4.94, "learning_rate": 6.7395237013596425e-06, "loss": 0.0613, "step": 5390 }, { "epoch": 4.94, "learning_rate": 6.736015875943722e-06, "loss": 0.0637, "step": 5391 }, { "epoch": 4.94, "learning_rate": 6.732508499988016e-06, "loss": 0.0634, "step": 5392 }, { "epoch": 4.94, "learning_rate": 6.729001573975495e-06, "loss": 0.0692, "step": 5393 }, { "epoch": 4.94, "learning_rate": 6.725495098389068e-06, "loss": 0.0592, "step": 5394 }, { "epoch": 4.95, "learning_rate": 6.7219890737115964e-06, "loss": 0.0652, "step": 5395 }, { "epoch": 4.95, "learning_rate": 6.718483500425868e-06, "loss": 0.0588, "step": 5396 }, { "epoch": 4.95, "learning_rate": 6.714978379014608e-06, "loss": 0.0586, "step": 5397 }, { "epoch": 4.95, "learning_rate": 6.711473709960478e-06, "loss": 0.0594, "step": 5398 }, { "epoch": 4.95, "learning_rate": 6.70796949374609e-06, "loss": 0.0675, "step": 5399 }, { "epoch": 4.95, "learning_rate": 6.704465730853981e-06, "loss": 0.0693, "step": 5400 }, { "epoch": 4.95, "learning_rate": 6.700962421766628e-06, "loss": 0.0662, "step": 5401 }, { "epoch": 4.95, "learning_rate": 6.697459566966447e-06, "loss": 0.0705, "step": 5402 }, { "epoch": 4.95, "learning_rate": 6.693957166935794e-06, "loss": 0.0674, "step": 5403 }, { "epoch": 4.95, "learning_rate": 6.690455222156959e-06, "loss": 0.0666, "step": 5404 }, { "epoch": 4.95, "learning_rate": 6.686953733112164e-06, "loss": 0.0709, "step": 5405 }, { "epoch": 4.96, "learning_rate": 6.683452700283586e-06, "loss": 0.0581, "step": 5406 }, { "epoch": 4.96, "learning_rate": 6.67995212415332e-06, "loss": 0.0733, "step": 5407 }, { "epoch": 4.96, "learning_rate": 6.6764520052034054e-06, "loss": 0.0712, "step": 5408 }, { "epoch": 4.96, "learning_rate": 6.6729523439158196e-06, "loss": 0.0671, "step": 5409 }, { "epoch": 4.96, "learning_rate": 6.669453140772477e-06, "loss": 0.0661, "step": 5410 }, { "epoch": 4.96, "learning_rate": 6.665954396255232e-06, "loss": 0.0692, "step": 5411 }, { "epoch": 4.96, "learning_rate": 6.6624561108458644e-06, "loss": 0.0544, "step": 5412 }, { "epoch": 4.96, "learning_rate": 6.6589582850261025e-06, "loss": 0.0717, "step": 5413 }, { "epoch": 4.96, "learning_rate": 6.6554609192776085e-06, "loss": 0.0579, "step": 5414 }, { "epoch": 4.96, "learning_rate": 6.651964014081979e-06, "loss": 0.0604, "step": 5415 }, { "epoch": 4.96, "learning_rate": 6.648467569920742e-06, "loss": 0.0718, "step": 5416 }, { "epoch": 4.97, "learning_rate": 6.644971587275379e-06, "loss": 0.0669, "step": 5417 }, { "epoch": 4.97, "learning_rate": 6.6414760666272915e-06, "loss": 0.0631, "step": 5418 }, { "epoch": 4.97, "learning_rate": 6.637981008457819e-06, "loss": 0.0678, "step": 5419 }, { "epoch": 4.97, "learning_rate": 6.634486413248249e-06, "loss": 0.0684, "step": 5420 }, { "epoch": 4.97, "learning_rate": 6.630992281479794e-06, "loss": 0.0674, "step": 5421 }, { "epoch": 4.97, "learning_rate": 6.627498613633607e-06, "loss": 0.0672, "step": 5422 }, { "epoch": 4.97, "learning_rate": 6.62400541019077e-06, "loss": 0.0744, "step": 5423 }, { "epoch": 4.97, "learning_rate": 6.620512671632317e-06, "loss": 0.0541, "step": 5424 }, { "epoch": 4.97, "learning_rate": 6.617020398439205e-06, "loss": 0.0583, "step": 5425 }, { "epoch": 4.97, "learning_rate": 6.613528591092326e-06, "loss": 0.0601, "step": 5426 }, { "epoch": 4.97, "learning_rate": 6.610037250072518e-06, "loss": 0.0713, "step": 5427 }, { "epoch": 4.98, "learning_rate": 6.606546375860549e-06, "loss": 0.061, "step": 5428 }, { "epoch": 4.98, "learning_rate": 6.6030559689371165e-06, "loss": 0.0625, "step": 5429 }, { "epoch": 4.98, "learning_rate": 6.599566029782863e-06, "loss": 0.0676, "step": 5430 }, { "epoch": 4.98, "learning_rate": 6.596076558878367e-06, "loss": 0.062, "step": 5431 }, { "epoch": 4.98, "learning_rate": 6.592587556704133e-06, "loss": 0.0616, "step": 5432 }, { "epoch": 4.98, "learning_rate": 6.589099023740613e-06, "loss": 0.0701, "step": 5433 }, { "epoch": 4.98, "learning_rate": 6.585610960468181e-06, "loss": 0.0618, "step": 5434 }, { "epoch": 4.98, "learning_rate": 6.582123367367159e-06, "loss": 0.063, "step": 5435 }, { "epoch": 4.98, "learning_rate": 6.578636244917799e-06, "loss": 0.0669, "step": 5436 }, { "epoch": 4.98, "learning_rate": 6.575149593600282e-06, "loss": 0.063, "step": 5437 }, { "epoch": 4.98, "learning_rate": 6.571663413894738e-06, "loss": 0.0582, "step": 5438 }, { "epoch": 4.99, "learning_rate": 6.568177706281221e-06, "loss": 0.0697, "step": 5439 }, { "epoch": 4.99, "learning_rate": 6.564692471239723e-06, "loss": 0.0617, "step": 5440 }, { "epoch": 4.99, "learning_rate": 6.561207709250167e-06, "loss": 0.0695, "step": 5441 }, { "epoch": 4.99, "learning_rate": 6.557723420792422e-06, "loss": 0.0662, "step": 5442 }, { "epoch": 4.99, "learning_rate": 6.554239606346282e-06, "loss": 0.0714, "step": 5443 }, { "epoch": 4.99, "learning_rate": 6.5507562663914745e-06, "loss": 0.075, "step": 5444 }, { "epoch": 4.99, "learning_rate": 6.547273401407675e-06, "loss": 0.0623, "step": 5445 }, { "epoch": 4.99, "learning_rate": 6.5437910118744764e-06, "loss": 0.0601, "step": 5446 }, { "epoch": 4.99, "learning_rate": 6.540309098271416e-06, "loss": 0.0642, "step": 5447 }, { "epoch": 4.99, "learning_rate": 6.53682766107796e-06, "loss": 0.063, "step": 5448 }, { "epoch": 4.99, "learning_rate": 6.53334670077352e-06, "loss": 0.0686, "step": 5449 }, { "epoch": 5.0, "learning_rate": 6.529866217837431e-06, "loss": 0.0754, "step": 5450 }, { "epoch": 5.0, "learning_rate": 6.526386212748961e-06, "loss": 0.07, "step": 5451 }, { "epoch": 5.0, "learning_rate": 6.522906685987327e-06, "loss": 0.0653, "step": 5452 }, { "epoch": 5.0, "learning_rate": 6.519427638031662e-06, "loss": 0.0633, "step": 5453 }, { "epoch": 5.0, "learning_rate": 6.5159490693610425e-06, "loss": 0.0639, "step": 5454 }, { "epoch": 5.0, "learning_rate": 6.51247098045448e-06, "loss": 0.0659, "step": 5455 }, { "epoch": 5.0, "learning_rate": 6.508993371790915e-06, "loss": 0.0547, "step": 5456 }, { "epoch": 5.0, "learning_rate": 6.505516243849227e-06, "loss": 0.0637, "step": 5457 }, { "epoch": 5.0, "learning_rate": 6.502039597108226e-06, "loss": 0.0558, "step": 5458 }, { "epoch": 5.0, "learning_rate": 6.4985634320466514e-06, "loss": 0.0471, "step": 5459 }, { "epoch": 5.0, "learning_rate": 6.495087749143191e-06, "loss": 0.0492, "step": 5460 }, { "epoch": 5.01, "learning_rate": 6.491612548876451e-06, "loss": 0.0542, "step": 5461 }, { "epoch": 5.01, "learning_rate": 6.488137831724972e-06, "loss": 0.0536, "step": 5462 }, { "epoch": 5.01, "learning_rate": 6.484663598167243e-06, "loss": 0.0435, "step": 5463 }, { "epoch": 5.01, "learning_rate": 6.48118984868167e-06, "loss": 0.0567, "step": 5464 }, { "epoch": 5.01, "learning_rate": 6.477716583746601e-06, "loss": 0.0723, "step": 5465 }, { "epoch": 5.01, "learning_rate": 6.474243803840308e-06, "loss": 0.0592, "step": 5466 }, { "epoch": 5.01, "learning_rate": 6.470771509441014e-06, "loss": 0.054, "step": 5467 }, { "epoch": 5.01, "learning_rate": 6.4672997010268564e-06, "loss": 0.0647, "step": 5468 }, { "epoch": 5.01, "learning_rate": 6.463828379075912e-06, "loss": 0.0519, "step": 5469 }, { "epoch": 5.01, "learning_rate": 6.4603575440662e-06, "loss": 0.0478, "step": 5470 }, { "epoch": 5.01, "learning_rate": 6.45688719647566e-06, "loss": 0.0543, "step": 5471 }, { "epoch": 5.02, "learning_rate": 6.45341733678217e-06, "loss": 0.0534, "step": 5472 }, { "epoch": 5.02, "learning_rate": 6.449947965463535e-06, "loss": 0.0554, "step": 5473 }, { "epoch": 5.02, "learning_rate": 6.4464790829975034e-06, "loss": 0.0554, "step": 5474 }, { "epoch": 5.02, "learning_rate": 6.4430106898617486e-06, "loss": 0.0629, "step": 5475 }, { "epoch": 5.02, "learning_rate": 6.439542786533879e-06, "loss": 0.054, "step": 5476 }, { "epoch": 5.02, "learning_rate": 6.436075373491436e-06, "loss": 0.0504, "step": 5477 }, { "epoch": 5.02, "learning_rate": 6.43260845121189e-06, "loss": 0.0539, "step": 5478 }, { "epoch": 5.02, "learning_rate": 6.429142020172649e-06, "loss": 0.0722, "step": 5479 }, { "epoch": 5.02, "learning_rate": 6.425676080851047e-06, "loss": 0.058, "step": 5480 }, { "epoch": 5.02, "learning_rate": 6.42221063372436e-06, "loss": 0.0562, "step": 5481 }, { "epoch": 5.02, "learning_rate": 6.418745679269785e-06, "loss": 0.0581, "step": 5482 }, { "epoch": 5.03, "learning_rate": 6.415281217964456e-06, "loss": 0.0531, "step": 5483 }, { "epoch": 5.03, "learning_rate": 6.411817250285446e-06, "loss": 0.0515, "step": 5484 }, { "epoch": 5.03, "learning_rate": 6.408353776709748e-06, "loss": 0.0632, "step": 5485 }, { "epoch": 5.03, "learning_rate": 6.404890797714293e-06, "loss": 0.0606, "step": 5486 }, { "epoch": 5.03, "learning_rate": 6.401428313775944e-06, "loss": 0.0514, "step": 5487 }, { "epoch": 5.03, "learning_rate": 6.397966325371496e-06, "loss": 0.057, "step": 5488 }, { "epoch": 5.03, "learning_rate": 6.394504832977675e-06, "loss": 0.0467, "step": 5489 }, { "epoch": 5.03, "learning_rate": 6.391043837071138e-06, "loss": 0.0497, "step": 5490 }, { "epoch": 5.03, "learning_rate": 6.387583338128471e-06, "loss": 0.0488, "step": 5491 }, { "epoch": 5.03, "learning_rate": 6.3841233366262025e-06, "loss": 0.0465, "step": 5492 }, { "epoch": 5.03, "learning_rate": 6.380663833040779e-06, "loss": 0.0538, "step": 5493 }, { "epoch": 5.04, "learning_rate": 6.377204827848583e-06, "loss": 0.0577, "step": 5494 }, { "epoch": 5.04, "learning_rate": 6.373746321525937e-06, "loss": 0.0569, "step": 5495 }, { "epoch": 5.04, "learning_rate": 6.370288314549083e-06, "loss": 0.0517, "step": 5496 }, { "epoch": 5.04, "learning_rate": 6.366830807394195e-06, "loss": 0.0526, "step": 5497 }, { "epoch": 5.04, "learning_rate": 6.363373800537388e-06, "loss": 0.0509, "step": 5498 }, { "epoch": 5.04, "learning_rate": 6.3599172944546984e-06, "loss": 0.0539, "step": 5499 }, { "epoch": 5.04, "learning_rate": 6.356461289622101e-06, "loss": 0.0615, "step": 5500 }, { "epoch": 5.04, "learning_rate": 6.353005786515494e-06, "loss": 0.04, "step": 5501 }, { "epoch": 5.04, "learning_rate": 6.34955078561071e-06, "loss": 0.0414, "step": 5502 }, { "epoch": 5.04, "learning_rate": 6.346096287383519e-06, "loss": 0.0544, "step": 5503 }, { "epoch": 5.04, "learning_rate": 6.342642292309609e-06, "loss": 0.0586, "step": 5504 }, { "epoch": 5.05, "learning_rate": 6.339188800864604e-06, "loss": 0.0576, "step": 5505 }, { "epoch": 5.05, "learning_rate": 6.3357358135240664e-06, "loss": 0.0549, "step": 5506 }, { "epoch": 5.05, "learning_rate": 6.33228333076348e-06, "loss": 0.0574, "step": 5507 }, { "epoch": 5.05, "learning_rate": 6.328831353058258e-06, "loss": 0.0476, "step": 5508 }, { "epoch": 5.05, "learning_rate": 6.3253798808837555e-06, "loss": 0.0505, "step": 5509 }, { "epoch": 5.05, "learning_rate": 6.3219289147152465e-06, "loss": 0.054, "step": 5510 }, { "epoch": 5.05, "learning_rate": 6.318478455027938e-06, "loss": 0.0565, "step": 5511 }, { "epoch": 5.05, "learning_rate": 6.315028502296965e-06, "loss": 0.0617, "step": 5512 }, { "epoch": 5.05, "learning_rate": 6.311579056997406e-06, "loss": 0.0552, "step": 5513 }, { "epoch": 5.05, "learning_rate": 6.3081301196042524e-06, "loss": 0.0602, "step": 5514 }, { "epoch": 5.05, "learning_rate": 6.304681690592431e-06, "loss": 0.0576, "step": 5515 }, { "epoch": 5.06, "learning_rate": 6.3012337704368075e-06, "loss": 0.0523, "step": 5516 }, { "epoch": 5.06, "learning_rate": 6.2977863596121704e-06, "loss": 0.065, "step": 5517 }, { "epoch": 5.06, "learning_rate": 6.29433945859323e-06, "loss": 0.0469, "step": 5518 }, { "epoch": 5.06, "learning_rate": 6.290893067854643e-06, "loss": 0.0563, "step": 5519 }, { "epoch": 5.06, "learning_rate": 6.287447187870986e-06, "loss": 0.0489, "step": 5520 }, { "epoch": 5.06, "learning_rate": 6.284001819116763e-06, "loss": 0.04, "step": 5521 }, { "epoch": 5.06, "learning_rate": 6.280556962066415e-06, "loss": 0.0524, "step": 5522 }, { "epoch": 5.06, "learning_rate": 6.277112617194305e-06, "loss": 0.0563, "step": 5523 }, { "epoch": 5.06, "learning_rate": 6.273668784974737e-06, "loss": 0.0648, "step": 5524 }, { "epoch": 5.06, "learning_rate": 6.270225465881932e-06, "loss": 0.0585, "step": 5525 }, { "epoch": 5.07, "learning_rate": 6.266782660390039e-06, "loss": 0.0499, "step": 5526 }, { "epoch": 5.07, "learning_rate": 6.263340368973154e-06, "loss": 0.0583, "step": 5527 }, { "epoch": 5.07, "learning_rate": 6.259898592105287e-06, "loss": 0.0609, "step": 5528 }, { "epoch": 5.07, "learning_rate": 6.256457330260377e-06, "loss": 0.0637, "step": 5529 }, { "epoch": 5.07, "learning_rate": 6.253016583912295e-06, "loss": 0.0501, "step": 5530 }, { "epoch": 5.07, "learning_rate": 6.2495763535348475e-06, "loss": 0.052, "step": 5531 }, { "epoch": 5.07, "learning_rate": 6.246136639601763e-06, "loss": 0.0481, "step": 5532 }, { "epoch": 5.07, "learning_rate": 6.242697442586693e-06, "loss": 0.0469, "step": 5533 }, { "epoch": 5.07, "learning_rate": 6.239258762963238e-06, "loss": 0.0483, "step": 5534 }, { "epoch": 5.07, "learning_rate": 6.235820601204906e-06, "loss": 0.0536, "step": 5535 }, { "epoch": 5.07, "learning_rate": 6.232382957785143e-06, "loss": 0.0476, "step": 5536 }, { "epoch": 5.08, "learning_rate": 6.228945833177319e-06, "loss": 0.0521, "step": 5537 }, { "epoch": 5.08, "learning_rate": 6.225509227854745e-06, "loss": 0.0563, "step": 5538 }, { "epoch": 5.08, "learning_rate": 6.222073142290644e-06, "loss": 0.0531, "step": 5539 }, { "epoch": 5.08, "learning_rate": 6.2186375769581755e-06, "loss": 0.0521, "step": 5540 }, { "epoch": 5.08, "learning_rate": 6.215202532330433e-06, "loss": 0.0533, "step": 5541 }, { "epoch": 5.08, "learning_rate": 6.211768008880427e-06, "loss": 0.0575, "step": 5542 }, { "epoch": 5.08, "learning_rate": 6.208334007081105e-06, "loss": 0.0458, "step": 5543 }, { "epoch": 5.08, "learning_rate": 6.204900527405333e-06, "loss": 0.0594, "step": 5544 }, { "epoch": 5.08, "learning_rate": 6.201467570325918e-06, "loss": 0.0513, "step": 5545 }, { "epoch": 5.08, "learning_rate": 6.198035136315585e-06, "loss": 0.0482, "step": 5546 }, { "epoch": 5.08, "learning_rate": 6.194603225846993e-06, "loss": 0.0485, "step": 5547 }, { "epoch": 5.09, "learning_rate": 6.19117183939272e-06, "loss": 0.0531, "step": 5548 }, { "epoch": 5.09, "learning_rate": 6.187740977425285e-06, "loss": 0.0594, "step": 5549 }, { "epoch": 5.09, "learning_rate": 6.184310640417125e-06, "loss": 0.053, "step": 5550 }, { "epoch": 5.09, "learning_rate": 6.180880828840603e-06, "loss": 0.051, "step": 5551 }, { "epoch": 5.09, "learning_rate": 6.177451543168022e-06, "loss": 0.0523, "step": 5552 }, { "epoch": 5.09, "learning_rate": 6.174022783871603e-06, "loss": 0.0553, "step": 5553 }, { "epoch": 5.09, "learning_rate": 6.170594551423494e-06, "loss": 0.0659, "step": 5554 }, { "epoch": 5.09, "learning_rate": 6.167166846295767e-06, "loss": 0.057, "step": 5555 }, { "epoch": 5.09, "learning_rate": 6.163739668960439e-06, "loss": 0.0521, "step": 5556 }, { "epoch": 5.09, "learning_rate": 6.1603130198894354e-06, "loss": 0.0517, "step": 5557 }, { "epoch": 5.09, "learning_rate": 6.156886899554614e-06, "loss": 0.0477, "step": 5558 }, { "epoch": 5.1, "learning_rate": 6.153461308427768e-06, "loss": 0.0551, "step": 5559 }, { "epoch": 5.1, "learning_rate": 6.150036246980608e-06, "loss": 0.0474, "step": 5560 }, { "epoch": 5.1, "learning_rate": 6.146611715684774e-06, "loss": 0.0556, "step": 5561 }, { "epoch": 5.1, "learning_rate": 6.143187715011833e-06, "loss": 0.0518, "step": 5562 }, { "epoch": 5.1, "learning_rate": 6.139764245433286e-06, "loss": 0.0496, "step": 5563 }, { "epoch": 5.1, "learning_rate": 6.136341307420548e-06, "loss": 0.0546, "step": 5564 }, { "epoch": 5.1, "learning_rate": 6.132918901444971e-06, "loss": 0.0627, "step": 5565 }, { "epoch": 5.1, "learning_rate": 6.129497027977829e-06, "loss": 0.0549, "step": 5566 }, { "epoch": 5.1, "learning_rate": 6.126075687490327e-06, "loss": 0.0608, "step": 5567 }, { "epoch": 5.1, "learning_rate": 6.122654880453591e-06, "loss": 0.0639, "step": 5568 }, { "epoch": 5.1, "learning_rate": 6.119234607338673e-06, "loss": 0.0629, "step": 5569 }, { "epoch": 5.11, "learning_rate": 6.115814868616563e-06, "loss": 0.0597, "step": 5570 }, { "epoch": 5.11, "learning_rate": 6.112395664758164e-06, "loss": 0.0471, "step": 5571 }, { "epoch": 5.11, "learning_rate": 6.108976996234307e-06, "loss": 0.06, "step": 5572 }, { "epoch": 5.11, "learning_rate": 6.105558863515762e-06, "loss": 0.0525, "step": 5573 }, { "epoch": 5.11, "learning_rate": 6.102141267073207e-06, "loss": 0.0515, "step": 5574 }, { "epoch": 5.11, "learning_rate": 6.098724207377261e-06, "loss": 0.0555, "step": 5575 }, { "epoch": 5.11, "learning_rate": 6.095307684898457e-06, "loss": 0.0558, "step": 5576 }, { "epoch": 5.11, "learning_rate": 6.091891700107267e-06, "loss": 0.0536, "step": 5577 }, { "epoch": 5.11, "learning_rate": 6.088476253474078e-06, "loss": 0.0522, "step": 5578 }, { "epoch": 5.11, "learning_rate": 6.08506134546921e-06, "loss": 0.0523, "step": 5579 }, { "epoch": 5.11, "learning_rate": 6.0816469765629e-06, "loss": 0.0574, "step": 5580 }, { "epoch": 5.12, "learning_rate": 6.078233147225323e-06, "loss": 0.0532, "step": 5581 }, { "epoch": 5.12, "learning_rate": 6.074819857926573e-06, "loss": 0.0531, "step": 5582 }, { "epoch": 5.12, "learning_rate": 6.071407109136662e-06, "loss": 0.0501, "step": 5583 }, { "epoch": 5.12, "learning_rate": 6.0679949013255455e-06, "loss": 0.0542, "step": 5584 }, { "epoch": 5.12, "learning_rate": 6.0645832349630895e-06, "loss": 0.0548, "step": 5585 }, { "epoch": 5.12, "learning_rate": 6.0611721105190935e-06, "loss": 0.0522, "step": 5586 }, { "epoch": 5.12, "learning_rate": 6.057761528463275e-06, "loss": 0.0572, "step": 5587 }, { "epoch": 5.12, "learning_rate": 6.054351489265284e-06, "loss": 0.048, "step": 5588 }, { "epoch": 5.12, "learning_rate": 6.050941993394694e-06, "loss": 0.0522, "step": 5589 }, { "epoch": 5.12, "learning_rate": 6.047533041320998e-06, "loss": 0.0518, "step": 5590 }, { "epoch": 5.12, "learning_rate": 6.044124633513624e-06, "loss": 0.0585, "step": 5591 }, { "epoch": 5.13, "learning_rate": 6.040716770441917e-06, "loss": 0.0426, "step": 5592 }, { "epoch": 5.13, "learning_rate": 6.037309452575152e-06, "loss": 0.0482, "step": 5593 }, { "epoch": 5.13, "learning_rate": 6.0339026803825204e-06, "loss": 0.0505, "step": 5594 }, { "epoch": 5.13, "learning_rate": 6.030496454333154e-06, "loss": 0.0474, "step": 5595 }, { "epoch": 5.13, "learning_rate": 6.027090774896095e-06, "loss": 0.0493, "step": 5596 }, { "epoch": 5.13, "learning_rate": 6.023685642540312e-06, "loss": 0.0515, "step": 5597 }, { "epoch": 5.13, "learning_rate": 6.02028105773471e-06, "loss": 0.0623, "step": 5598 }, { "epoch": 5.13, "learning_rate": 6.016877020948107e-06, "loss": 0.0454, "step": 5599 }, { "epoch": 5.13, "learning_rate": 6.013473532649246e-06, "loss": 0.0481, "step": 5600 }, { "epoch": 5.13, "learning_rate": 6.010070593306796e-06, "loss": 0.0549, "step": 5601 }, { "epoch": 5.13, "learning_rate": 6.006668203389359e-06, "loss": 0.0567, "step": 5602 }, { "epoch": 5.14, "learning_rate": 6.00326636336545e-06, "loss": 0.0523, "step": 5603 }, { "epoch": 5.14, "learning_rate": 5.999865073703508e-06, "loss": 0.0457, "step": 5604 }, { "epoch": 5.14, "learning_rate": 5.99646433487191e-06, "loss": 0.0466, "step": 5605 }, { "epoch": 5.14, "learning_rate": 5.993064147338943e-06, "loss": 0.0548, "step": 5606 }, { "epoch": 5.14, "learning_rate": 5.989664511572819e-06, "loss": 0.0626, "step": 5607 }, { "epoch": 5.14, "learning_rate": 5.986265428041682e-06, "loss": 0.0404, "step": 5608 }, { "epoch": 5.14, "learning_rate": 5.9828668972135984e-06, "loss": 0.0507, "step": 5609 }, { "epoch": 5.14, "learning_rate": 5.97946891955655e-06, "loss": 0.057, "step": 5610 }, { "epoch": 5.14, "learning_rate": 5.976071495538454e-06, "loss": 0.0609, "step": 5611 }, { "epoch": 5.14, "learning_rate": 5.9726746256271374e-06, "loss": 0.0536, "step": 5612 }, { "epoch": 5.14, "learning_rate": 5.969278310290367e-06, "loss": 0.057, "step": 5613 }, { "epoch": 5.15, "learning_rate": 5.965882549995825e-06, "loss": 0.0557, "step": 5614 }, { "epoch": 5.15, "learning_rate": 5.962487345211112e-06, "loss": 0.0618, "step": 5615 }, { "epoch": 5.15, "learning_rate": 5.9590926964037645e-06, "loss": 0.0483, "step": 5616 }, { "epoch": 5.15, "learning_rate": 5.955698604041231e-06, "loss": 0.0511, "step": 5617 }, { "epoch": 5.15, "learning_rate": 5.952305068590891e-06, "loss": 0.0458, "step": 5618 }, { "epoch": 5.15, "learning_rate": 5.948912090520038e-06, "loss": 0.0529, "step": 5619 }, { "epoch": 5.15, "learning_rate": 5.945519670295904e-06, "loss": 0.0641, "step": 5620 }, { "epoch": 5.15, "learning_rate": 5.942127808385631e-06, "loss": 0.0492, "step": 5621 }, { "epoch": 5.15, "learning_rate": 5.938736505256283e-06, "loss": 0.0474, "step": 5622 }, { "epoch": 5.15, "learning_rate": 5.935345761374862e-06, "loss": 0.0565, "step": 5623 }, { "epoch": 5.15, "learning_rate": 5.93195557720828e-06, "loss": 0.0546, "step": 5624 }, { "epoch": 5.16, "learning_rate": 5.928565953223373e-06, "loss": 0.0516, "step": 5625 }, { "epoch": 5.16, "learning_rate": 5.925176889886901e-06, "loss": 0.0576, "step": 5626 }, { "epoch": 5.16, "learning_rate": 5.921788387665554e-06, "loss": 0.0527, "step": 5627 }, { "epoch": 5.16, "learning_rate": 5.918400447025934e-06, "loss": 0.0452, "step": 5628 }, { "epoch": 5.16, "learning_rate": 5.91501306843457e-06, "loss": 0.048, "step": 5629 }, { "epoch": 5.16, "learning_rate": 5.911626252357918e-06, "loss": 0.0514, "step": 5630 }, { "epoch": 5.16, "learning_rate": 5.9082399992623485e-06, "loss": 0.0537, "step": 5631 }, { "epoch": 5.16, "learning_rate": 5.904854309614162e-06, "loss": 0.0591, "step": 5632 }, { "epoch": 5.16, "learning_rate": 5.901469183879573e-06, "loss": 0.0571, "step": 5633 }, { "epoch": 5.16, "learning_rate": 5.898084622524729e-06, "loss": 0.0573, "step": 5634 }, { "epoch": 5.16, "learning_rate": 5.894700626015691e-06, "loss": 0.0539, "step": 5635 }, { "epoch": 5.17, "learning_rate": 5.891317194818447e-06, "loss": 0.0508, "step": 5636 }, { "epoch": 5.17, "learning_rate": 5.8879343293989e-06, "loss": 0.0489, "step": 5637 }, { "epoch": 5.17, "learning_rate": 5.884552030222888e-06, "loss": 0.0621, "step": 5638 }, { "epoch": 5.17, "learning_rate": 5.8811702977561606e-06, "loss": 0.0484, "step": 5639 }, { "epoch": 5.17, "learning_rate": 5.877789132464387e-06, "loss": 0.0505, "step": 5640 }, { "epoch": 5.17, "learning_rate": 5.874408534813174e-06, "loss": 0.0649, "step": 5641 }, { "epoch": 5.17, "learning_rate": 5.871028505268032e-06, "loss": 0.0504, "step": 5642 }, { "epoch": 5.17, "learning_rate": 5.867649044294406e-06, "loss": 0.0507, "step": 5643 }, { "epoch": 5.17, "learning_rate": 5.8642701523576496e-06, "loss": 0.0615, "step": 5644 }, { "epoch": 5.17, "learning_rate": 5.860891829923057e-06, "loss": 0.0631, "step": 5645 }, { "epoch": 5.18, "learning_rate": 5.857514077455826e-06, "loss": 0.0468, "step": 5646 }, { "epoch": 5.18, "learning_rate": 5.8541368954210805e-06, "loss": 0.0534, "step": 5647 }, { "epoch": 5.18, "learning_rate": 5.8507602842838775e-06, "loss": 0.0665, "step": 5648 }, { "epoch": 5.18, "learning_rate": 5.84738424450918e-06, "loss": 0.0537, "step": 5649 }, { "epoch": 5.18, "learning_rate": 5.84400877656188e-06, "loss": 0.0532, "step": 5650 }, { "epoch": 5.18, "learning_rate": 5.8406338809067874e-06, "loss": 0.0491, "step": 5651 }, { "epoch": 5.18, "learning_rate": 5.8372595580086415e-06, "loss": 0.0558, "step": 5652 }, { "epoch": 5.18, "learning_rate": 5.833885808332091e-06, "loss": 0.0476, "step": 5653 }, { "epoch": 5.18, "learning_rate": 5.830512632341707e-06, "loss": 0.0546, "step": 5654 }, { "epoch": 5.18, "learning_rate": 5.827140030501997e-06, "loss": 0.0593, "step": 5655 }, { "epoch": 5.18, "learning_rate": 5.823768003277372e-06, "loss": 0.0575, "step": 5656 }, { "epoch": 5.19, "learning_rate": 5.82039655113217e-06, "loss": 0.0537, "step": 5657 }, { "epoch": 5.19, "learning_rate": 5.817025674530647e-06, "loss": 0.0619, "step": 5658 }, { "epoch": 5.19, "learning_rate": 5.813655373936988e-06, "loss": 0.0514, "step": 5659 }, { "epoch": 5.19, "learning_rate": 5.810285649815293e-06, "loss": 0.0628, "step": 5660 }, { "epoch": 5.19, "learning_rate": 5.806916502629576e-06, "loss": 0.051, "step": 5661 }, { "epoch": 5.19, "learning_rate": 5.803547932843787e-06, "loss": 0.058, "step": 5662 }, { "epoch": 5.19, "learning_rate": 5.8001799409217815e-06, "loss": 0.0498, "step": 5663 }, { "epoch": 5.19, "learning_rate": 5.796812527327349e-06, "loss": 0.0512, "step": 5664 }, { "epoch": 5.19, "learning_rate": 5.793445692524188e-06, "loss": 0.0502, "step": 5665 }, { "epoch": 5.19, "learning_rate": 5.790079436975917e-06, "loss": 0.0532, "step": 5666 }, { "epoch": 5.19, "learning_rate": 5.786713761146089e-06, "loss": 0.0687, "step": 5667 }, { "epoch": 5.2, "learning_rate": 5.78334866549816e-06, "loss": 0.0458, "step": 5668 }, { "epoch": 5.2, "learning_rate": 5.779984150495516e-06, "loss": 0.0426, "step": 5669 }, { "epoch": 5.2, "learning_rate": 5.7766202166014615e-06, "loss": 0.0479, "step": 5670 }, { "epoch": 5.2, "learning_rate": 5.77325686427922e-06, "loss": 0.061, "step": 5671 }, { "epoch": 5.2, "learning_rate": 5.76989409399193e-06, "loss": 0.052, "step": 5672 }, { "epoch": 5.2, "learning_rate": 5.766531906202664e-06, "loss": 0.0586, "step": 5673 }, { "epoch": 5.2, "learning_rate": 5.763170301374399e-06, "loss": 0.0668, "step": 5674 }, { "epoch": 5.2, "learning_rate": 5.75980927997004e-06, "loss": 0.0506, "step": 5675 }, { "epoch": 5.2, "learning_rate": 5.756448842452405e-06, "loss": 0.0566, "step": 5676 }, { "epoch": 5.2, "learning_rate": 5.753088989284241e-06, "loss": 0.0487, "step": 5677 }, { "epoch": 5.2, "learning_rate": 5.749729720928211e-06, "loss": 0.0455, "step": 5678 }, { "epoch": 5.21, "learning_rate": 5.746371037846889e-06, "loss": 0.0456, "step": 5679 }, { "epoch": 5.21, "learning_rate": 5.743012940502783e-06, "loss": 0.0447, "step": 5680 }, { "epoch": 5.21, "learning_rate": 5.7396554293583105e-06, "loss": 0.064, "step": 5681 }, { "epoch": 5.21, "learning_rate": 5.73629850487581e-06, "loss": 0.0445, "step": 5682 }, { "epoch": 5.21, "learning_rate": 5.732942167517537e-06, "loss": 0.0636, "step": 5683 }, { "epoch": 5.21, "learning_rate": 5.729586417745674e-06, "loss": 0.052, "step": 5684 }, { "epoch": 5.21, "learning_rate": 5.726231256022316e-06, "loss": 0.0478, "step": 5685 }, { "epoch": 5.21, "learning_rate": 5.722876682809476e-06, "loss": 0.0506, "step": 5686 }, { "epoch": 5.21, "learning_rate": 5.7195226985690935e-06, "loss": 0.0527, "step": 5687 }, { "epoch": 5.21, "learning_rate": 5.71616930376302e-06, "loss": 0.0604, "step": 5688 }, { "epoch": 5.21, "learning_rate": 5.712816498853026e-06, "loss": 0.0645, "step": 5689 }, { "epoch": 5.22, "learning_rate": 5.7094642843008e-06, "loss": 0.0487, "step": 5690 }, { "epoch": 5.22, "learning_rate": 5.706112660567959e-06, "loss": 0.0582, "step": 5691 }, { "epoch": 5.22, "learning_rate": 5.702761628116029e-06, "loss": 0.0486, "step": 5692 }, { "epoch": 5.22, "learning_rate": 5.6994111874064505e-06, "loss": 0.0546, "step": 5693 }, { "epoch": 5.22, "learning_rate": 5.696061338900601e-06, "loss": 0.0593, "step": 5694 }, { "epoch": 5.22, "learning_rate": 5.692712083059755e-06, "loss": 0.053, "step": 5695 }, { "epoch": 5.22, "learning_rate": 5.68936342034512e-06, "loss": 0.0604, "step": 5696 }, { "epoch": 5.22, "learning_rate": 5.686015351217808e-06, "loss": 0.0503, "step": 5697 }, { "epoch": 5.22, "learning_rate": 5.682667876138871e-06, "loss": 0.0416, "step": 5698 }, { "epoch": 5.22, "learning_rate": 5.6793209955692595e-06, "loss": 0.0524, "step": 5699 }, { "epoch": 5.22, "learning_rate": 5.675974709969847e-06, "loss": 0.055, "step": 5700 }, { "epoch": 5.23, "learning_rate": 5.672629019801428e-06, "loss": 0.0472, "step": 5701 }, { "epoch": 5.23, "learning_rate": 5.669283925524716e-06, "loss": 0.0411, "step": 5702 }, { "epoch": 5.23, "learning_rate": 5.66593942760034e-06, "loss": 0.0565, "step": 5703 }, { "epoch": 5.23, "learning_rate": 5.662595526488841e-06, "loss": 0.0587, "step": 5704 }, { "epoch": 5.23, "learning_rate": 5.659252222650695e-06, "loss": 0.0534, "step": 5705 }, { "epoch": 5.23, "learning_rate": 5.655909516546278e-06, "loss": 0.0567, "step": 5706 }, { "epoch": 5.23, "learning_rate": 5.652567408635887e-06, "loss": 0.0539, "step": 5707 }, { "epoch": 5.23, "learning_rate": 5.64922589937975e-06, "loss": 0.0542, "step": 5708 }, { "epoch": 5.23, "learning_rate": 5.645884989237993e-06, "loss": 0.0514, "step": 5709 }, { "epoch": 5.23, "learning_rate": 5.642544678670676e-06, "loss": 0.0483, "step": 5710 }, { "epoch": 5.23, "learning_rate": 5.639204968137768e-06, "loss": 0.0531, "step": 5711 }, { "epoch": 5.24, "learning_rate": 5.635865858099152e-06, "loss": 0.0546, "step": 5712 }, { "epoch": 5.24, "learning_rate": 5.63252734901464e-06, "loss": 0.0521, "step": 5713 }, { "epoch": 5.24, "learning_rate": 5.629189441343952e-06, "loss": 0.0534, "step": 5714 }, { "epoch": 5.24, "learning_rate": 5.625852135546724e-06, "loss": 0.052, "step": 5715 }, { "epoch": 5.24, "learning_rate": 5.6225154320825205e-06, "loss": 0.0523, "step": 5716 }, { "epoch": 5.24, "learning_rate": 5.619179331410809e-06, "loss": 0.0437, "step": 5717 }, { "epoch": 5.24, "learning_rate": 5.61584383399098e-06, "loss": 0.0529, "step": 5718 }, { "epoch": 5.24, "learning_rate": 5.612508940282348e-06, "loss": 0.0588, "step": 5719 }, { "epoch": 5.24, "learning_rate": 5.6091746507441334e-06, "loss": 0.0542, "step": 5720 }, { "epoch": 5.24, "learning_rate": 5.605840965835476e-06, "loss": 0.0478, "step": 5721 }, { "epoch": 5.24, "learning_rate": 5.602507886015434e-06, "loss": 0.052, "step": 5722 }, { "epoch": 5.25, "learning_rate": 5.599175411742986e-06, "loss": 0.0476, "step": 5723 }, { "epoch": 5.25, "learning_rate": 5.595843543477022e-06, "loss": 0.0501, "step": 5724 }, { "epoch": 5.25, "learning_rate": 5.59251228167635e-06, "loss": 0.0573, "step": 5725 }, { "epoch": 5.25, "learning_rate": 5.58918162679969e-06, "loss": 0.0442, "step": 5726 }, { "epoch": 5.25, "learning_rate": 5.58585157930569e-06, "loss": 0.0455, "step": 5727 }, { "epoch": 5.25, "learning_rate": 5.582522139652906e-06, "loss": 0.0557, "step": 5728 }, { "epoch": 5.25, "learning_rate": 5.579193308299805e-06, "loss": 0.0471, "step": 5729 }, { "epoch": 5.25, "learning_rate": 5.575865085704786e-06, "loss": 0.0533, "step": 5730 }, { "epoch": 5.25, "learning_rate": 5.572537472326152e-06, "loss": 0.0513, "step": 5731 }, { "epoch": 5.25, "learning_rate": 5.569210468622121e-06, "loss": 0.0521, "step": 5732 }, { "epoch": 5.25, "learning_rate": 5.565884075050833e-06, "loss": 0.056, "step": 5733 }, { "epoch": 5.26, "learning_rate": 5.562558292070347e-06, "loss": 0.0463, "step": 5734 }, { "epoch": 5.26, "learning_rate": 5.55923312013863e-06, "loss": 0.0523, "step": 5735 }, { "epoch": 5.26, "learning_rate": 5.555908559713561e-06, "loss": 0.056, "step": 5736 }, { "epoch": 5.26, "learning_rate": 5.552584611252954e-06, "loss": 0.055, "step": 5737 }, { "epoch": 5.26, "learning_rate": 5.54926127521452e-06, "loss": 0.0519, "step": 5738 }, { "epoch": 5.26, "learning_rate": 5.545938552055894e-06, "loss": 0.057, "step": 5739 }, { "epoch": 5.26, "learning_rate": 5.542616442234618e-06, "loss": 0.0543, "step": 5740 }, { "epoch": 5.26, "learning_rate": 5.5392949462081665e-06, "loss": 0.0498, "step": 5741 }, { "epoch": 5.26, "learning_rate": 5.5359740644339155e-06, "loss": 0.0559, "step": 5742 }, { "epoch": 5.26, "learning_rate": 5.532653797369154e-06, "loss": 0.0555, "step": 5743 }, { "epoch": 5.26, "learning_rate": 5.529334145471101e-06, "loss": 0.062, "step": 5744 }, { "epoch": 5.27, "learning_rate": 5.5260151091968816e-06, "loss": 0.0643, "step": 5745 }, { "epoch": 5.27, "learning_rate": 5.522696689003533e-06, "loss": 0.0578, "step": 5746 }, { "epoch": 5.27, "learning_rate": 5.519378885348009e-06, "loss": 0.0516, "step": 5747 }, { "epoch": 5.27, "learning_rate": 5.5160616986871896e-06, "loss": 0.0468, "step": 5748 }, { "epoch": 5.27, "learning_rate": 5.512745129477855e-06, "loss": 0.0548, "step": 5749 }, { "epoch": 5.27, "learning_rate": 5.509429178176704e-06, "loss": 0.0553, "step": 5750 }, { "epoch": 5.27, "learning_rate": 5.5061138452403615e-06, "loss": 0.0621, "step": 5751 }, { "epoch": 5.27, "learning_rate": 5.5027991311253494e-06, "loss": 0.0517, "step": 5752 }, { "epoch": 5.27, "learning_rate": 5.499485036288122e-06, "loss": 0.0498, "step": 5753 }, { "epoch": 5.27, "learning_rate": 5.496171561185035e-06, "loss": 0.0614, "step": 5754 }, { "epoch": 5.27, "learning_rate": 5.49285870627236e-06, "loss": 0.0539, "step": 5755 }, { "epoch": 5.28, "learning_rate": 5.489546472006297e-06, "loss": 0.055, "step": 5756 }, { "epoch": 5.28, "learning_rate": 5.4862348588429444e-06, "loss": 0.0502, "step": 5757 }, { "epoch": 5.28, "learning_rate": 5.482923867238317e-06, "loss": 0.051, "step": 5758 }, { "epoch": 5.28, "learning_rate": 5.479613497648357e-06, "loss": 0.0568, "step": 5759 }, { "epoch": 5.28, "learning_rate": 5.476303750528907e-06, "loss": 0.0506, "step": 5760 }, { "epoch": 5.28, "learning_rate": 5.4729946263357255e-06, "loss": 0.0479, "step": 5761 }, { "epoch": 5.28, "learning_rate": 5.469686125524497e-06, "loss": 0.0436, "step": 5762 }, { "epoch": 5.28, "learning_rate": 5.4663782485508075e-06, "loss": 0.0564, "step": 5763 }, { "epoch": 5.28, "learning_rate": 5.46307099587016e-06, "loss": 0.0527, "step": 5764 }, { "epoch": 5.28, "learning_rate": 5.459764367937972e-06, "loss": 0.0462, "step": 5765 }, { "epoch": 5.29, "learning_rate": 5.45645836520958e-06, "loss": 0.0609, "step": 5766 }, { "epoch": 5.29, "learning_rate": 5.453152988140229e-06, "loss": 0.0508, "step": 5767 }, { "epoch": 5.29, "learning_rate": 5.449848237185076e-06, "loss": 0.0478, "step": 5768 }, { "epoch": 5.29, "learning_rate": 5.4465441127991995e-06, "loss": 0.0478, "step": 5769 }, { "epoch": 5.29, "learning_rate": 5.443240615437586e-06, "loss": 0.0571, "step": 5770 }, { "epoch": 5.29, "learning_rate": 5.439937745555135e-06, "loss": 0.032, "step": 5771 }, { "epoch": 5.29, "learning_rate": 5.43663550360666e-06, "loss": 0.0405, "step": 5772 }, { "epoch": 5.29, "learning_rate": 5.4333338900468935e-06, "loss": 0.0472, "step": 5773 }, { "epoch": 5.29, "learning_rate": 5.430032905330477e-06, "loss": 0.0517, "step": 5774 }, { "epoch": 5.29, "learning_rate": 5.426732549911959e-06, "loss": 0.0468, "step": 5775 }, { "epoch": 5.29, "learning_rate": 5.423432824245819e-06, "loss": 0.054, "step": 5776 }, { "epoch": 5.3, "learning_rate": 5.420133728786432e-06, "loss": 0.0604, "step": 5777 }, { "epoch": 5.3, "learning_rate": 5.416835263988096e-06, "loss": 0.0584, "step": 5778 }, { "epoch": 5.3, "learning_rate": 5.4135374303050136e-06, "loss": 0.0459, "step": 5779 }, { "epoch": 5.3, "learning_rate": 5.4102402281913135e-06, "loss": 0.06, "step": 5780 }, { "epoch": 5.3, "learning_rate": 5.406943658101027e-06, "loss": 0.0558, "step": 5781 }, { "epoch": 5.3, "learning_rate": 5.403647720488099e-06, "loss": 0.0503, "step": 5782 }, { "epoch": 5.3, "learning_rate": 5.4003524158063956e-06, "loss": 0.0582, "step": 5783 }, { "epoch": 5.3, "learning_rate": 5.397057744509686e-06, "loss": 0.056, "step": 5784 }, { "epoch": 5.3, "learning_rate": 5.393763707051658e-06, "loss": 0.0563, "step": 5785 }, { "epoch": 5.3, "learning_rate": 5.390470303885904e-06, "loss": 0.0601, "step": 5786 }, { "epoch": 5.3, "learning_rate": 5.387177535465945e-06, "loss": 0.0557, "step": 5787 }, { "epoch": 5.31, "learning_rate": 5.3838854022452e-06, "loss": 0.0511, "step": 5788 }, { "epoch": 5.31, "learning_rate": 5.380593904677006e-06, "loss": 0.0517, "step": 5789 }, { "epoch": 5.31, "learning_rate": 5.377303043214608e-06, "loss": 0.0529, "step": 5790 }, { "epoch": 5.31, "learning_rate": 5.374012818311174e-06, "loss": 0.0617, "step": 5791 }, { "epoch": 5.31, "learning_rate": 5.370723230419774e-06, "loss": 0.0555, "step": 5792 }, { "epoch": 5.31, "learning_rate": 5.367434279993392e-06, "loss": 0.0477, "step": 5793 }, { "epoch": 5.31, "learning_rate": 5.364145967484932e-06, "loss": 0.0579, "step": 5794 }, { "epoch": 5.31, "learning_rate": 5.360858293347196e-06, "loss": 0.0538, "step": 5795 }, { "epoch": 5.31, "learning_rate": 5.3575712580329164e-06, "loss": 0.0536, "step": 5796 }, { "epoch": 5.31, "learning_rate": 5.354284861994723e-06, "loss": 0.048, "step": 5797 }, { "epoch": 5.31, "learning_rate": 5.350999105685156e-06, "loss": 0.0448, "step": 5798 }, { "epoch": 5.32, "learning_rate": 5.347713989556684e-06, "loss": 0.0562, "step": 5799 }, { "epoch": 5.32, "learning_rate": 5.344429514061668e-06, "loss": 0.0553, "step": 5800 }, { "epoch": 5.32, "learning_rate": 5.3411456796524e-06, "loss": 0.0559, "step": 5801 }, { "epoch": 5.32, "learning_rate": 5.337862486781068e-06, "loss": 0.0516, "step": 5802 }, { "epoch": 5.32, "learning_rate": 5.334579935899777e-06, "loss": 0.0468, "step": 5803 }, { "epoch": 5.32, "learning_rate": 5.33129802746054e-06, "loss": 0.0571, "step": 5804 }, { "epoch": 5.32, "learning_rate": 5.328016761915293e-06, "loss": 0.0494, "step": 5805 }, { "epoch": 5.32, "learning_rate": 5.324736139715875e-06, "loss": 0.0464, "step": 5806 }, { "epoch": 5.32, "learning_rate": 5.321456161314029e-06, "loss": 0.0509, "step": 5807 }, { "epoch": 5.32, "learning_rate": 5.3181768271614285e-06, "loss": 0.0491, "step": 5808 }, { "epoch": 5.32, "learning_rate": 5.314898137709643e-06, "loss": 0.0478, "step": 5809 }, { "epoch": 5.33, "learning_rate": 5.311620093410157e-06, "loss": 0.0531, "step": 5810 }, { "epoch": 5.33, "learning_rate": 5.308342694714363e-06, "loss": 0.0521, "step": 5811 }, { "epoch": 5.33, "learning_rate": 5.305065942073576e-06, "loss": 0.0508, "step": 5812 }, { "epoch": 5.33, "learning_rate": 5.301789835939012e-06, "loss": 0.0521, "step": 5813 }, { "epoch": 5.33, "learning_rate": 5.298514376761798e-06, "loss": 0.0497, "step": 5814 }, { "epoch": 5.33, "learning_rate": 5.295239564992975e-06, "loss": 0.0564, "step": 5815 }, { "epoch": 5.33, "learning_rate": 5.291965401083497e-06, "loss": 0.0541, "step": 5816 }, { "epoch": 5.33, "learning_rate": 5.2886918854842224e-06, "loss": 0.0514, "step": 5817 }, { "epoch": 5.33, "learning_rate": 5.285419018645926e-06, "loss": 0.049, "step": 5818 }, { "epoch": 5.33, "learning_rate": 5.28214680101929e-06, "loss": 0.0545, "step": 5819 }, { "epoch": 5.33, "learning_rate": 5.278875233054912e-06, "loss": 0.0531, "step": 5820 }, { "epoch": 5.34, "learning_rate": 5.2756043152032934e-06, "loss": 0.0441, "step": 5821 }, { "epoch": 5.34, "learning_rate": 5.2723340479148446e-06, "loss": 0.0499, "step": 5822 }, { "epoch": 5.34, "learning_rate": 5.269064431639901e-06, "loss": 0.0462, "step": 5823 }, { "epoch": 5.34, "learning_rate": 5.265795466828693e-06, "loss": 0.0635, "step": 5824 }, { "epoch": 5.34, "learning_rate": 5.2625271539313624e-06, "loss": 0.0628, "step": 5825 }, { "epoch": 5.34, "learning_rate": 5.259259493397974e-06, "loss": 0.0483, "step": 5826 }, { "epoch": 5.34, "learning_rate": 5.255992485678492e-06, "loss": 0.0529, "step": 5827 }, { "epoch": 5.34, "learning_rate": 5.25272613122279e-06, "loss": 0.0542, "step": 5828 }, { "epoch": 5.34, "learning_rate": 5.249460430480651e-06, "loss": 0.0488, "step": 5829 }, { "epoch": 5.34, "learning_rate": 5.246195383901782e-06, "loss": 0.0558, "step": 5830 }, { "epoch": 5.34, "learning_rate": 5.242930991935783e-06, "loss": 0.0501, "step": 5831 }, { "epoch": 5.35, "learning_rate": 5.239667255032168e-06, "loss": 0.0573, "step": 5832 }, { "epoch": 5.35, "learning_rate": 5.236404173640371e-06, "loss": 0.0506, "step": 5833 }, { "epoch": 5.35, "learning_rate": 5.233141748209721e-06, "loss": 0.0589, "step": 5834 }, { "epoch": 5.35, "learning_rate": 5.2298799791894686e-06, "loss": 0.0555, "step": 5835 }, { "epoch": 5.35, "learning_rate": 5.226618867028761e-06, "loss": 0.0577, "step": 5836 }, { "epoch": 5.35, "learning_rate": 5.2233584121766734e-06, "loss": 0.0487, "step": 5837 }, { "epoch": 5.35, "learning_rate": 5.22009861508217e-06, "loss": 0.0458, "step": 5838 }, { "epoch": 5.35, "learning_rate": 5.216839476194144e-06, "loss": 0.0427, "step": 5839 }, { "epoch": 5.35, "learning_rate": 5.2135809959613825e-06, "loss": 0.0565, "step": 5840 }, { "epoch": 5.35, "learning_rate": 5.210323174832586e-06, "loss": 0.0567, "step": 5841 }, { "epoch": 5.35, "learning_rate": 5.207066013256375e-06, "loss": 0.0497, "step": 5842 }, { "epoch": 5.36, "learning_rate": 5.203809511681261e-06, "loss": 0.0429, "step": 5843 }, { "epoch": 5.36, "learning_rate": 5.2005536705556745e-06, "loss": 0.0642, "step": 5844 }, { "epoch": 5.36, "learning_rate": 5.197298490327962e-06, "loss": 0.0464, "step": 5845 }, { "epoch": 5.36, "learning_rate": 5.194043971446366e-06, "loss": 0.0543, "step": 5846 }, { "epoch": 5.36, "learning_rate": 5.19079011435904e-06, "loss": 0.0432, "step": 5847 }, { "epoch": 5.36, "learning_rate": 5.187536919514058e-06, "loss": 0.0545, "step": 5848 }, { "epoch": 5.36, "learning_rate": 5.184284387359388e-06, "loss": 0.0482, "step": 5849 }, { "epoch": 5.36, "learning_rate": 5.181032518342913e-06, "loss": 0.0484, "step": 5850 }, { "epoch": 5.36, "learning_rate": 5.177781312912431e-06, "loss": 0.056, "step": 5851 }, { "epoch": 5.36, "learning_rate": 5.174530771515638e-06, "loss": 0.0458, "step": 5852 }, { "epoch": 5.36, "learning_rate": 5.1712808946001435e-06, "loss": 0.0426, "step": 5853 }, { "epoch": 5.37, "learning_rate": 5.168031682613463e-06, "loss": 0.0561, "step": 5854 }, { "epoch": 5.37, "learning_rate": 5.164783136003027e-06, "loss": 0.0572, "step": 5855 }, { "epoch": 5.37, "learning_rate": 5.1615352552161706e-06, "loss": 0.0505, "step": 5856 }, { "epoch": 5.37, "learning_rate": 5.158288040700127e-06, "loss": 0.054, "step": 5857 }, { "epoch": 5.37, "learning_rate": 5.155041492902059e-06, "loss": 0.0427, "step": 5858 }, { "epoch": 5.37, "learning_rate": 5.15179561226902e-06, "loss": 0.0403, "step": 5859 }, { "epoch": 5.37, "learning_rate": 5.14855039924798e-06, "loss": 0.0535, "step": 5860 }, { "epoch": 5.37, "learning_rate": 5.145305854285807e-06, "loss": 0.0499, "step": 5861 }, { "epoch": 5.37, "learning_rate": 5.1420619778292934e-06, "loss": 0.06, "step": 5862 }, { "epoch": 5.37, "learning_rate": 5.138818770325129e-06, "loss": 0.0564, "step": 5863 }, { "epoch": 5.37, "learning_rate": 5.135576232219906e-06, "loss": 0.0509, "step": 5864 }, { "epoch": 5.38, "learning_rate": 5.132334363960138e-06, "loss": 0.0456, "step": 5865 }, { "epoch": 5.38, "learning_rate": 5.129093165992241e-06, "loss": 0.0624, "step": 5866 }, { "epoch": 5.38, "learning_rate": 5.125852638762534e-06, "loss": 0.0546, "step": 5867 }, { "epoch": 5.38, "learning_rate": 5.1226127827172445e-06, "loss": 0.0467, "step": 5868 }, { "epoch": 5.38, "learning_rate": 5.119373598302515e-06, "loss": 0.0498, "step": 5869 }, { "epoch": 5.38, "learning_rate": 5.116135085964391e-06, "loss": 0.0538, "step": 5870 }, { "epoch": 5.38, "learning_rate": 5.1128972461488205e-06, "loss": 0.0483, "step": 5871 }, { "epoch": 5.38, "learning_rate": 5.109660079301668e-06, "loss": 0.0476, "step": 5872 }, { "epoch": 5.38, "learning_rate": 5.1064235858687e-06, "loss": 0.0464, "step": 5873 }, { "epoch": 5.38, "learning_rate": 5.103187766295592e-06, "loss": 0.0475, "step": 5874 }, { "epoch": 5.38, "learning_rate": 5.099952621027919e-06, "loss": 0.051, "step": 5875 }, { "epoch": 5.39, "learning_rate": 5.09671815051118e-06, "loss": 0.0593, "step": 5876 }, { "epoch": 5.39, "learning_rate": 5.093484355190767e-06, "loss": 0.0516, "step": 5877 }, { "epoch": 5.39, "learning_rate": 5.09025123551198e-06, "loss": 0.0566, "step": 5878 }, { "epoch": 5.39, "learning_rate": 5.087018791920031e-06, "loss": 0.0545, "step": 5879 }, { "epoch": 5.39, "learning_rate": 5.08378702486004e-06, "loss": 0.0553, "step": 5880 }, { "epoch": 5.39, "learning_rate": 5.080555934777028e-06, "loss": 0.0475, "step": 5881 }, { "epoch": 5.39, "learning_rate": 5.077325522115924e-06, "loss": 0.051, "step": 5882 }, { "epoch": 5.39, "learning_rate": 5.074095787321569e-06, "loss": 0.0571, "step": 5883 }, { "epoch": 5.39, "learning_rate": 5.070866730838703e-06, "loss": 0.0557, "step": 5884 }, { "epoch": 5.39, "learning_rate": 5.067638353111982e-06, "loss": 0.055, "step": 5885 }, { "epoch": 5.4, "learning_rate": 5.0644106545859605e-06, "loss": 0.0523, "step": 5886 }, { "epoch": 5.4, "learning_rate": 5.061183635705098e-06, "loss": 0.0584, "step": 5887 }, { "epoch": 5.4, "learning_rate": 5.057957296913771e-06, "loss": 0.0524, "step": 5888 }, { "epoch": 5.4, "learning_rate": 5.0547316386562505e-06, "loss": 0.0518, "step": 5889 }, { "epoch": 5.4, "learning_rate": 5.051506661376725e-06, "loss": 0.0441, "step": 5890 }, { "epoch": 5.4, "learning_rate": 5.048282365519279e-06, "loss": 0.0477, "step": 5891 }, { "epoch": 5.4, "learning_rate": 5.045058751527909e-06, "loss": 0.0569, "step": 5892 }, { "epoch": 5.4, "learning_rate": 5.041835819846512e-06, "loss": 0.0569, "step": 5893 }, { "epoch": 5.4, "learning_rate": 5.038613570918903e-06, "loss": 0.0516, "step": 5894 }, { "epoch": 5.4, "learning_rate": 5.035392005188791e-06, "loss": 0.0526, "step": 5895 }, { "epoch": 5.4, "learning_rate": 5.032171123099789e-06, "loss": 0.0573, "step": 5896 }, { "epoch": 5.41, "learning_rate": 5.028950925095435e-06, "loss": 0.0487, "step": 5897 }, { "epoch": 5.41, "learning_rate": 5.02573141161915e-06, "loss": 0.0482, "step": 5898 }, { "epoch": 5.41, "learning_rate": 5.022512583114274e-06, "loss": 0.0638, "step": 5899 }, { "epoch": 5.41, "learning_rate": 5.019294440024045e-06, "loss": 0.06, "step": 5900 }, { "epoch": 5.41, "learning_rate": 5.016076982791618e-06, "loss": 0.0531, "step": 5901 }, { "epoch": 5.41, "learning_rate": 5.01286021186004e-06, "loss": 0.0574, "step": 5902 }, { "epoch": 5.41, "learning_rate": 5.009644127672275e-06, "loss": 0.0574, "step": 5903 }, { "epoch": 5.41, "learning_rate": 5.006428730671179e-06, "loss": 0.0596, "step": 5904 }, { "epoch": 5.41, "learning_rate": 5.003214021299532e-06, "loss": 0.0566, "step": 5905 }, { "epoch": 5.41, "learning_rate": 5.000000000000003e-06, "loss": 0.0556, "step": 5906 }, { "epoch": 5.41, "learning_rate": 4.996786667215169e-06, "loss": 0.0516, "step": 5907 }, { "epoch": 5.42, "learning_rate": 4.993574023387524e-06, "loss": 0.0557, "step": 5908 }, { "epoch": 5.42, "learning_rate": 4.9903620689594534e-06, "loss": 0.0484, "step": 5909 }, { "epoch": 5.42, "learning_rate": 4.9871508043732545e-06, "loss": 0.0479, "step": 5910 }, { "epoch": 5.42, "learning_rate": 4.983940230071121e-06, "loss": 0.0448, "step": 5911 }, { "epoch": 5.42, "learning_rate": 4.98073034649517e-06, "loss": 0.0436, "step": 5912 }, { "epoch": 5.42, "learning_rate": 4.977521154087405e-06, "loss": 0.0575, "step": 5913 }, { "epoch": 5.42, "learning_rate": 4.974312653289739e-06, "loss": 0.0462, "step": 5914 }, { "epoch": 5.42, "learning_rate": 4.971104844543998e-06, "loss": 0.0538, "step": 5915 }, { "epoch": 5.42, "learning_rate": 4.967897728291905e-06, "loss": 0.0498, "step": 5916 }, { "epoch": 5.42, "learning_rate": 4.964691304975089e-06, "loss": 0.0495, "step": 5917 }, { "epoch": 5.42, "learning_rate": 4.961485575035078e-06, "loss": 0.0543, "step": 5918 }, { "epoch": 5.43, "learning_rate": 4.95828053891332e-06, "loss": 0.0537, "step": 5919 }, { "epoch": 5.43, "learning_rate": 4.955076197051154e-06, "loss": 0.0523, "step": 5920 }, { "epoch": 5.43, "learning_rate": 4.951872549889823e-06, "loss": 0.0465, "step": 5921 }, { "epoch": 5.43, "learning_rate": 4.948669597870486e-06, "loss": 0.0579, "step": 5922 }, { "epoch": 5.43, "learning_rate": 4.9454673414341945e-06, "loss": 0.0485, "step": 5923 }, { "epoch": 5.43, "learning_rate": 4.9422657810219114e-06, "loss": 0.0638, "step": 5924 }, { "epoch": 5.43, "learning_rate": 4.939064917074494e-06, "loss": 0.0492, "step": 5925 }, { "epoch": 5.43, "learning_rate": 4.93586475003272e-06, "loss": 0.048, "step": 5926 }, { "epoch": 5.43, "learning_rate": 4.9326652803372534e-06, "loss": 0.0481, "step": 5927 }, { "epoch": 5.43, "learning_rate": 4.929466508428678e-06, "loss": 0.0538, "step": 5928 }, { "epoch": 5.43, "learning_rate": 4.926268434747473e-06, "loss": 0.0588, "step": 5929 }, { "epoch": 5.44, "learning_rate": 4.923071059734015e-06, "loss": 0.0652, "step": 5930 }, { "epoch": 5.44, "learning_rate": 4.9198743838286015e-06, "loss": 0.0527, "step": 5931 }, { "epoch": 5.44, "learning_rate": 4.916678407471417e-06, "loss": 0.0579, "step": 5932 }, { "epoch": 5.44, "learning_rate": 4.9134831311025625e-06, "loss": 0.0473, "step": 5933 }, { "epoch": 5.44, "learning_rate": 4.910288555162035e-06, "loss": 0.0515, "step": 5934 }, { "epoch": 5.44, "learning_rate": 4.907094680089736e-06, "loss": 0.0529, "step": 5935 }, { "epoch": 5.44, "learning_rate": 4.903901506325468e-06, "loss": 0.0498, "step": 5936 }, { "epoch": 5.44, "learning_rate": 4.900709034308948e-06, "loss": 0.0456, "step": 5937 }, { "epoch": 5.44, "learning_rate": 4.897517264479785e-06, "loss": 0.0534, "step": 5938 }, { "epoch": 5.44, "learning_rate": 4.894326197277491e-06, "loss": 0.048, "step": 5939 }, { "epoch": 5.44, "learning_rate": 4.891135833141495e-06, "loss": 0.0525, "step": 5940 }, { "epoch": 5.45, "learning_rate": 4.887946172511112e-06, "loss": 0.0445, "step": 5941 }, { "epoch": 5.45, "learning_rate": 4.8847572158255705e-06, "loss": 0.0492, "step": 5942 }, { "epoch": 5.45, "learning_rate": 4.881568963523993e-06, "loss": 0.0527, "step": 5943 }, { "epoch": 5.45, "learning_rate": 4.878381416045422e-06, "loss": 0.0543, "step": 5944 }, { "epoch": 5.45, "learning_rate": 4.875194573828786e-06, "loss": 0.0497, "step": 5945 }, { "epoch": 5.45, "learning_rate": 4.87200843731292e-06, "loss": 0.0555, "step": 5946 }, { "epoch": 5.45, "learning_rate": 4.868823006936572e-06, "loss": 0.0461, "step": 5947 }, { "epoch": 5.45, "learning_rate": 4.8656382831383795e-06, "loss": 0.0518, "step": 5948 }, { "epoch": 5.45, "learning_rate": 4.86245426635689e-06, "loss": 0.0512, "step": 5949 }, { "epoch": 5.45, "learning_rate": 4.859270957030547e-06, "loss": 0.0413, "step": 5950 }, { "epoch": 5.45, "learning_rate": 4.85608835559771e-06, "loss": 0.0505, "step": 5951 }, { "epoch": 5.46, "learning_rate": 4.8529064624966295e-06, "loss": 0.0601, "step": 5952 }, { "epoch": 5.46, "learning_rate": 4.849725278165457e-06, "loss": 0.0573, "step": 5953 }, { "epoch": 5.46, "learning_rate": 4.846544803042258e-06, "loss": 0.0492, "step": 5954 }, { "epoch": 5.46, "learning_rate": 4.843365037564989e-06, "loss": 0.0514, "step": 5955 }, { "epoch": 5.46, "learning_rate": 4.840185982171515e-06, "loss": 0.0442, "step": 5956 }, { "epoch": 5.46, "learning_rate": 4.837007637299596e-06, "loss": 0.0556, "step": 5957 }, { "epoch": 5.46, "learning_rate": 4.833830003386907e-06, "loss": 0.0544, "step": 5958 }, { "epoch": 5.46, "learning_rate": 4.830653080871015e-06, "loss": 0.0669, "step": 5959 }, { "epoch": 5.46, "learning_rate": 4.827476870189387e-06, "loss": 0.0484, "step": 5960 }, { "epoch": 5.46, "learning_rate": 4.824301371779403e-06, "loss": 0.0493, "step": 5961 }, { "epoch": 5.46, "learning_rate": 4.821126586078336e-06, "loss": 0.0604, "step": 5962 }, { "epoch": 5.47, "learning_rate": 4.817952513523364e-06, "loss": 0.0565, "step": 5963 }, { "epoch": 5.47, "learning_rate": 4.814779154551561e-06, "loss": 0.0586, "step": 5964 }, { "epoch": 5.47, "learning_rate": 4.811606509599917e-06, "loss": 0.0666, "step": 5965 }, { "epoch": 5.47, "learning_rate": 4.808434579105308e-06, "loss": 0.0546, "step": 5966 }, { "epoch": 5.47, "learning_rate": 4.80526336350452e-06, "loss": 0.0535, "step": 5967 }, { "epoch": 5.47, "learning_rate": 4.8020928632342345e-06, "loss": 0.0498, "step": 5968 }, { "epoch": 5.47, "learning_rate": 4.798923078731047e-06, "loss": 0.0478, "step": 5969 }, { "epoch": 5.47, "learning_rate": 4.7957540104314385e-06, "loss": 0.0567, "step": 5970 }, { "epoch": 5.47, "learning_rate": 4.7925856587718045e-06, "loss": 0.0587, "step": 5971 }, { "epoch": 5.47, "learning_rate": 4.789418024188436e-06, "loss": 0.0483, "step": 5972 }, { "epoch": 5.47, "learning_rate": 4.786251107117519e-06, "loss": 0.0476, "step": 5973 }, { "epoch": 5.48, "learning_rate": 4.783084907995156e-06, "loss": 0.0562, "step": 5974 }, { "epoch": 5.48, "learning_rate": 4.7799194272573346e-06, "loss": 0.0548, "step": 5975 }, { "epoch": 5.48, "learning_rate": 4.776754665339957e-06, "loss": 0.0514, "step": 5976 }, { "epoch": 5.48, "learning_rate": 4.7735906226788185e-06, "loss": 0.0456, "step": 5977 }, { "epoch": 5.48, "learning_rate": 4.770427299709611e-06, "loss": 0.0537, "step": 5978 }, { "epoch": 5.48, "learning_rate": 4.7672646968679436e-06, "loss": 0.0468, "step": 5979 }, { "epoch": 5.48, "learning_rate": 4.76410281458931e-06, "loss": 0.0453, "step": 5980 }, { "epoch": 5.48, "learning_rate": 4.760941653309111e-06, "loss": 0.0519, "step": 5981 }, { "epoch": 5.48, "learning_rate": 4.757781213462646e-06, "loss": 0.0518, "step": 5982 }, { "epoch": 5.48, "learning_rate": 4.754621495485121e-06, "loss": 0.0537, "step": 5983 }, { "epoch": 5.48, "learning_rate": 4.751462499811638e-06, "loss": 0.0461, "step": 5984 }, { "epoch": 5.49, "learning_rate": 4.748304226877194e-06, "loss": 0.0519, "step": 5985 }, { "epoch": 5.49, "learning_rate": 4.745146677116701e-06, "loss": 0.0531, "step": 5986 }, { "epoch": 5.49, "learning_rate": 4.7419898509649585e-06, "loss": 0.0536, "step": 5987 }, { "epoch": 5.49, "learning_rate": 4.7388337488566725e-06, "loss": 0.0566, "step": 5988 }, { "epoch": 5.49, "learning_rate": 4.7356783712264405e-06, "loss": 0.052, "step": 5989 }, { "epoch": 5.49, "learning_rate": 4.7325237185087795e-06, "loss": 0.0462, "step": 5990 }, { "epoch": 5.49, "learning_rate": 4.729369791138085e-06, "loss": 0.0519, "step": 5991 }, { "epoch": 5.49, "learning_rate": 4.726216589548667e-06, "loss": 0.0539, "step": 5992 }, { "epoch": 5.49, "learning_rate": 4.723064114174723e-06, "loss": 0.0596, "step": 5993 }, { "epoch": 5.49, "learning_rate": 4.71991236545037e-06, "loss": 0.0493, "step": 5994 }, { "epoch": 5.49, "learning_rate": 4.716761343809605e-06, "loss": 0.0596, "step": 5995 }, { "epoch": 5.5, "learning_rate": 4.713611049686331e-06, "loss": 0.0592, "step": 5996 }, { "epoch": 5.5, "learning_rate": 4.710461483514362e-06, "loss": 0.0484, "step": 5997 }, { "epoch": 5.5, "learning_rate": 4.7073126457273945e-06, "loss": 0.0494, "step": 5998 }, { "epoch": 5.5, "learning_rate": 4.704164536759037e-06, "loss": 0.0561, "step": 5999 }, { "epoch": 5.5, "learning_rate": 4.701017157042788e-06, "loss": 0.0543, "step": 6000 }, { "epoch": 5.5, "learning_rate": 4.697870507012057e-06, "loss": 0.0473, "step": 6001 }, { "epoch": 5.5, "learning_rate": 4.694724587100146e-06, "loss": 0.0509, "step": 6002 }, { "epoch": 5.5, "learning_rate": 4.69157939774025e-06, "loss": 0.046, "step": 6003 }, { "epoch": 5.5, "learning_rate": 4.6884349393654825e-06, "loss": 0.0495, "step": 6004 }, { "epoch": 5.5, "learning_rate": 4.685291212408839e-06, "loss": 0.0548, "step": 6005 }, { "epoch": 5.51, "learning_rate": 4.682148217303219e-06, "loss": 0.0613, "step": 6006 }, { "epoch": 5.51, "learning_rate": 4.6790059544814194e-06, "loss": 0.0484, "step": 6007 }, { "epoch": 5.51, "learning_rate": 4.675864424376146e-06, "loss": 0.0535, "step": 6008 }, { "epoch": 5.51, "learning_rate": 4.672723627419995e-06, "loss": 0.0586, "step": 6009 }, { "epoch": 5.51, "learning_rate": 4.669583564045457e-06, "loss": 0.0521, "step": 6010 }, { "epoch": 5.51, "learning_rate": 4.666444234684936e-06, "loss": 0.0561, "step": 6011 }, { "epoch": 5.51, "learning_rate": 4.663305639770724e-06, "loss": 0.0516, "step": 6012 }, { "epoch": 5.51, "learning_rate": 4.660167779735011e-06, "loss": 0.0577, "step": 6013 }, { "epoch": 5.51, "learning_rate": 4.657030655009896e-06, "loss": 0.0407, "step": 6014 }, { "epoch": 5.51, "learning_rate": 4.653894266027367e-06, "loss": 0.0446, "step": 6015 }, { "epoch": 5.51, "learning_rate": 4.650758613219311e-06, "loss": 0.0475, "step": 6016 }, { "epoch": 5.52, "learning_rate": 4.6476236970175235e-06, "loss": 0.0549, "step": 6017 }, { "epoch": 5.52, "learning_rate": 4.644489517853688e-06, "loss": 0.0465, "step": 6018 }, { "epoch": 5.52, "learning_rate": 4.641356076159387e-06, "loss": 0.0413, "step": 6019 }, { "epoch": 5.52, "learning_rate": 4.638223372366111e-06, "loss": 0.0506, "step": 6020 }, { "epoch": 5.52, "learning_rate": 4.635091406905235e-06, "loss": 0.0571, "step": 6021 }, { "epoch": 5.52, "learning_rate": 4.63196018020805e-06, "loss": 0.053, "step": 6022 }, { "epoch": 5.52, "learning_rate": 4.628829692705726e-06, "loss": 0.0495, "step": 6023 }, { "epoch": 5.52, "learning_rate": 4.625699944829346e-06, "loss": 0.0555, "step": 6024 }, { "epoch": 5.52, "learning_rate": 4.622570937009879e-06, "loss": 0.0534, "step": 6025 }, { "epoch": 5.52, "learning_rate": 4.619442669678206e-06, "loss": 0.0508, "step": 6026 }, { "epoch": 5.52, "learning_rate": 4.616315143265096e-06, "loss": 0.0417, "step": 6027 }, { "epoch": 5.53, "learning_rate": 4.613188358201212e-06, "loss": 0.0604, "step": 6028 }, { "epoch": 5.53, "learning_rate": 4.610062314917134e-06, "loss": 0.0559, "step": 6029 }, { "epoch": 5.53, "learning_rate": 4.606937013843319e-06, "loss": 0.0494, "step": 6030 }, { "epoch": 5.53, "learning_rate": 4.603812455410132e-06, "loss": 0.0563, "step": 6031 }, { "epoch": 5.53, "learning_rate": 4.600688640047831e-06, "loss": 0.0499, "step": 6032 }, { "epoch": 5.53, "learning_rate": 4.5975655681865805e-06, "loss": 0.0565, "step": 6033 }, { "epoch": 5.53, "learning_rate": 4.594443240256433e-06, "loss": 0.0572, "step": 6034 }, { "epoch": 5.53, "learning_rate": 4.5913216566873395e-06, "loss": 0.0439, "step": 6035 }, { "epoch": 5.53, "learning_rate": 4.588200817909157e-06, "loss": 0.0511, "step": 6036 }, { "epoch": 5.53, "learning_rate": 4.585080724351633e-06, "loss": 0.0508, "step": 6037 }, { "epoch": 5.53, "learning_rate": 4.581961376444413e-06, "loss": 0.043, "step": 6038 }, { "epoch": 5.54, "learning_rate": 4.5788427746170335e-06, "loss": 0.0489, "step": 6039 }, { "epoch": 5.54, "learning_rate": 4.575724919298946e-06, "loss": 0.0515, "step": 6040 }, { "epoch": 5.54, "learning_rate": 4.572607810919484e-06, "loss": 0.0559, "step": 6041 }, { "epoch": 5.54, "learning_rate": 4.569491449907878e-06, "loss": 0.0562, "step": 6042 }, { "epoch": 5.54, "learning_rate": 4.566375836693268e-06, "loss": 0.0501, "step": 6043 }, { "epoch": 5.54, "learning_rate": 4.563260971704678e-06, "loss": 0.0459, "step": 6044 }, { "epoch": 5.54, "learning_rate": 4.560146855371036e-06, "loss": 0.0569, "step": 6045 }, { "epoch": 5.54, "learning_rate": 4.5570334881211605e-06, "loss": 0.0517, "step": 6046 }, { "epoch": 5.54, "learning_rate": 4.553920870383779e-06, "loss": 0.0558, "step": 6047 }, { "epoch": 5.54, "learning_rate": 4.550809002587504e-06, "loss": 0.0553, "step": 6048 }, { "epoch": 5.54, "learning_rate": 4.547697885160845e-06, "loss": 0.054, "step": 6049 }, { "epoch": 5.55, "learning_rate": 4.544587518532221e-06, "loss": 0.0531, "step": 6050 }, { "epoch": 5.55, "learning_rate": 4.541477903129933e-06, "loss": 0.0446, "step": 6051 }, { "epoch": 5.55, "learning_rate": 4.538369039382184e-06, "loss": 0.0559, "step": 6052 }, { "epoch": 5.55, "learning_rate": 4.5352609277170725e-06, "loss": 0.0551, "step": 6053 }, { "epoch": 5.55, "learning_rate": 4.5321535685626e-06, "loss": 0.0565, "step": 6054 }, { "epoch": 5.55, "learning_rate": 4.529046962346653e-06, "loss": 0.0557, "step": 6055 }, { "epoch": 5.55, "learning_rate": 4.525941109497022e-06, "loss": 0.0506, "step": 6056 }, { "epoch": 5.55, "learning_rate": 4.522836010441396e-06, "loss": 0.0493, "step": 6057 }, { "epoch": 5.55, "learning_rate": 4.519731665607353e-06, "loss": 0.0504, "step": 6058 }, { "epoch": 5.55, "learning_rate": 4.516628075422368e-06, "loss": 0.0534, "step": 6059 }, { "epoch": 5.55, "learning_rate": 4.51352524031382e-06, "loss": 0.045, "step": 6060 }, { "epoch": 5.56, "learning_rate": 4.510423160708975e-06, "loss": 0.0455, "step": 6061 }, { "epoch": 5.56, "learning_rate": 4.507321837034997e-06, "loss": 0.0448, "step": 6062 }, { "epoch": 5.56, "learning_rate": 4.504221269718951e-06, "loss": 0.052, "step": 6063 }, { "epoch": 5.56, "learning_rate": 4.5011214591877914e-06, "loss": 0.0513, "step": 6064 }, { "epoch": 5.56, "learning_rate": 4.4980224058683755e-06, "loss": 0.0561, "step": 6065 }, { "epoch": 5.56, "learning_rate": 4.49492411018745e-06, "loss": 0.0471, "step": 6066 }, { "epoch": 5.56, "learning_rate": 4.4918265725716525e-06, "loss": 0.0541, "step": 6067 }, { "epoch": 5.56, "learning_rate": 4.488729793447535e-06, "loss": 0.0476, "step": 6068 }, { "epoch": 5.56, "learning_rate": 4.485633773241526e-06, "loss": 0.0414, "step": 6069 }, { "epoch": 5.56, "learning_rate": 4.4825385123799585e-06, "loss": 0.0545, "step": 6070 }, { "epoch": 5.56, "learning_rate": 4.479444011289054e-06, "loss": 0.0607, "step": 6071 }, { "epoch": 5.57, "learning_rate": 4.476350270394942e-06, "loss": 0.0636, "step": 6072 }, { "epoch": 5.57, "learning_rate": 4.473257290123637e-06, "loss": 0.0513, "step": 6073 }, { "epoch": 5.57, "learning_rate": 4.470165070901047e-06, "loss": 0.058, "step": 6074 }, { "epoch": 5.57, "learning_rate": 4.467073613152987e-06, "loss": 0.0468, "step": 6075 }, { "epoch": 5.57, "learning_rate": 4.463982917305155e-06, "loss": 0.0454, "step": 6076 }, { "epoch": 5.57, "learning_rate": 4.460892983783153e-06, "loss": 0.0556, "step": 6077 }, { "epoch": 5.57, "learning_rate": 4.457803813012466e-06, "loss": 0.043, "step": 6078 }, { "epoch": 5.57, "learning_rate": 4.454715405418489e-06, "loss": 0.0508, "step": 6079 }, { "epoch": 5.57, "learning_rate": 4.451627761426506e-06, "loss": 0.05, "step": 6080 }, { "epoch": 5.57, "learning_rate": 4.448540881461689e-06, "loss": 0.0538, "step": 6081 }, { "epoch": 5.57, "learning_rate": 4.445454765949111e-06, "loss": 0.0547, "step": 6082 }, { "epoch": 5.58, "learning_rate": 4.442369415313744e-06, "loss": 0.056, "step": 6083 }, { "epoch": 5.58, "learning_rate": 4.439284829980447e-06, "loss": 0.0476, "step": 6084 }, { "epoch": 5.58, "learning_rate": 4.436201010373973e-06, "loss": 0.0489, "step": 6085 }, { "epoch": 5.58, "learning_rate": 4.433117956918979e-06, "loss": 0.0506, "step": 6086 }, { "epoch": 5.58, "learning_rate": 4.43003567004001e-06, "loss": 0.0558, "step": 6087 }, { "epoch": 5.58, "learning_rate": 4.426954150161503e-06, "loss": 0.0408, "step": 6088 }, { "epoch": 5.58, "learning_rate": 4.423873397707789e-06, "loss": 0.0519, "step": 6089 }, { "epoch": 5.58, "learning_rate": 4.420793413103106e-06, "loss": 0.0484, "step": 6090 }, { "epoch": 5.58, "learning_rate": 4.417714196771571e-06, "loss": 0.0466, "step": 6091 }, { "epoch": 5.58, "learning_rate": 4.414635749137198e-06, "loss": 0.0481, "step": 6092 }, { "epoch": 5.58, "learning_rate": 4.411558070623907e-06, "loss": 0.0608, "step": 6093 }, { "epoch": 5.59, "learning_rate": 4.408481161655499e-06, "loss": 0.0508, "step": 6094 }, { "epoch": 5.59, "learning_rate": 4.405405022655672e-06, "loss": 0.0529, "step": 6095 }, { "epoch": 5.59, "learning_rate": 4.402329654048018e-06, "loss": 0.0488, "step": 6096 }, { "epoch": 5.59, "learning_rate": 4.399255056256028e-06, "loss": 0.0493, "step": 6097 }, { "epoch": 5.59, "learning_rate": 4.396181229703085e-06, "loss": 0.0427, "step": 6098 }, { "epoch": 5.59, "learning_rate": 4.393108174812456e-06, "loss": 0.0521, "step": 6099 }, { "epoch": 5.59, "learning_rate": 4.390035892007318e-06, "loss": 0.058, "step": 6100 }, { "epoch": 5.59, "learning_rate": 4.38696438171073e-06, "loss": 0.0477, "step": 6101 }, { "epoch": 5.59, "learning_rate": 4.383893644345644e-06, "loss": 0.0448, "step": 6102 }, { "epoch": 5.59, "learning_rate": 4.380823680334919e-06, "loss": 0.0605, "step": 6103 }, { "epoch": 5.59, "learning_rate": 4.377754490101291e-06, "loss": 0.0478, "step": 6104 }, { "epoch": 5.6, "learning_rate": 4.374686074067395e-06, "loss": 0.0484, "step": 6105 }, { "epoch": 5.6, "learning_rate": 4.3716184326557676e-06, "loss": 0.0528, "step": 6106 }, { "epoch": 5.6, "learning_rate": 4.3685515662888245e-06, "loss": 0.0533, "step": 6107 }, { "epoch": 5.6, "learning_rate": 4.3654854753888905e-06, "loss": 0.0434, "step": 6108 }, { "epoch": 5.6, "learning_rate": 4.362420160378172e-06, "loss": 0.058, "step": 6109 }, { "epoch": 5.6, "learning_rate": 4.359355621678765e-06, "loss": 0.0454, "step": 6110 }, { "epoch": 5.6, "learning_rate": 4.356291859712676e-06, "loss": 0.0502, "step": 6111 }, { "epoch": 5.6, "learning_rate": 4.353228874901789e-06, "loss": 0.0505, "step": 6112 }, { "epoch": 5.6, "learning_rate": 4.350166667667888e-06, "loss": 0.0428, "step": 6113 }, { "epoch": 5.6, "learning_rate": 4.3471052384326416e-06, "loss": 0.0644, "step": 6114 }, { "epoch": 5.6, "learning_rate": 4.344044587617626e-06, "loss": 0.051, "step": 6115 }, { "epoch": 5.61, "learning_rate": 4.340984715644298e-06, "loss": 0.0451, "step": 6116 }, { "epoch": 5.61, "learning_rate": 4.337925622934008e-06, "loss": 0.0516, "step": 6117 }, { "epoch": 5.61, "learning_rate": 4.334867309908009e-06, "loss": 0.0564, "step": 6118 }, { "epoch": 5.61, "learning_rate": 4.331809776987436e-06, "loss": 0.0503, "step": 6119 }, { "epoch": 5.61, "learning_rate": 4.328753024593319e-06, "loss": 0.0483, "step": 6120 }, { "epoch": 5.61, "learning_rate": 4.32569705314658e-06, "loss": 0.0584, "step": 6121 }, { "epoch": 5.61, "learning_rate": 4.322641863068044e-06, "loss": 0.0403, "step": 6122 }, { "epoch": 5.61, "learning_rate": 4.319587454778411e-06, "loss": 0.0535, "step": 6123 }, { "epoch": 5.61, "learning_rate": 4.3165338286982825e-06, "loss": 0.0452, "step": 6124 }, { "epoch": 5.61, "learning_rate": 4.313480985248157e-06, "loss": 0.0578, "step": 6125 }, { "epoch": 5.62, "learning_rate": 4.3104289248484184e-06, "loss": 0.0421, "step": 6126 }, { "epoch": 5.62, "learning_rate": 4.307377647919343e-06, "loss": 0.0489, "step": 6127 }, { "epoch": 5.62, "learning_rate": 4.304327154881096e-06, "loss": 0.0483, "step": 6128 }, { "epoch": 5.62, "learning_rate": 4.301277446153749e-06, "loss": 0.0516, "step": 6129 }, { "epoch": 5.62, "learning_rate": 4.2982285221572505e-06, "loss": 0.0452, "step": 6130 }, { "epoch": 5.62, "learning_rate": 4.295180383311443e-06, "loss": 0.0482, "step": 6131 }, { "epoch": 5.62, "learning_rate": 4.292133030036072e-06, "loss": 0.0475, "step": 6132 }, { "epoch": 5.62, "learning_rate": 4.289086462750762e-06, "loss": 0.0558, "step": 6133 }, { "epoch": 5.62, "learning_rate": 4.286040681875036e-06, "loss": 0.0546, "step": 6134 }, { "epoch": 5.62, "learning_rate": 4.282995687828302e-06, "loss": 0.0481, "step": 6135 }, { "epoch": 5.62, "learning_rate": 4.2799514810298725e-06, "loss": 0.0481, "step": 6136 }, { "epoch": 5.63, "learning_rate": 4.27690806189894e-06, "loss": 0.0478, "step": 6137 }, { "epoch": 5.63, "learning_rate": 4.273865430854593e-06, "loss": 0.0508, "step": 6138 }, { "epoch": 5.63, "learning_rate": 4.2708235883158055e-06, "loss": 0.0409, "step": 6139 }, { "epoch": 5.63, "learning_rate": 4.267782534701456e-06, "loss": 0.0443, "step": 6140 }, { "epoch": 5.63, "learning_rate": 4.264742270430304e-06, "loss": 0.0507, "step": 6141 }, { "epoch": 5.63, "learning_rate": 4.261702795920998e-06, "loss": 0.0546, "step": 6142 }, { "epoch": 5.63, "learning_rate": 4.25866411159209e-06, "loss": 0.0565, "step": 6143 }, { "epoch": 5.63, "learning_rate": 4.255626217862014e-06, "loss": 0.0472, "step": 6144 }, { "epoch": 5.63, "learning_rate": 4.25258911514909e-06, "loss": 0.0502, "step": 6145 }, { "epoch": 5.63, "learning_rate": 4.249552803871545e-06, "loss": 0.0515, "step": 6146 }, { "epoch": 5.63, "learning_rate": 4.2465172844474845e-06, "loss": 0.0552, "step": 6147 }, { "epoch": 5.64, "learning_rate": 4.243482557294905e-06, "loss": 0.055, "step": 6148 }, { "epoch": 5.64, "learning_rate": 4.240448622831703e-06, "loss": 0.0522, "step": 6149 }, { "epoch": 5.64, "learning_rate": 4.237415481475654e-06, "loss": 0.0581, "step": 6150 }, { "epoch": 5.64, "learning_rate": 4.23438313364444e-06, "loss": 0.0514, "step": 6151 }, { "epoch": 5.64, "learning_rate": 4.231351579755616e-06, "loss": 0.053, "step": 6152 }, { "epoch": 5.64, "learning_rate": 4.228320820226636e-06, "loss": 0.0479, "step": 6153 }, { "epoch": 5.64, "learning_rate": 4.22529085547485e-06, "loss": 0.0517, "step": 6154 }, { "epoch": 5.64, "learning_rate": 4.222261685917489e-06, "loss": 0.0393, "step": 6155 }, { "epoch": 5.64, "learning_rate": 4.219233311971678e-06, "loss": 0.0479, "step": 6156 }, { "epoch": 5.64, "learning_rate": 4.216205734054436e-06, "loss": 0.0486, "step": 6157 }, { "epoch": 5.64, "learning_rate": 4.2131789525826685e-06, "loss": 0.0561, "step": 6158 }, { "epoch": 5.65, "learning_rate": 4.2101529679731715e-06, "loss": 0.0493, "step": 6159 }, { "epoch": 5.65, "learning_rate": 4.207127780642628e-06, "loss": 0.0555, "step": 6160 }, { "epoch": 5.65, "learning_rate": 4.2041033910076235e-06, "loss": 0.0509, "step": 6161 }, { "epoch": 5.65, "learning_rate": 4.20107979948462e-06, "loss": 0.0447, "step": 6162 }, { "epoch": 5.65, "learning_rate": 4.198057006489973e-06, "loss": 0.0532, "step": 6163 }, { "epoch": 5.65, "learning_rate": 4.195035012439936e-06, "loss": 0.0541, "step": 6164 }, { "epoch": 5.65, "learning_rate": 4.192013817750644e-06, "loss": 0.0551, "step": 6165 }, { "epoch": 5.65, "learning_rate": 4.188993422838123e-06, "loss": 0.0516, "step": 6166 }, { "epoch": 5.65, "learning_rate": 4.185973828118288e-06, "loss": 0.0442, "step": 6167 }, { "epoch": 5.65, "learning_rate": 4.182955034006954e-06, "loss": 0.0533, "step": 6168 }, { "epoch": 5.65, "learning_rate": 4.17993704091981e-06, "loss": 0.0492, "step": 6169 }, { "epoch": 5.66, "learning_rate": 4.176919849272447e-06, "loss": 0.0443, "step": 6170 }, { "epoch": 5.66, "learning_rate": 4.173903459480336e-06, "loss": 0.053, "step": 6171 }, { "epoch": 5.66, "learning_rate": 4.1708878719588505e-06, "loss": 0.0595, "step": 6172 }, { "epoch": 5.66, "learning_rate": 4.167873087123241e-06, "loss": 0.0485, "step": 6173 }, { "epoch": 5.66, "learning_rate": 4.164859105388649e-06, "loss": 0.0516, "step": 6174 }, { "epoch": 5.66, "learning_rate": 4.161845927170117e-06, "loss": 0.0589, "step": 6175 }, { "epoch": 5.66, "learning_rate": 4.158833552882563e-06, "loss": 0.0476, "step": 6176 }, { "epoch": 5.66, "learning_rate": 4.155821982940801e-06, "loss": 0.0492, "step": 6177 }, { "epoch": 5.66, "learning_rate": 4.152811217759529e-06, "loss": 0.0472, "step": 6178 }, { "epoch": 5.66, "learning_rate": 4.149801257753346e-06, "loss": 0.045, "step": 6179 }, { "epoch": 5.66, "learning_rate": 4.146792103336728e-06, "loss": 0.0576, "step": 6180 }, { "epoch": 5.67, "learning_rate": 4.143783754924042e-06, "loss": 0.0527, "step": 6181 }, { "epoch": 5.67, "learning_rate": 4.1407762129295525e-06, "loss": 0.0434, "step": 6182 }, { "epoch": 5.67, "learning_rate": 4.137769477767404e-06, "loss": 0.0509, "step": 6183 }, { "epoch": 5.67, "learning_rate": 4.134763549851631e-06, "loss": 0.045, "step": 6184 }, { "epoch": 5.67, "learning_rate": 4.131758429596158e-06, "loss": 0.0527, "step": 6185 }, { "epoch": 5.67, "learning_rate": 4.1287541174148035e-06, "loss": 0.0526, "step": 6186 }, { "epoch": 5.67, "learning_rate": 4.125750613721268e-06, "loss": 0.0448, "step": 6187 }, { "epoch": 5.67, "learning_rate": 4.1227479189291374e-06, "loss": 0.0399, "step": 6188 }, { "epoch": 5.67, "learning_rate": 4.1197460334519015e-06, "loss": 0.0456, "step": 6189 }, { "epoch": 5.67, "learning_rate": 4.116744957702922e-06, "loss": 0.0427, "step": 6190 }, { "epoch": 5.67, "learning_rate": 4.113744692095456e-06, "loss": 0.0548, "step": 6191 }, { "epoch": 5.68, "learning_rate": 4.110745237042652e-06, "loss": 0.0464, "step": 6192 }, { "epoch": 5.68, "learning_rate": 4.107746592957545e-06, "loss": 0.0421, "step": 6193 }, { "epoch": 5.68, "learning_rate": 4.104748760253049e-06, "loss": 0.0471, "step": 6194 }, { "epoch": 5.68, "learning_rate": 4.1017517393419826e-06, "loss": 0.0553, "step": 6195 }, { "epoch": 5.68, "learning_rate": 4.09875553063704e-06, "loss": 0.0506, "step": 6196 }, { "epoch": 5.68, "learning_rate": 4.095760134550811e-06, "loss": 0.0474, "step": 6197 }, { "epoch": 5.68, "learning_rate": 4.09276555149577e-06, "loss": 0.0534, "step": 6198 }, { "epoch": 5.68, "learning_rate": 4.0897717818842755e-06, "loss": 0.0517, "step": 6199 }, { "epoch": 5.68, "learning_rate": 4.086778826128584e-06, "loss": 0.0391, "step": 6200 }, { "epoch": 5.68, "learning_rate": 4.083786684640832e-06, "loss": 0.0436, "step": 6201 }, { "epoch": 5.68, "learning_rate": 4.0807953578330475e-06, "loss": 0.0496, "step": 6202 }, { "epoch": 5.69, "learning_rate": 4.07780484611714e-06, "loss": 0.0411, "step": 6203 }, { "epoch": 5.69, "learning_rate": 4.0748151499049185e-06, "loss": 0.0597, "step": 6204 }, { "epoch": 5.69, "learning_rate": 4.07182626960807e-06, "loss": 0.0504, "step": 6205 }, { "epoch": 5.69, "learning_rate": 4.0688382056381685e-06, "loss": 0.0513, "step": 6206 }, { "epoch": 5.69, "learning_rate": 4.065850958406686e-06, "loss": 0.0464, "step": 6207 }, { "epoch": 5.69, "learning_rate": 4.0628645283249716e-06, "loss": 0.0507, "step": 6208 }, { "epoch": 5.69, "learning_rate": 4.059878915804266e-06, "loss": 0.0448, "step": 6209 }, { "epoch": 5.69, "learning_rate": 4.056894121255693e-06, "loss": 0.0492, "step": 6210 }, { "epoch": 5.69, "learning_rate": 4.053910145090275e-06, "loss": 0.0538, "step": 6211 }, { "epoch": 5.69, "learning_rate": 4.050926987718911e-06, "loss": 0.0493, "step": 6212 }, { "epoch": 5.69, "learning_rate": 4.047944649552385e-06, "loss": 0.0442, "step": 6213 }, { "epoch": 5.7, "learning_rate": 4.044963131001384e-06, "loss": 0.0501, "step": 6214 }, { "epoch": 5.7, "learning_rate": 4.041982432476466e-06, "loss": 0.0594, "step": 6215 }, { "epoch": 5.7, "learning_rate": 4.039002554388082e-06, "loss": 0.0525, "step": 6216 }, { "epoch": 5.7, "learning_rate": 4.036023497146567e-06, "loss": 0.0561, "step": 6217 }, { "epoch": 5.7, "learning_rate": 4.0330452611621525e-06, "loss": 0.0577, "step": 6218 }, { "epoch": 5.7, "learning_rate": 4.0300678468449495e-06, "loss": 0.0535, "step": 6219 }, { "epoch": 5.7, "learning_rate": 4.02709125460495e-06, "loss": 0.0584, "step": 6220 }, { "epoch": 5.7, "learning_rate": 4.024115484852047e-06, "loss": 0.0473, "step": 6221 }, { "epoch": 5.7, "learning_rate": 4.02114053799601e-06, "loss": 0.0466, "step": 6222 }, { "epoch": 5.7, "learning_rate": 4.0181664144464984e-06, "loss": 0.0622, "step": 6223 }, { "epoch": 5.7, "learning_rate": 4.015193114613053e-06, "loss": 0.0521, "step": 6224 }, { "epoch": 5.71, "learning_rate": 4.012220638905113e-06, "loss": 0.0482, "step": 6225 }, { "epoch": 5.71, "learning_rate": 4.009248987731995e-06, "loss": 0.0514, "step": 6226 }, { "epoch": 5.71, "learning_rate": 4.006278161502904e-06, "loss": 0.0533, "step": 6227 }, { "epoch": 5.71, "learning_rate": 4.003308160626925e-06, "loss": 0.0477, "step": 6228 }, { "epoch": 5.71, "learning_rate": 4.000338985513046e-06, "loss": 0.0414, "step": 6229 }, { "epoch": 5.71, "learning_rate": 3.997370636570127e-06, "loss": 0.0419, "step": 6230 }, { "epoch": 5.71, "learning_rate": 3.994403114206914e-06, "loss": 0.0474, "step": 6231 }, { "epoch": 5.71, "learning_rate": 3.991436418832051e-06, "loss": 0.058, "step": 6232 }, { "epoch": 5.71, "learning_rate": 3.9884705508540575e-06, "loss": 0.0505, "step": 6233 }, { "epoch": 5.71, "learning_rate": 3.985505510681339e-06, "loss": 0.046, "step": 6234 }, { "epoch": 5.71, "learning_rate": 3.9825412987221965e-06, "loss": 0.0419, "step": 6235 }, { "epoch": 5.72, "learning_rate": 3.979577915384807e-06, "loss": 0.0466, "step": 6236 }, { "epoch": 5.72, "learning_rate": 3.976615361077234e-06, "loss": 0.0485, "step": 6237 }, { "epoch": 5.72, "learning_rate": 3.973653636207437e-06, "loss": 0.0433, "step": 6238 }, { "epoch": 5.72, "learning_rate": 3.9706927411832486e-06, "loss": 0.0433, "step": 6239 }, { "epoch": 5.72, "learning_rate": 3.967732676412397e-06, "loss": 0.0567, "step": 6240 }, { "epoch": 5.72, "learning_rate": 3.96477344230249e-06, "loss": 0.0472, "step": 6241 }, { "epoch": 5.72, "learning_rate": 3.961815039261019e-06, "loss": 0.057, "step": 6242 }, { "epoch": 5.72, "learning_rate": 3.958857467695372e-06, "loss": 0.0486, "step": 6243 }, { "epoch": 5.72, "learning_rate": 3.95590072801281e-06, "loss": 0.0429, "step": 6244 }, { "epoch": 5.72, "learning_rate": 3.952944820620485e-06, "loss": 0.0489, "step": 6245 }, { "epoch": 5.73, "learning_rate": 3.9499897459254375e-06, "loss": 0.0422, "step": 6246 }, { "epoch": 5.73, "learning_rate": 3.947035504334589e-06, "loss": 0.053, "step": 6247 }, { "epoch": 5.73, "learning_rate": 3.944082096254746e-06, "loss": 0.0483, "step": 6248 }, { "epoch": 5.73, "learning_rate": 3.941129522092598e-06, "loss": 0.0482, "step": 6249 }, { "epoch": 5.73, "learning_rate": 3.938177782254731e-06, "loss": 0.0475, "step": 6250 }, { "epoch": 5.73, "learning_rate": 3.935226877147603e-06, "loss": 0.0471, "step": 6251 }, { "epoch": 5.73, "learning_rate": 3.932276807177561e-06, "loss": 0.0504, "step": 6252 }, { "epoch": 5.73, "learning_rate": 3.9293275727508425e-06, "loss": 0.0444, "step": 6253 }, { "epoch": 5.73, "learning_rate": 3.926379174273566e-06, "loss": 0.0583, "step": 6254 }, { "epoch": 5.73, "learning_rate": 3.923431612151731e-06, "loss": 0.0508, "step": 6255 }, { "epoch": 5.73, "learning_rate": 3.920484886791225e-06, "loss": 0.047, "step": 6256 }, { "epoch": 5.74, "learning_rate": 3.917538998597825e-06, "loss": 0.0518, "step": 6257 }, { "epoch": 5.74, "learning_rate": 3.914593947977186e-06, "loss": 0.0463, "step": 6258 }, { "epoch": 5.74, "learning_rate": 3.911649735334853e-06, "loss": 0.0486, "step": 6259 }, { "epoch": 5.74, "learning_rate": 3.908706361076244e-06, "loss": 0.0495, "step": 6260 }, { "epoch": 5.74, "learning_rate": 3.905763825606681e-06, "loss": 0.0495, "step": 6261 }, { "epoch": 5.74, "learning_rate": 3.902822129331355e-06, "loss": 0.0478, "step": 6262 }, { "epoch": 5.74, "learning_rate": 3.899881272655342e-06, "loss": 0.0521, "step": 6263 }, { "epoch": 5.74, "learning_rate": 3.896941255983616e-06, "loss": 0.0612, "step": 6264 }, { "epoch": 5.74, "learning_rate": 3.894002079721021e-06, "loss": 0.0454, "step": 6265 }, { "epoch": 5.74, "learning_rate": 3.8910637442722886e-06, "loss": 0.0516, "step": 6266 }, { "epoch": 5.74, "learning_rate": 3.888126250042036e-06, "loss": 0.0486, "step": 6267 }, { "epoch": 5.75, "learning_rate": 3.88518959743477e-06, "loss": 0.0436, "step": 6268 }, { "epoch": 5.75, "learning_rate": 3.882253786854871e-06, "loss": 0.0446, "step": 6269 }, { "epoch": 5.75, "learning_rate": 3.879318818706609e-06, "loss": 0.0432, "step": 6270 }, { "epoch": 5.75, "learning_rate": 3.876384693394143e-06, "loss": 0.0448, "step": 6271 }, { "epoch": 5.75, "learning_rate": 3.873451411321506e-06, "loss": 0.0499, "step": 6272 }, { "epoch": 5.75, "learning_rate": 3.870518972892622e-06, "loss": 0.0508, "step": 6273 }, { "epoch": 5.75, "learning_rate": 3.867587378511291e-06, "loss": 0.0581, "step": 6274 }, { "epoch": 5.75, "learning_rate": 3.864656628581208e-06, "loss": 0.0488, "step": 6275 }, { "epoch": 5.75, "learning_rate": 3.861726723505947e-06, "loss": 0.062, "step": 6276 }, { "epoch": 5.75, "learning_rate": 3.858797663688955e-06, "loss": 0.0528, "step": 6277 }, { "epoch": 5.75, "learning_rate": 3.855869449533584e-06, "loss": 0.047, "step": 6278 }, { "epoch": 5.76, "learning_rate": 3.852942081443053e-06, "loss": 0.0513, "step": 6279 }, { "epoch": 5.76, "learning_rate": 3.850015559820465e-06, "loss": 0.0584, "step": 6280 }, { "epoch": 5.76, "learning_rate": 3.847089885068817e-06, "loss": 0.0444, "step": 6281 }, { "epoch": 5.76, "learning_rate": 3.844165057590978e-06, "loss": 0.054, "step": 6282 }, { "epoch": 5.76, "learning_rate": 3.841241077789709e-06, "loss": 0.0431, "step": 6283 }, { "epoch": 5.76, "learning_rate": 3.8383179460676515e-06, "loss": 0.0568, "step": 6284 }, { "epoch": 5.76, "learning_rate": 3.835395662827323e-06, "loss": 0.0519, "step": 6285 }, { "epoch": 5.76, "learning_rate": 3.832474228471137e-06, "loss": 0.0515, "step": 6286 }, { "epoch": 5.76, "learning_rate": 3.829553643401383e-06, "loss": 0.0484, "step": 6287 }, { "epoch": 5.76, "learning_rate": 3.826633908020228e-06, "loss": 0.0507, "step": 6288 }, { "epoch": 5.76, "learning_rate": 3.823715022729735e-06, "loss": 0.0506, "step": 6289 }, { "epoch": 5.77, "learning_rate": 3.820796987931842e-06, "loss": 0.0479, "step": 6290 }, { "epoch": 5.77, "learning_rate": 3.817879804028368e-06, "loss": 0.0495, "step": 6291 }, { "epoch": 5.77, "learning_rate": 3.814963471421017e-06, "loss": 0.0551, "step": 6292 }, { "epoch": 5.77, "learning_rate": 3.812047990511383e-06, "loss": 0.0621, "step": 6293 }, { "epoch": 5.77, "learning_rate": 3.8091333617009306e-06, "loss": 0.0479, "step": 6294 }, { "epoch": 5.77, "learning_rate": 3.8062195853910123e-06, "loss": 0.0435, "step": 6295 }, { "epoch": 5.77, "learning_rate": 3.8033066619828686e-06, "loss": 0.0414, "step": 6296 }, { "epoch": 5.77, "learning_rate": 3.8003945918776143e-06, "loss": 0.0518, "step": 6297 }, { "epoch": 5.77, "learning_rate": 3.7974833754762506e-06, "loss": 0.0523, "step": 6298 }, { "epoch": 5.77, "learning_rate": 3.7945730131796577e-06, "loss": 0.0458, "step": 6299 }, { "epoch": 5.77, "learning_rate": 3.7916635053886066e-06, "loss": 0.0484, "step": 6300 }, { "epoch": 5.78, "learning_rate": 3.7887548525037434e-06, "loss": 0.0521, "step": 6301 }, { "epoch": 5.78, "learning_rate": 3.7858470549255923e-06, "loss": 0.0486, "step": 6302 }, { "epoch": 5.78, "learning_rate": 3.782940113054574e-06, "loss": 0.048, "step": 6303 }, { "epoch": 5.78, "learning_rate": 3.7800340272909784e-06, "loss": 0.0539, "step": 6304 }, { "epoch": 5.78, "learning_rate": 3.7771287980349836e-06, "loss": 0.0454, "step": 6305 }, { "epoch": 5.78, "learning_rate": 3.7742244256866445e-06, "loss": 0.0441, "step": 6306 }, { "epoch": 5.78, "learning_rate": 3.771320910645908e-06, "loss": 0.0466, "step": 6307 }, { "epoch": 5.78, "learning_rate": 3.7684182533125922e-06, "loss": 0.0508, "step": 6308 }, { "epoch": 5.78, "learning_rate": 3.7655164540864007e-06, "loss": 0.0505, "step": 6309 }, { "epoch": 5.78, "learning_rate": 3.762615513366925e-06, "loss": 0.0492, "step": 6310 }, { "epoch": 5.78, "learning_rate": 3.7597154315536287e-06, "loss": 0.0545, "step": 6311 }, { "epoch": 5.79, "learning_rate": 3.756816209045864e-06, "loss": 0.0509, "step": 6312 }, { "epoch": 5.79, "learning_rate": 3.7539178462428582e-06, "loss": 0.0513, "step": 6313 }, { "epoch": 5.79, "learning_rate": 3.75102034354373e-06, "loss": 0.0516, "step": 6314 }, { "epoch": 5.79, "learning_rate": 3.748123701347471e-06, "loss": 0.0462, "step": 6315 }, { "epoch": 5.79, "learning_rate": 3.7452279200529584e-06, "loss": 0.049, "step": 6316 }, { "epoch": 5.79, "learning_rate": 3.7423330000589455e-06, "loss": 0.0488, "step": 6317 }, { "epoch": 5.79, "learning_rate": 3.7394389417640786e-06, "loss": 0.0545, "step": 6318 }, { "epoch": 5.79, "learning_rate": 3.7365457455668742e-06, "loss": 0.0472, "step": 6319 }, { "epoch": 5.79, "learning_rate": 3.733653411865731e-06, "loss": 0.0515, "step": 6320 }, { "epoch": 5.79, "learning_rate": 3.730761941058938e-06, "loss": 0.0562, "step": 6321 }, { "epoch": 5.79, "learning_rate": 3.7278713335446557e-06, "loss": 0.0441, "step": 6322 }, { "epoch": 5.8, "learning_rate": 3.724981589720926e-06, "loss": 0.0479, "step": 6323 }, { "epoch": 5.8, "learning_rate": 3.7220927099856842e-06, "loss": 0.0527, "step": 6324 }, { "epoch": 5.8, "learning_rate": 3.719204694736728e-06, "loss": 0.0456, "step": 6325 }, { "epoch": 5.8, "learning_rate": 3.7163175443717526e-06, "loss": 0.0533, "step": 6326 }, { "epoch": 5.8, "learning_rate": 3.713431259288325e-06, "loss": 0.0415, "step": 6327 }, { "epoch": 5.8, "learning_rate": 3.71054583988389e-06, "loss": 0.0542, "step": 6328 }, { "epoch": 5.8, "learning_rate": 3.707661286555787e-06, "loss": 0.0471, "step": 6329 }, { "epoch": 5.8, "learning_rate": 3.7047775997012224e-06, "loss": 0.04, "step": 6330 }, { "epoch": 5.8, "learning_rate": 3.7018947797172864e-06, "loss": 0.0438, "step": 6331 }, { "epoch": 5.8, "learning_rate": 3.6990128270009585e-06, "loss": 0.0487, "step": 6332 }, { "epoch": 5.8, "learning_rate": 3.696131741949087e-06, "loss": 0.0464, "step": 6333 }, { "epoch": 5.81, "learning_rate": 3.6932515249584044e-06, "loss": 0.0525, "step": 6334 }, { "epoch": 5.81, "learning_rate": 3.690372176425531e-06, "loss": 0.0351, "step": 6335 }, { "epoch": 5.81, "learning_rate": 3.6874936967469575e-06, "loss": 0.0507, "step": 6336 }, { "epoch": 5.81, "learning_rate": 3.6846160863190595e-06, "loss": 0.0546, "step": 6337 }, { "epoch": 5.81, "learning_rate": 3.68173934553809e-06, "loss": 0.0471, "step": 6338 }, { "epoch": 5.81, "learning_rate": 3.6788634748001894e-06, "loss": 0.0444, "step": 6339 }, { "epoch": 5.81, "learning_rate": 3.675988474501373e-06, "loss": 0.0457, "step": 6340 }, { "epoch": 5.81, "learning_rate": 3.6731143450375306e-06, "loss": 0.0397, "step": 6341 }, { "epoch": 5.81, "learning_rate": 3.6702410868044456e-06, "loss": 0.046, "step": 6342 }, { "epoch": 5.81, "learning_rate": 3.667368700197773e-06, "loss": 0.0519, "step": 6343 }, { "epoch": 5.81, "learning_rate": 3.664497185613046e-06, "loss": 0.0506, "step": 6344 }, { "epoch": 5.82, "learning_rate": 3.6616265434456787e-06, "loss": 0.0524, "step": 6345 }, { "epoch": 5.82, "learning_rate": 3.6587567740909744e-06, "loss": 0.0526, "step": 6346 }, { "epoch": 5.82, "learning_rate": 3.6558878779441033e-06, "loss": 0.0542, "step": 6347 }, { "epoch": 5.82, "learning_rate": 3.653019855400123e-06, "loss": 0.0534, "step": 6348 }, { "epoch": 5.82, "learning_rate": 3.6501527068539633e-06, "loss": 0.0493, "step": 6349 }, { "epoch": 5.82, "learning_rate": 3.6472864327004464e-06, "loss": 0.0554, "step": 6350 }, { "epoch": 5.82, "learning_rate": 3.6444210333342644e-06, "loss": 0.0412, "step": 6351 }, { "epoch": 5.82, "learning_rate": 3.641556509149987e-06, "loss": 0.0584, "step": 6352 }, { "epoch": 5.82, "learning_rate": 3.6386928605420748e-06, "loss": 0.0441, "step": 6353 }, { "epoch": 5.82, "learning_rate": 3.6358300879048556e-06, "loss": 0.0464, "step": 6354 }, { "epoch": 5.82, "learning_rate": 3.6329681916325444e-06, "loss": 0.0474, "step": 6355 }, { "epoch": 5.83, "learning_rate": 3.630107172119226e-06, "loss": 0.052, "step": 6356 }, { "epoch": 5.83, "learning_rate": 3.6272470297588814e-06, "loss": 0.0516, "step": 6357 }, { "epoch": 5.83, "learning_rate": 3.6243877649453553e-06, "loss": 0.0491, "step": 6358 }, { "epoch": 5.83, "learning_rate": 3.6215293780723726e-06, "loss": 0.0435, "step": 6359 }, { "epoch": 5.83, "learning_rate": 3.61867186953355e-06, "loss": 0.0446, "step": 6360 }, { "epoch": 5.83, "learning_rate": 3.615815239722371e-06, "loss": 0.0456, "step": 6361 }, { "epoch": 5.83, "learning_rate": 3.612959489032203e-06, "loss": 0.04, "step": 6362 }, { "epoch": 5.83, "learning_rate": 3.610104617856286e-06, "loss": 0.0496, "step": 6363 }, { "epoch": 5.83, "learning_rate": 3.6072506265877515e-06, "loss": 0.0427, "step": 6364 }, { "epoch": 5.83, "learning_rate": 3.604397515619599e-06, "loss": 0.0437, "step": 6365 }, { "epoch": 5.84, "learning_rate": 3.6015452853447077e-06, "loss": 0.0512, "step": 6366 }, { "epoch": 5.84, "learning_rate": 3.598693936155844e-06, "loss": 0.0524, "step": 6367 }, { "epoch": 5.84, "learning_rate": 3.5958434684456444e-06, "loss": 0.0416, "step": 6368 }, { "epoch": 5.84, "learning_rate": 3.5929938826066223e-06, "loss": 0.0468, "step": 6369 }, { "epoch": 5.84, "learning_rate": 3.5901451790311825e-06, "loss": 0.044, "step": 6370 }, { "epoch": 5.84, "learning_rate": 3.587297358111591e-06, "loss": 0.0583, "step": 6371 }, { "epoch": 5.84, "learning_rate": 3.58445042024001e-06, "loss": 0.0422, "step": 6372 }, { "epoch": 5.84, "learning_rate": 3.581604365808465e-06, "loss": 0.0388, "step": 6373 }, { "epoch": 5.84, "learning_rate": 3.5787591952088663e-06, "loss": 0.0496, "step": 6374 }, { "epoch": 5.84, "learning_rate": 3.575914908833006e-06, "loss": 0.0506, "step": 6375 }, { "epoch": 5.84, "learning_rate": 3.5730715070725486e-06, "loss": 0.0593, "step": 6376 }, { "epoch": 5.85, "learning_rate": 3.570228990319036e-06, "loss": 0.0534, "step": 6377 }, { "epoch": 5.85, "learning_rate": 3.5673873589638964e-06, "loss": 0.047, "step": 6378 }, { "epoch": 5.85, "learning_rate": 3.5645466133984286e-06, "loss": 0.0529, "step": 6379 }, { "epoch": 5.85, "learning_rate": 3.5617067540138107e-06, "loss": 0.0508, "step": 6380 }, { "epoch": 5.85, "learning_rate": 3.5588677812010973e-06, "loss": 0.0517, "step": 6381 }, { "epoch": 5.85, "learning_rate": 3.5560296953512296e-06, "loss": 0.0598, "step": 6382 }, { "epoch": 5.85, "learning_rate": 3.5531924968550167e-06, "loss": 0.0428, "step": 6383 }, { "epoch": 5.85, "learning_rate": 3.550356186103148e-06, "loss": 0.0454, "step": 6384 }, { "epoch": 5.85, "learning_rate": 3.5475207634861943e-06, "loss": 0.0461, "step": 6385 }, { "epoch": 5.85, "learning_rate": 3.5446862293946018e-06, "loss": 0.0496, "step": 6386 }, { "epoch": 5.85, "learning_rate": 3.5418525842186923e-06, "loss": 0.0459, "step": 6387 }, { "epoch": 5.86, "learning_rate": 3.5390198283486653e-06, "loss": 0.0612, "step": 6388 }, { "epoch": 5.86, "learning_rate": 3.5361879621746043e-06, "loss": 0.0403, "step": 6389 }, { "epoch": 5.86, "learning_rate": 3.5333569860864646e-06, "loss": 0.0434, "step": 6390 }, { "epoch": 5.86, "learning_rate": 3.530526900474075e-06, "loss": 0.0486, "step": 6391 }, { "epoch": 5.86, "learning_rate": 3.5276977057271534e-06, "loss": 0.0512, "step": 6392 }, { "epoch": 5.86, "learning_rate": 3.5248694022352847e-06, "loss": 0.0444, "step": 6393 }, { "epoch": 5.86, "learning_rate": 3.5220419903879356e-06, "loss": 0.0497, "step": 6394 }, { "epoch": 5.86, "learning_rate": 3.519215470574445e-06, "loss": 0.0473, "step": 6395 }, { "epoch": 5.86, "learning_rate": 3.5163898431840404e-06, "loss": 0.0444, "step": 6396 }, { "epoch": 5.86, "learning_rate": 3.5135651086058155e-06, "loss": 0.0458, "step": 6397 }, { "epoch": 5.86, "learning_rate": 3.51074126722874e-06, "loss": 0.0469, "step": 6398 }, { "epoch": 5.87, "learning_rate": 3.507918319441672e-06, "loss": 0.0536, "step": 6399 }, { "epoch": 5.87, "learning_rate": 3.5050962656333376e-06, "loss": 0.049, "step": 6400 }, { "epoch": 5.87, "learning_rate": 3.502275106192341e-06, "loss": 0.0516, "step": 6401 }, { "epoch": 5.87, "learning_rate": 3.4994548415071615e-06, "loss": 0.0444, "step": 6402 }, { "epoch": 5.87, "learning_rate": 3.4966354719661645e-06, "loss": 0.0446, "step": 6403 }, { "epoch": 5.87, "learning_rate": 3.493816997957582e-06, "loss": 0.0504, "step": 6404 }, { "epoch": 5.87, "learning_rate": 3.4909994198695253e-06, "loss": 0.0452, "step": 6405 }, { "epoch": 5.87, "learning_rate": 3.48818273808998e-06, "loss": 0.0537, "step": 6406 }, { "epoch": 5.87, "learning_rate": 3.485366953006819e-06, "loss": 0.0505, "step": 6407 }, { "epoch": 5.87, "learning_rate": 3.4825520650077804e-06, "loss": 0.0449, "step": 6408 }, { "epoch": 5.87, "learning_rate": 3.47973807448048e-06, "loss": 0.045, "step": 6409 }, { "epoch": 5.88, "learning_rate": 3.4769249818124172e-06, "loss": 0.05, "step": 6410 }, { "epoch": 5.88, "learning_rate": 3.474112787390963e-06, "loss": 0.0437, "step": 6411 }, { "epoch": 5.88, "learning_rate": 3.4713014916033582e-06, "loss": 0.0519, "step": 6412 }, { "epoch": 5.88, "learning_rate": 3.4684910948367347e-06, "loss": 0.057, "step": 6413 }, { "epoch": 5.88, "learning_rate": 3.4656815974780855e-06, "loss": 0.0551, "step": 6414 }, { "epoch": 5.88, "learning_rate": 3.462872999914294e-06, "loss": 0.0454, "step": 6415 }, { "epoch": 5.88, "learning_rate": 3.4600653025321085e-06, "loss": 0.0548, "step": 6416 }, { "epoch": 5.88, "learning_rate": 3.4572585057181527e-06, "loss": 0.0438, "step": 6417 }, { "epoch": 5.88, "learning_rate": 3.4544526098589393e-06, "loss": 0.0484, "step": 6418 }, { "epoch": 5.88, "learning_rate": 3.451647615340844e-06, "loss": 0.0439, "step": 6419 }, { "epoch": 5.88, "learning_rate": 3.4488435225501204e-06, "loss": 0.0476, "step": 6420 }, { "epoch": 5.89, "learning_rate": 3.4460403318729064e-06, "loss": 0.0498, "step": 6421 }, { "epoch": 5.89, "learning_rate": 3.443238043695205e-06, "loss": 0.0487, "step": 6422 }, { "epoch": 5.89, "learning_rate": 3.440436658402898e-06, "loss": 0.0392, "step": 6423 }, { "epoch": 5.89, "learning_rate": 3.4376361763817513e-06, "loss": 0.0424, "step": 6424 }, { "epoch": 5.89, "learning_rate": 3.4348365980173948e-06, "loss": 0.0411, "step": 6425 }, { "epoch": 5.89, "learning_rate": 3.432037923695338e-06, "loss": 0.05, "step": 6426 }, { "epoch": 5.89, "learning_rate": 3.4292401538009658e-06, "loss": 0.0454, "step": 6427 }, { "epoch": 5.89, "learning_rate": 3.426443288719544e-06, "loss": 0.0517, "step": 6428 }, { "epoch": 5.89, "learning_rate": 3.423647328836207e-06, "loss": 0.0385, "step": 6429 }, { "epoch": 5.89, "learning_rate": 3.420852274535963e-06, "loss": 0.0421, "step": 6430 }, { "epoch": 5.89, "learning_rate": 3.418058126203705e-06, "loss": 0.0397, "step": 6431 }, { "epoch": 5.9, "learning_rate": 3.415264884224193e-06, "loss": 0.0506, "step": 6432 }, { "epoch": 5.9, "learning_rate": 3.4124725489820643e-06, "loss": 0.0442, "step": 6433 }, { "epoch": 5.9, "learning_rate": 3.4096811208618288e-06, "loss": 0.0372, "step": 6434 }, { "epoch": 5.9, "learning_rate": 3.406890600247881e-06, "loss": 0.0519, "step": 6435 }, { "epoch": 5.9, "learning_rate": 3.404100987524479e-06, "loss": 0.0463, "step": 6436 }, { "epoch": 5.9, "learning_rate": 3.4013122830757626e-06, "loss": 0.0545, "step": 6437 }, { "epoch": 5.9, "learning_rate": 3.3985244872857404e-06, "loss": 0.0497, "step": 6438 }, { "epoch": 5.9, "learning_rate": 3.3957376005383068e-06, "loss": 0.04, "step": 6439 }, { "epoch": 5.9, "learning_rate": 3.392951623217221e-06, "loss": 0.0554, "step": 6440 }, { "epoch": 5.9, "learning_rate": 3.390166555706117e-06, "loss": 0.0452, "step": 6441 }, { "epoch": 5.9, "learning_rate": 3.3873823983885135e-06, "loss": 0.0393, "step": 6442 }, { "epoch": 5.91, "learning_rate": 3.3845991516477927e-06, "loss": 0.0473, "step": 6443 }, { "epoch": 5.91, "learning_rate": 3.3818168158672182e-06, "loss": 0.0477, "step": 6444 }, { "epoch": 5.91, "learning_rate": 3.37903539142992e-06, "loss": 0.0517, "step": 6445 }, { "epoch": 5.91, "learning_rate": 3.3762548787189177e-06, "loss": 0.053, "step": 6446 }, { "epoch": 5.91, "learning_rate": 3.37347527811709e-06, "loss": 0.0535, "step": 6447 }, { "epoch": 5.91, "learning_rate": 3.370696590007194e-06, "loss": 0.0443, "step": 6448 }, { "epoch": 5.91, "learning_rate": 3.36791881477187e-06, "loss": 0.0474, "step": 6449 }, { "epoch": 5.91, "learning_rate": 3.3651419527936223e-06, "loss": 0.0504, "step": 6450 }, { "epoch": 5.91, "learning_rate": 3.3623660044548335e-06, "loss": 0.0554, "step": 6451 }, { "epoch": 5.91, "learning_rate": 3.359590970137756e-06, "loss": 0.0457, "step": 6452 }, { "epoch": 5.91, "learning_rate": 3.356816850224527e-06, "loss": 0.0475, "step": 6453 }, { "epoch": 5.92, "learning_rate": 3.354043645097147e-06, "loss": 0.051, "step": 6454 }, { "epoch": 5.92, "learning_rate": 3.351271355137492e-06, "loss": 0.0461, "step": 6455 }, { "epoch": 5.92, "learning_rate": 3.348499980727321e-06, "loss": 0.0473, "step": 6456 }, { "epoch": 5.92, "learning_rate": 3.3457295222482532e-06, "loss": 0.0499, "step": 6457 }, { "epoch": 5.92, "learning_rate": 3.342959980081796e-06, "loss": 0.0482, "step": 6458 }, { "epoch": 5.92, "learning_rate": 3.34019135460932e-06, "loss": 0.0521, "step": 6459 }, { "epoch": 5.92, "learning_rate": 3.33742364621207e-06, "loss": 0.0502, "step": 6460 }, { "epoch": 5.92, "learning_rate": 3.3346568552711745e-06, "loss": 0.0419, "step": 6461 }, { "epoch": 5.92, "learning_rate": 3.3318909821676214e-06, "loss": 0.0415, "step": 6462 }, { "epoch": 5.92, "learning_rate": 3.3291260272822857e-06, "loss": 0.0526, "step": 6463 }, { "epoch": 5.92, "learning_rate": 3.326361990995908e-06, "loss": 0.0498, "step": 6464 }, { "epoch": 5.93, "learning_rate": 3.3235988736891035e-06, "loss": 0.0388, "step": 6465 }, { "epoch": 5.93, "learning_rate": 3.320836675742358e-06, "loss": 0.045, "step": 6466 }, { "epoch": 5.93, "learning_rate": 3.3180753975360415e-06, "loss": 0.045, "step": 6467 }, { "epoch": 5.93, "learning_rate": 3.315315039450385e-06, "loss": 0.0446, "step": 6468 }, { "epoch": 5.93, "learning_rate": 3.3125556018655013e-06, "loss": 0.0438, "step": 6469 }, { "epoch": 5.93, "learning_rate": 3.309797085161367e-06, "loss": 0.0472, "step": 6470 }, { "epoch": 5.93, "learning_rate": 3.307039489717846e-06, "loss": 0.044, "step": 6471 }, { "epoch": 5.93, "learning_rate": 3.304282815914662e-06, "loss": 0.0527, "step": 6472 }, { "epoch": 5.93, "learning_rate": 3.3015270641314167e-06, "loss": 0.0439, "step": 6473 }, { "epoch": 5.93, "learning_rate": 3.298772234747589e-06, "loss": 0.0492, "step": 6474 }, { "epoch": 5.93, "learning_rate": 3.2960183281425263e-06, "loss": 0.0587, "step": 6475 }, { "epoch": 5.94, "learning_rate": 3.293265344695448e-06, "loss": 0.042, "step": 6476 }, { "epoch": 5.94, "learning_rate": 3.2905132847854448e-06, "loss": 0.0498, "step": 6477 }, { "epoch": 5.94, "learning_rate": 3.28776214879149e-06, "loss": 0.0467, "step": 6478 }, { "epoch": 5.94, "learning_rate": 3.2850119370924205e-06, "loss": 0.0455, "step": 6479 }, { "epoch": 5.94, "learning_rate": 3.2822626500669454e-06, "loss": 0.0457, "step": 6480 }, { "epoch": 5.94, "learning_rate": 3.2795142880936546e-06, "loss": 0.0435, "step": 6481 }, { "epoch": 5.94, "learning_rate": 3.276766851551004e-06, "loss": 0.0516, "step": 6482 }, { "epoch": 5.94, "learning_rate": 3.2740203408173233e-06, "loss": 0.0444, "step": 6483 }, { "epoch": 5.94, "learning_rate": 3.2712747562708115e-06, "loss": 0.0506, "step": 6484 }, { "epoch": 5.94, "learning_rate": 3.2685300982895496e-06, "loss": 0.0492, "step": 6485 }, { "epoch": 5.95, "learning_rate": 3.265786367251483e-06, "loss": 0.0338, "step": 6486 }, { "epoch": 5.95, "learning_rate": 3.2630435635344283e-06, "loss": 0.0478, "step": 6487 }, { "epoch": 5.95, "learning_rate": 3.260301687516083e-06, "loss": 0.0474, "step": 6488 }, { "epoch": 5.95, "learning_rate": 3.2575607395740083e-06, "loss": 0.0429, "step": 6489 }, { "epoch": 5.95, "learning_rate": 3.254820720085643e-06, "loss": 0.0453, "step": 6490 }, { "epoch": 5.95, "learning_rate": 3.25208162942829e-06, "loss": 0.0491, "step": 6491 }, { "epoch": 5.95, "learning_rate": 3.249343467979137e-06, "loss": 0.0506, "step": 6492 }, { "epoch": 5.95, "learning_rate": 3.246606236115235e-06, "loss": 0.0512, "step": 6493 }, { "epoch": 5.95, "learning_rate": 3.243869934213508e-06, "loss": 0.0459, "step": 6494 }, { "epoch": 5.95, "learning_rate": 3.2411345626507496e-06, "loss": 0.042, "step": 6495 }, { "epoch": 5.95, "learning_rate": 3.2384001218036353e-06, "loss": 0.0483, "step": 6496 }, { "epoch": 5.96, "learning_rate": 3.2356666120487012e-06, "loss": 0.0489, "step": 6497 }, { "epoch": 5.96, "learning_rate": 3.232934033762358e-06, "loss": 0.046, "step": 6498 }, { "epoch": 5.96, "learning_rate": 3.2302023873208966e-06, "loss": 0.0376, "step": 6499 }, { "epoch": 5.96, "learning_rate": 3.227471673100464e-06, "loss": 0.0425, "step": 6500 }, { "epoch": 5.96, "learning_rate": 3.224741891477096e-06, "loss": 0.0438, "step": 6501 }, { "epoch": 5.96, "learning_rate": 3.222013042826687e-06, "loss": 0.0542, "step": 6502 }, { "epoch": 5.96, "learning_rate": 3.2192851275250048e-06, "loss": 0.046, "step": 6503 }, { "epoch": 5.96, "learning_rate": 3.216558145947698e-06, "loss": 0.0448, "step": 6504 }, { "epoch": 5.96, "learning_rate": 3.2138320984702763e-06, "loss": 0.0446, "step": 6505 }, { "epoch": 5.96, "learning_rate": 3.2111069854681232e-06, "loss": 0.0541, "step": 6506 }, { "epoch": 5.96, "learning_rate": 3.208382807316498e-06, "loss": 0.0514, "step": 6507 }, { "epoch": 5.97, "learning_rate": 3.2056595643905274e-06, "loss": 0.0474, "step": 6508 }, { "epoch": 5.97, "learning_rate": 3.2029372570652052e-06, "loss": 0.0435, "step": 6509 }, { "epoch": 5.97, "learning_rate": 3.2002158857154076e-06, "loss": 0.0414, "step": 6510 }, { "epoch": 5.97, "learning_rate": 3.197495450715874e-06, "loss": 0.0447, "step": 6511 }, { "epoch": 5.97, "learning_rate": 3.1947759524412105e-06, "loss": 0.0487, "step": 6512 }, { "epoch": 5.97, "learning_rate": 3.1920573912659082e-06, "loss": 0.0527, "step": 6513 }, { "epoch": 5.97, "learning_rate": 3.189339767564318e-06, "loss": 0.0451, "step": 6514 }, { "epoch": 5.97, "learning_rate": 3.1866230817106626e-06, "loss": 0.0527, "step": 6515 }, { "epoch": 5.97, "learning_rate": 3.1839073340790373e-06, "loss": 0.048, "step": 6516 }, { "epoch": 5.97, "learning_rate": 3.181192525043413e-06, "loss": 0.0472, "step": 6517 }, { "epoch": 5.97, "learning_rate": 3.178478654977624e-06, "loss": 0.0431, "step": 6518 }, { "epoch": 5.98, "learning_rate": 3.175765724255374e-06, "loss": 0.0457, "step": 6519 }, { "epoch": 5.98, "learning_rate": 3.17305373325025e-06, "loss": 0.0373, "step": 6520 }, { "epoch": 5.98, "learning_rate": 3.1703426823356965e-06, "loss": 0.0563, "step": 6521 }, { "epoch": 5.98, "learning_rate": 3.1676325718850343e-06, "loss": 0.0509, "step": 6522 }, { "epoch": 5.98, "learning_rate": 3.164923402271449e-06, "loss": 0.0553, "step": 6523 }, { "epoch": 5.98, "learning_rate": 3.1622151738680096e-06, "loss": 0.0483, "step": 6524 }, { "epoch": 5.98, "learning_rate": 3.1595078870476413e-06, "loss": 0.0496, "step": 6525 }, { "epoch": 5.98, "learning_rate": 3.156801542183148e-06, "loss": 0.0447, "step": 6526 }, { "epoch": 5.98, "learning_rate": 3.154096139647196e-06, "loss": 0.0502, "step": 6527 }, { "epoch": 5.98, "learning_rate": 3.1513916798123355e-06, "loss": 0.0487, "step": 6528 }, { "epoch": 5.98, "learning_rate": 3.1486881630509736e-06, "loss": 0.046, "step": 6529 }, { "epoch": 5.99, "learning_rate": 3.1459855897353895e-06, "loss": 0.0447, "step": 6530 }, { "epoch": 5.99, "learning_rate": 3.1432839602377442e-06, "loss": 0.0487, "step": 6531 }, { "epoch": 5.99, "learning_rate": 3.1405832749300546e-06, "loss": 0.0478, "step": 6532 }, { "epoch": 5.99, "learning_rate": 3.137883534184213e-06, "loss": 0.0474, "step": 6533 }, { "epoch": 5.99, "learning_rate": 3.135184738371979e-06, "loss": 0.0525, "step": 6534 }, { "epoch": 5.99, "learning_rate": 3.132486887864992e-06, "loss": 0.0471, "step": 6535 }, { "epoch": 5.99, "learning_rate": 3.1297899830347477e-06, "loss": 0.0477, "step": 6536 }, { "epoch": 5.99, "learning_rate": 3.1270940242526183e-06, "loss": 0.0492, "step": 6537 }, { "epoch": 5.99, "learning_rate": 3.1243990118898505e-06, "loss": 0.0428, "step": 6538 }, { "epoch": 5.99, "learning_rate": 3.121704946317551e-06, "loss": 0.0448, "step": 6539 }, { "epoch": 5.99, "learning_rate": 3.1190118279067017e-06, "loss": 0.0415, "step": 6540 }, { "epoch": 6.0, "learning_rate": 3.1163196570281496e-06, "loss": 0.0503, "step": 6541 }, { "epoch": 6.0, "learning_rate": 3.11362843405262e-06, "loss": 0.0399, "step": 6542 }, { "epoch": 6.0, "learning_rate": 3.1109381593506972e-06, "loss": 0.0514, "step": 6543 }, { "epoch": 6.0, "learning_rate": 3.108248833292846e-06, "loss": 0.0452, "step": 6544 }, { "epoch": 6.0, "learning_rate": 3.1055604562493903e-06, "loss": 0.0483, "step": 6545 }, { "epoch": 6.0, "learning_rate": 3.1028730285905263e-06, "loss": 0.0446, "step": 6546 }, { "epoch": 6.0, "learning_rate": 3.100186550686324e-06, "loss": 0.033, "step": 6547 }, { "epoch": 6.0, "learning_rate": 3.0975010229067192e-06, "loss": 0.04, "step": 6548 }, { "epoch": 6.0, "learning_rate": 3.0948164456215124e-06, "loss": 0.0309, "step": 6549 }, { "epoch": 6.0, "learning_rate": 3.092132819200383e-06, "loss": 0.0374, "step": 6550 }, { "epoch": 6.0, "learning_rate": 3.089450144012871e-06, "loss": 0.0314, "step": 6551 }, { "epoch": 6.01, "learning_rate": 3.086768420428392e-06, "loss": 0.0311, "step": 6552 }, { "epoch": 6.01, "learning_rate": 3.0840876488162254e-06, "loss": 0.0325, "step": 6553 }, { "epoch": 6.01, "learning_rate": 3.0814078295455207e-06, "loss": 0.0317, "step": 6554 }, { "epoch": 6.01, "learning_rate": 3.0787289629852955e-06, "loss": 0.0324, "step": 6555 }, { "epoch": 6.01, "learning_rate": 3.0760510495044414e-06, "loss": 0.0402, "step": 6556 }, { "epoch": 6.01, "learning_rate": 3.0733740894717135e-06, "loss": 0.0335, "step": 6557 }, { "epoch": 6.01, "learning_rate": 3.0706980832557366e-06, "loss": 0.0354, "step": 6558 }, { "epoch": 6.01, "learning_rate": 3.0680230312250015e-06, "loss": 0.0276, "step": 6559 }, { "epoch": 6.01, "learning_rate": 3.065348933747876e-06, "loss": 0.0364, "step": 6560 }, { "epoch": 6.01, "learning_rate": 3.0626757911925898e-06, "loss": 0.0326, "step": 6561 }, { "epoch": 6.01, "learning_rate": 3.0600036039272385e-06, "loss": 0.0344, "step": 6562 }, { "epoch": 6.02, "learning_rate": 3.057332372319797e-06, "loss": 0.0283, "step": 6563 }, { "epoch": 6.02, "learning_rate": 3.0546620967380968e-06, "loss": 0.0351, "step": 6564 }, { "epoch": 6.02, "learning_rate": 3.0519927775498448e-06, "loss": 0.0289, "step": 6565 }, { "epoch": 6.02, "learning_rate": 3.0493244151226096e-06, "loss": 0.0323, "step": 6566 }, { "epoch": 6.02, "learning_rate": 3.04665700982384e-06, "loss": 0.0319, "step": 6567 }, { "epoch": 6.02, "learning_rate": 3.043990562020842e-06, "loss": 0.0297, "step": 6568 }, { "epoch": 6.02, "learning_rate": 3.0413250720807886e-06, "loss": 0.0394, "step": 6569 }, { "epoch": 6.02, "learning_rate": 3.0386605403707347e-06, "loss": 0.0383, "step": 6570 }, { "epoch": 6.02, "learning_rate": 3.0359969672575886e-06, "loss": 0.031, "step": 6571 }, { "epoch": 6.02, "learning_rate": 3.0333343531081337e-06, "loss": 0.0332, "step": 6572 }, { "epoch": 6.02, "learning_rate": 3.0306726982890156e-06, "loss": 0.0381, "step": 6573 }, { "epoch": 6.03, "learning_rate": 3.028012003166758e-06, "loss": 0.0295, "step": 6574 }, { "epoch": 6.03, "learning_rate": 3.025352268107745e-06, "loss": 0.0412, "step": 6575 }, { "epoch": 6.03, "learning_rate": 3.022693493478226e-06, "loss": 0.0354, "step": 6576 }, { "epoch": 6.03, "learning_rate": 3.0200356796443276e-06, "loss": 0.0397, "step": 6577 }, { "epoch": 6.03, "learning_rate": 3.017378826972036e-06, "loss": 0.0349, "step": 6578 }, { "epoch": 6.03, "learning_rate": 3.0147229358272077e-06, "loss": 0.03, "step": 6579 }, { "epoch": 6.03, "learning_rate": 3.0120680065755637e-06, "loss": 0.036, "step": 6580 }, { "epoch": 6.03, "learning_rate": 3.009414039582701e-06, "loss": 0.0386, "step": 6581 }, { "epoch": 6.03, "learning_rate": 3.0067610352140765e-06, "loss": 0.0323, "step": 6582 }, { "epoch": 6.03, "learning_rate": 3.0041089938350166e-06, "loss": 0.0329, "step": 6583 }, { "epoch": 6.03, "learning_rate": 3.0014579158107116e-06, "loss": 0.0336, "step": 6584 }, { "epoch": 6.04, "learning_rate": 2.9988078015062295e-06, "loss": 0.0343, "step": 6585 }, { "epoch": 6.04, "learning_rate": 2.9961586512864947e-06, "loss": 0.0303, "step": 6586 }, { "epoch": 6.04, "learning_rate": 2.9935104655163017e-06, "loss": 0.0313, "step": 6587 }, { "epoch": 6.04, "learning_rate": 2.990863244560318e-06, "loss": 0.0397, "step": 6588 }, { "epoch": 6.04, "learning_rate": 2.9882169887830682e-06, "loss": 0.0373, "step": 6589 }, { "epoch": 6.04, "learning_rate": 2.9855716985489556e-06, "loss": 0.0332, "step": 6590 }, { "epoch": 6.04, "learning_rate": 2.982927374222241e-06, "loss": 0.0362, "step": 6591 }, { "epoch": 6.04, "learning_rate": 2.9802840161670532e-06, "loss": 0.0351, "step": 6592 }, { "epoch": 6.04, "learning_rate": 2.977641624747396e-06, "loss": 0.0297, "step": 6593 }, { "epoch": 6.04, "learning_rate": 2.975000200327128e-06, "loss": 0.0266, "step": 6594 }, { "epoch": 6.04, "learning_rate": 2.972359743269988e-06, "loss": 0.0332, "step": 6595 }, { "epoch": 6.05, "learning_rate": 2.969720253939572e-06, "loss": 0.0325, "step": 6596 }, { "epoch": 6.05, "learning_rate": 2.967081732699345e-06, "loss": 0.0342, "step": 6597 }, { "epoch": 6.05, "learning_rate": 2.9644441799126346e-06, "loss": 0.0336, "step": 6598 }, { "epoch": 6.05, "learning_rate": 2.9618075959426484e-06, "loss": 0.03, "step": 6599 }, { "epoch": 6.05, "learning_rate": 2.9591719811524467e-06, "loss": 0.0338, "step": 6600 }, { "epoch": 6.05, "learning_rate": 2.9565373359049586e-06, "loss": 0.0279, "step": 6601 }, { "epoch": 6.05, "learning_rate": 2.953903660562989e-06, "loss": 0.0407, "step": 6602 }, { "epoch": 6.05, "learning_rate": 2.9512709554892007e-06, "loss": 0.0322, "step": 6603 }, { "epoch": 6.05, "learning_rate": 2.9486392210461225e-06, "loss": 0.0361, "step": 6604 }, { "epoch": 6.05, "learning_rate": 2.946008457596151e-06, "loss": 0.031, "step": 6605 }, { "epoch": 6.05, "learning_rate": 2.943378665501555e-06, "loss": 0.0314, "step": 6606 }, { "epoch": 6.06, "learning_rate": 2.940749845124463e-06, "loss": 0.0352, "step": 6607 }, { "epoch": 6.06, "learning_rate": 2.938121996826866e-06, "loss": 0.0308, "step": 6608 }, { "epoch": 6.06, "learning_rate": 2.935495120970635e-06, "loss": 0.0387, "step": 6609 }, { "epoch": 6.06, "learning_rate": 2.9328692179174933e-06, "loss": 0.03, "step": 6610 }, { "epoch": 6.06, "learning_rate": 2.930244288029037e-06, "loss": 0.0364, "step": 6611 }, { "epoch": 6.06, "learning_rate": 2.927620331666723e-06, "loss": 0.0412, "step": 6612 }, { "epoch": 6.06, "learning_rate": 2.924997349191884e-06, "loss": 0.0295, "step": 6613 }, { "epoch": 6.06, "learning_rate": 2.9223753409657087e-06, "loss": 0.0291, "step": 6614 }, { "epoch": 6.06, "learning_rate": 2.919754307349256e-06, "loss": 0.032, "step": 6615 }, { "epoch": 6.06, "learning_rate": 2.917134248703447e-06, "loss": 0.0353, "step": 6616 }, { "epoch": 6.07, "learning_rate": 2.914515165389078e-06, "loss": 0.026, "step": 6617 }, { "epoch": 6.07, "learning_rate": 2.9118970577668005e-06, "loss": 0.0384, "step": 6618 }, { "epoch": 6.07, "learning_rate": 2.9092799261971326e-06, "loss": 0.0273, "step": 6619 }, { "epoch": 6.07, "learning_rate": 2.906663771040468e-06, "loss": 0.0364, "step": 6620 }, { "epoch": 6.07, "learning_rate": 2.9040485926570538e-06, "loss": 0.0295, "step": 6621 }, { "epoch": 6.07, "learning_rate": 2.901434391407011e-06, "loss": 0.0327, "step": 6622 }, { "epoch": 6.07, "learning_rate": 2.8988211676503162e-06, "loss": 0.0249, "step": 6623 }, { "epoch": 6.07, "learning_rate": 2.896208921746828e-06, "loss": 0.0276, "step": 6624 }, { "epoch": 6.07, "learning_rate": 2.893597654056254e-06, "loss": 0.0314, "step": 6625 }, { "epoch": 6.07, "learning_rate": 2.8909873649381714e-06, "loss": 0.0355, "step": 6626 }, { "epoch": 6.07, "learning_rate": 2.8883780547520314e-06, "loss": 0.0321, "step": 6627 }, { "epoch": 6.08, "learning_rate": 2.88576972385714e-06, "loss": 0.0352, "step": 6628 }, { "epoch": 6.08, "learning_rate": 2.883162372612672e-06, "loss": 0.0284, "step": 6629 }, { "epoch": 6.08, "learning_rate": 2.8805560013776657e-06, "loss": 0.0278, "step": 6630 }, { "epoch": 6.08, "learning_rate": 2.87795061051103e-06, "loss": 0.0286, "step": 6631 }, { "epoch": 6.08, "learning_rate": 2.8753462003715295e-06, "loss": 0.0305, "step": 6632 }, { "epoch": 6.08, "learning_rate": 2.8727427713178058e-06, "loss": 0.0268, "step": 6633 }, { "epoch": 6.08, "learning_rate": 2.870140323708356e-06, "loss": 0.0351, "step": 6634 }, { "epoch": 6.08, "learning_rate": 2.8675388579015407e-06, "loss": 0.0401, "step": 6635 }, { "epoch": 6.08, "learning_rate": 2.864938374255596e-06, "loss": 0.0274, "step": 6636 }, { "epoch": 6.08, "learning_rate": 2.8623388731286097e-06, "loss": 0.0363, "step": 6637 }, { "epoch": 6.08, "learning_rate": 2.8597403548785464e-06, "loss": 0.0262, "step": 6638 }, { "epoch": 6.09, "learning_rate": 2.8571428198632288e-06, "loss": 0.0338, "step": 6639 }, { "epoch": 6.09, "learning_rate": 2.854546268440339e-06, "loss": 0.0292, "step": 6640 }, { "epoch": 6.09, "learning_rate": 2.85195070096744e-06, "loss": 0.0299, "step": 6641 }, { "epoch": 6.09, "learning_rate": 2.8493561178019413e-06, "loss": 0.0334, "step": 6642 }, { "epoch": 6.09, "learning_rate": 2.846762519301127e-06, "loss": 0.0307, "step": 6643 }, { "epoch": 6.09, "learning_rate": 2.8441699058221406e-06, "loss": 0.0348, "step": 6644 }, { "epoch": 6.09, "learning_rate": 2.841578277721998e-06, "loss": 0.0257, "step": 6645 }, { "epoch": 6.09, "learning_rate": 2.8389876353575707e-06, "loss": 0.0292, "step": 6646 }, { "epoch": 6.09, "learning_rate": 2.836397979085599e-06, "loss": 0.0322, "step": 6647 }, { "epoch": 6.09, "learning_rate": 2.8338093092626815e-06, "loss": 0.0344, "step": 6648 }, { "epoch": 6.09, "learning_rate": 2.831221626245293e-06, "loss": 0.0362, "step": 6649 }, { "epoch": 6.1, "learning_rate": 2.828634930389762e-06, "loss": 0.0347, "step": 6650 }, { "epoch": 6.1, "learning_rate": 2.8260492220522795e-06, "loss": 0.0343, "step": 6651 }, { "epoch": 6.1, "learning_rate": 2.823464501588913e-06, "loss": 0.0364, "step": 6652 }, { "epoch": 6.1, "learning_rate": 2.820880769355582e-06, "loss": 0.0348, "step": 6653 }, { "epoch": 6.1, "learning_rate": 2.8182980257080748e-06, "loss": 0.0311, "step": 6654 }, { "epoch": 6.1, "learning_rate": 2.815716271002039e-06, "loss": 0.0439, "step": 6655 }, { "epoch": 6.1, "learning_rate": 2.8131355055929965e-06, "loss": 0.0367, "step": 6656 }, { "epoch": 6.1, "learning_rate": 2.8105557298363227e-06, "loss": 0.0263, "step": 6657 }, { "epoch": 6.1, "learning_rate": 2.8079769440872582e-06, "loss": 0.033, "step": 6658 }, { "epoch": 6.1, "learning_rate": 2.805399148700915e-06, "loss": 0.0288, "step": 6659 }, { "epoch": 6.1, "learning_rate": 2.8028223440322586e-06, "loss": 0.0306, "step": 6660 }, { "epoch": 6.11, "learning_rate": 2.800246530436125e-06, "loss": 0.0365, "step": 6661 }, { "epoch": 6.11, "learning_rate": 2.7976717082672065e-06, "loss": 0.0388, "step": 6662 }, { "epoch": 6.11, "learning_rate": 2.795097877880071e-06, "loss": 0.0321, "step": 6663 }, { "epoch": 6.11, "learning_rate": 2.792525039629138e-06, "loss": 0.0309, "step": 6664 }, { "epoch": 6.11, "learning_rate": 2.789953193868694e-06, "loss": 0.0334, "step": 6665 }, { "epoch": 6.11, "learning_rate": 2.787382340952893e-06, "loss": 0.0314, "step": 6666 }, { "epoch": 6.11, "learning_rate": 2.784812481235749e-06, "loss": 0.0358, "step": 6667 }, { "epoch": 6.11, "learning_rate": 2.7822436150711374e-06, "loss": 0.0328, "step": 6668 }, { "epoch": 6.11, "learning_rate": 2.779675742812795e-06, "loss": 0.0375, "step": 6669 }, { "epoch": 6.11, "learning_rate": 2.777108864814333e-06, "loss": 0.0303, "step": 6670 }, { "epoch": 6.11, "learning_rate": 2.7745429814292147e-06, "loss": 0.0343, "step": 6671 }, { "epoch": 6.12, "learning_rate": 2.7719780930107687e-06, "loss": 0.033, "step": 6672 }, { "epoch": 6.12, "learning_rate": 2.769414199912186e-06, "loss": 0.0374, "step": 6673 }, { "epoch": 6.12, "learning_rate": 2.766851302486527e-06, "loss": 0.024, "step": 6674 }, { "epoch": 6.12, "learning_rate": 2.764289401086706e-06, "loss": 0.0309, "step": 6675 }, { "epoch": 6.12, "learning_rate": 2.7617284960655076e-06, "loss": 0.0368, "step": 6676 }, { "epoch": 6.12, "learning_rate": 2.759168587775576e-06, "loss": 0.0343, "step": 6677 }, { "epoch": 6.12, "learning_rate": 2.7566096765694128e-06, "loss": 0.0239, "step": 6678 }, { "epoch": 6.12, "learning_rate": 2.7540517627993947e-06, "loss": 0.0377, "step": 6679 }, { "epoch": 6.12, "learning_rate": 2.7514948468177506e-06, "loss": 0.0355, "step": 6680 }, { "epoch": 6.12, "learning_rate": 2.7489389289765733e-06, "loss": 0.037, "step": 6681 }, { "epoch": 6.12, "learning_rate": 2.7463840096278238e-06, "loss": 0.0379, "step": 6682 }, { "epoch": 6.13, "learning_rate": 2.7438300891233183e-06, "loss": 0.0283, "step": 6683 }, { "epoch": 6.13, "learning_rate": 2.7412771678147442e-06, "loss": 0.0355, "step": 6684 }, { "epoch": 6.13, "learning_rate": 2.738725246053644e-06, "loss": 0.0352, "step": 6685 }, { "epoch": 6.13, "learning_rate": 2.736174324191424e-06, "loss": 0.0386, "step": 6686 }, { "epoch": 6.13, "learning_rate": 2.7336244025793512e-06, "loss": 0.0361, "step": 6687 }, { "epoch": 6.13, "learning_rate": 2.7310754815685627e-06, "loss": 0.034, "step": 6688 }, { "epoch": 6.13, "learning_rate": 2.7285275615100503e-06, "loss": 0.0313, "step": 6689 }, { "epoch": 6.13, "learning_rate": 2.7259806427546654e-06, "loss": 0.0342, "step": 6690 }, { "epoch": 6.13, "learning_rate": 2.7234347256531345e-06, "loss": 0.0379, "step": 6691 }, { "epoch": 6.13, "learning_rate": 2.7208898105560332e-06, "loss": 0.0297, "step": 6692 }, { "epoch": 6.13, "learning_rate": 2.7183458978138045e-06, "loss": 0.0325, "step": 6693 }, { "epoch": 6.14, "learning_rate": 2.715802987776749e-06, "loss": 0.0348, "step": 6694 }, { "epoch": 6.14, "learning_rate": 2.7132610807950397e-06, "loss": 0.0326, "step": 6695 }, { "epoch": 6.14, "learning_rate": 2.7107201772187018e-06, "loss": 0.0305, "step": 6696 }, { "epoch": 6.14, "learning_rate": 2.7081802773976207e-06, "loss": 0.0428, "step": 6697 }, { "epoch": 6.14, "learning_rate": 2.7056413816815562e-06, "loss": 0.0302, "step": 6698 }, { "epoch": 6.14, "learning_rate": 2.703103490420117e-06, "loss": 0.0283, "step": 6699 }, { "epoch": 6.14, "learning_rate": 2.700566603962779e-06, "loss": 0.0397, "step": 6700 }, { "epoch": 6.14, "learning_rate": 2.6980307226588753e-06, "loss": 0.0349, "step": 6701 }, { "epoch": 6.14, "learning_rate": 2.695495846857611e-06, "loss": 0.0314, "step": 6702 }, { "epoch": 6.14, "learning_rate": 2.6929619769080418e-06, "loss": 0.0333, "step": 6703 }, { "epoch": 6.14, "learning_rate": 2.690429113159091e-06, "loss": 0.0364, "step": 6704 }, { "epoch": 6.15, "learning_rate": 2.6878972559595363e-06, "loss": 0.032, "step": 6705 }, { "epoch": 6.15, "learning_rate": 2.6853664056580287e-06, "loss": 0.0281, "step": 6706 }, { "epoch": 6.15, "learning_rate": 2.6828365626030716e-06, "loss": 0.0394, "step": 6707 }, { "epoch": 6.15, "learning_rate": 2.6803077271430276e-06, "loss": 0.0401, "step": 6708 }, { "epoch": 6.15, "learning_rate": 2.6777798996261306e-06, "loss": 0.0316, "step": 6709 }, { "epoch": 6.15, "learning_rate": 2.6752530804004684e-06, "loss": 0.0298, "step": 6710 }, { "epoch": 6.15, "learning_rate": 2.6727272698139906e-06, "loss": 0.0318, "step": 6711 }, { "epoch": 6.15, "learning_rate": 2.6702024682145047e-06, "loss": 0.0303, "step": 6712 }, { "epoch": 6.15, "learning_rate": 2.667678675949691e-06, "loss": 0.0326, "step": 6713 }, { "epoch": 6.15, "learning_rate": 2.6651558933670786e-06, "loss": 0.0295, "step": 6714 }, { "epoch": 6.15, "learning_rate": 2.6626341208140603e-06, "loss": 0.0357, "step": 6715 }, { "epoch": 6.16, "learning_rate": 2.660113358637898e-06, "loss": 0.0394, "step": 6716 }, { "epoch": 6.16, "learning_rate": 2.6575936071857035e-06, "loss": 0.0277, "step": 6717 }, { "epoch": 6.16, "learning_rate": 2.655074866804451e-06, "loss": 0.0315, "step": 6718 }, { "epoch": 6.16, "learning_rate": 2.6525571378409865e-06, "loss": 0.0317, "step": 6719 }, { "epoch": 6.16, "learning_rate": 2.650040420642003e-06, "loss": 0.0344, "step": 6720 }, { "epoch": 6.16, "learning_rate": 2.6475247155540583e-06, "loss": 0.0332, "step": 6721 }, { "epoch": 6.16, "learning_rate": 2.64501002292358e-06, "loss": 0.0279, "step": 6722 }, { "epoch": 6.16, "learning_rate": 2.642496343096842e-06, "loss": 0.0344, "step": 6723 }, { "epoch": 6.16, "learning_rate": 2.639983676419985e-06, "loss": 0.0289, "step": 6724 }, { "epoch": 6.16, "learning_rate": 2.637472023239016e-06, "loss": 0.0362, "step": 6725 }, { "epoch": 6.16, "learning_rate": 2.634961383899791e-06, "loss": 0.0361, "step": 6726 }, { "epoch": 6.17, "learning_rate": 2.632451758748038e-06, "loss": 0.0339, "step": 6727 }, { "epoch": 6.17, "learning_rate": 2.629943148129338e-06, "loss": 0.0389, "step": 6728 }, { "epoch": 6.17, "learning_rate": 2.6274355523891293e-06, "loss": 0.0289, "step": 6729 }, { "epoch": 6.17, "learning_rate": 2.6249289718727224e-06, "loss": 0.0375, "step": 6730 }, { "epoch": 6.17, "learning_rate": 2.6224234069252773e-06, "loss": 0.0378, "step": 6731 }, { "epoch": 6.17, "learning_rate": 2.619918857891818e-06, "loss": 0.0326, "step": 6732 }, { "epoch": 6.17, "learning_rate": 2.6174153251172253e-06, "loss": 0.0367, "step": 6733 }, { "epoch": 6.17, "learning_rate": 2.6149128089462485e-06, "loss": 0.0375, "step": 6734 }, { "epoch": 6.17, "learning_rate": 2.6124113097234894e-06, "loss": 0.0312, "step": 6735 }, { "epoch": 6.17, "learning_rate": 2.6099108277934105e-06, "loss": 0.0317, "step": 6736 }, { "epoch": 6.18, "learning_rate": 2.607411363500334e-06, "loss": 0.0402, "step": 6737 }, { "epoch": 6.18, "learning_rate": 2.604912917188448e-06, "loss": 0.0252, "step": 6738 }, { "epoch": 6.18, "learning_rate": 2.6024154892017938e-06, "loss": 0.0327, "step": 6739 }, { "epoch": 6.18, "learning_rate": 2.5999190798842712e-06, "loss": 0.0322, "step": 6740 }, { "epoch": 6.18, "learning_rate": 2.597423689579649e-06, "loss": 0.0293, "step": 6741 }, { "epoch": 6.18, "learning_rate": 2.594929318631547e-06, "loss": 0.0337, "step": 6742 }, { "epoch": 6.18, "learning_rate": 2.5924359673834485e-06, "loss": 0.0386, "step": 6743 }, { "epoch": 6.18, "learning_rate": 2.58994363617869e-06, "loss": 0.0266, "step": 6744 }, { "epoch": 6.18, "learning_rate": 2.587452325360481e-06, "loss": 0.0275, "step": 6745 }, { "epoch": 6.18, "learning_rate": 2.5849620352718775e-06, "loss": 0.0303, "step": 6746 }, { "epoch": 6.18, "learning_rate": 2.5824727662557992e-06, "loss": 0.0369, "step": 6747 }, { "epoch": 6.19, "learning_rate": 2.5799845186550286e-06, "loss": 0.0281, "step": 6748 }, { "epoch": 6.19, "learning_rate": 2.5774972928122045e-06, "loss": 0.037, "step": 6749 }, { "epoch": 6.19, "learning_rate": 2.5750110890698233e-06, "loss": 0.0318, "step": 6750 }, { "epoch": 6.19, "learning_rate": 2.57252590777024e-06, "loss": 0.0338, "step": 6751 }, { "epoch": 6.19, "learning_rate": 2.5700417492556783e-06, "loss": 0.0333, "step": 6752 }, { "epoch": 6.19, "learning_rate": 2.5675586138682108e-06, "loss": 0.0278, "step": 6753 }, { "epoch": 6.19, "learning_rate": 2.565076501949769e-06, "loss": 0.025, "step": 6754 }, { "epoch": 6.19, "learning_rate": 2.5625954138421528e-06, "loss": 0.037, "step": 6755 }, { "epoch": 6.19, "learning_rate": 2.5601153498870137e-06, "loss": 0.0342, "step": 6756 }, { "epoch": 6.19, "learning_rate": 2.5576363104258616e-06, "loss": 0.0278, "step": 6757 }, { "epoch": 6.19, "learning_rate": 2.5551582958000664e-06, "loss": 0.0282, "step": 6758 }, { "epoch": 6.2, "learning_rate": 2.552681306350864e-06, "loss": 0.0306, "step": 6759 }, { "epoch": 6.2, "learning_rate": 2.5502053424193383e-06, "loss": 0.0356, "step": 6760 }, { "epoch": 6.2, "learning_rate": 2.547730404346439e-06, "loss": 0.0287, "step": 6761 }, { "epoch": 6.2, "learning_rate": 2.5452564924729685e-06, "loss": 0.037, "step": 6762 }, { "epoch": 6.2, "learning_rate": 2.5427836071395973e-06, "loss": 0.0356, "step": 6763 }, { "epoch": 6.2, "learning_rate": 2.5403117486868432e-06, "loss": 0.0322, "step": 6764 }, { "epoch": 6.2, "learning_rate": 2.5378409174550957e-06, "loss": 0.0316, "step": 6765 }, { "epoch": 6.2, "learning_rate": 2.5353711137845893e-06, "loss": 0.0307, "step": 6766 }, { "epoch": 6.2, "learning_rate": 2.5329023380154238e-06, "loss": 0.0326, "step": 6767 }, { "epoch": 6.2, "learning_rate": 2.5304345904875605e-06, "loss": 0.0289, "step": 6768 }, { "epoch": 6.2, "learning_rate": 2.527967871540811e-06, "loss": 0.029, "step": 6769 }, { "epoch": 6.21, "learning_rate": 2.525502181514855e-06, "loss": 0.0386, "step": 6770 }, { "epoch": 6.21, "learning_rate": 2.5230375207492218e-06, "loss": 0.0221, "step": 6771 }, { "epoch": 6.21, "learning_rate": 2.5205738895833e-06, "loss": 0.0348, "step": 6772 }, { "epoch": 6.21, "learning_rate": 2.518111288356345e-06, "loss": 0.036, "step": 6773 }, { "epoch": 6.21, "learning_rate": 2.5156497174074613e-06, "loss": 0.0291, "step": 6774 }, { "epoch": 6.21, "learning_rate": 2.5131891770756133e-06, "loss": 0.0235, "step": 6775 }, { "epoch": 6.21, "learning_rate": 2.510729667699624e-06, "loss": 0.0299, "step": 6776 }, { "epoch": 6.21, "learning_rate": 2.5082711896181775e-06, "loss": 0.0382, "step": 6777 }, { "epoch": 6.21, "learning_rate": 2.505813743169815e-06, "loss": 0.0276, "step": 6778 }, { "epoch": 6.21, "learning_rate": 2.503357328692927e-06, "loss": 0.0289, "step": 6779 }, { "epoch": 6.21, "learning_rate": 2.5009019465257776e-06, "loss": 0.0249, "step": 6780 }, { "epoch": 6.22, "learning_rate": 2.498447597006476e-06, "loss": 0.0303, "step": 6781 }, { "epoch": 6.22, "learning_rate": 2.4959942804729933e-06, "loss": 0.0323, "step": 6782 }, { "epoch": 6.22, "learning_rate": 2.493541997263156e-06, "loss": 0.035, "step": 6783 }, { "epoch": 6.22, "learning_rate": 2.491090747714655e-06, "loss": 0.0304, "step": 6784 }, { "epoch": 6.22, "learning_rate": 2.488640532165034e-06, "loss": 0.0376, "step": 6785 }, { "epoch": 6.22, "learning_rate": 2.4861913509516913e-06, "loss": 0.0345, "step": 6786 }, { "epoch": 6.22, "learning_rate": 2.4837432044118914e-06, "loss": 0.0305, "step": 6787 }, { "epoch": 6.22, "learning_rate": 2.481296092882749e-06, "loss": 0.0327, "step": 6788 }, { "epoch": 6.22, "learning_rate": 2.4788500167012375e-06, "loss": 0.026, "step": 6789 }, { "epoch": 6.22, "learning_rate": 2.4764049762041874e-06, "loss": 0.0274, "step": 6790 }, { "epoch": 6.22, "learning_rate": 2.4739609717282933e-06, "loss": 0.0245, "step": 6791 }, { "epoch": 6.23, "learning_rate": 2.4715180036100972e-06, "loss": 0.0357, "step": 6792 }, { "epoch": 6.23, "learning_rate": 2.4690760721860053e-06, "loss": 0.0381, "step": 6793 }, { "epoch": 6.23, "learning_rate": 2.4666351777922746e-06, "loss": 0.0331, "step": 6794 }, { "epoch": 6.23, "learning_rate": 2.4641953207650304e-06, "loss": 0.0294, "step": 6795 }, { "epoch": 6.23, "learning_rate": 2.4617565014402445e-06, "loss": 0.0355, "step": 6796 }, { "epoch": 6.23, "learning_rate": 2.4593187201537463e-06, "loss": 0.0321, "step": 6797 }, { "epoch": 6.23, "learning_rate": 2.456881977241231e-06, "loss": 0.0331, "step": 6798 }, { "epoch": 6.23, "learning_rate": 2.4544462730382433e-06, "loss": 0.0297, "step": 6799 }, { "epoch": 6.23, "learning_rate": 2.452011607880187e-06, "loss": 0.0374, "step": 6800 }, { "epoch": 6.23, "learning_rate": 2.449577982102318e-06, "loss": 0.0293, "step": 6801 }, { "epoch": 6.23, "learning_rate": 2.4471453960397616e-06, "loss": 0.0289, "step": 6802 }, { "epoch": 6.24, "learning_rate": 2.444713850027488e-06, "loss": 0.0337, "step": 6803 }, { "epoch": 6.24, "learning_rate": 2.4422833444003256e-06, "loss": 0.0312, "step": 6804 }, { "epoch": 6.24, "learning_rate": 2.4398538794929683e-06, "loss": 0.0346, "step": 6805 }, { "epoch": 6.24, "learning_rate": 2.437425455639957e-06, "loss": 0.0335, "step": 6806 }, { "epoch": 6.24, "learning_rate": 2.4349980731756895e-06, "loss": 0.0369, "step": 6807 }, { "epoch": 6.24, "learning_rate": 2.432571732434431e-06, "loss": 0.0322, "step": 6808 }, { "epoch": 6.24, "learning_rate": 2.4301464337502912e-06, "loss": 0.034, "step": 6809 }, { "epoch": 6.24, "learning_rate": 2.427722177457239e-06, "loss": 0.0269, "step": 6810 }, { "epoch": 6.24, "learning_rate": 2.425298963889107e-06, "loss": 0.0309, "step": 6811 }, { "epoch": 6.24, "learning_rate": 2.4228767933795725e-06, "loss": 0.0355, "step": 6812 }, { "epoch": 6.24, "learning_rate": 2.4204556662621824e-06, "loss": 0.0283, "step": 6813 }, { "epoch": 6.25, "learning_rate": 2.4180355828703307e-06, "loss": 0.0332, "step": 6814 }, { "epoch": 6.25, "learning_rate": 2.4156165435372647e-06, "loss": 0.0406, "step": 6815 }, { "epoch": 6.25, "learning_rate": 2.4131985485961006e-06, "loss": 0.0349, "step": 6816 }, { "epoch": 6.25, "learning_rate": 2.4107815983798e-06, "loss": 0.0365, "step": 6817 }, { "epoch": 6.25, "learning_rate": 2.408365693221182e-06, "loss": 0.0317, "step": 6818 }, { "epoch": 6.25, "learning_rate": 2.405950833452928e-06, "loss": 0.0319, "step": 6819 }, { "epoch": 6.25, "learning_rate": 2.4035370194075703e-06, "loss": 0.0301, "step": 6820 }, { "epoch": 6.25, "learning_rate": 2.401124251417497e-06, "loss": 0.031, "step": 6821 }, { "epoch": 6.25, "learning_rate": 2.3987125298149503e-06, "loss": 0.0319, "step": 6822 }, { "epoch": 6.25, "learning_rate": 2.396301854932038e-06, "loss": 0.0344, "step": 6823 }, { "epoch": 6.25, "learning_rate": 2.3938922271007147e-06, "loss": 0.0326, "step": 6824 }, { "epoch": 6.26, "learning_rate": 2.3914836466527914e-06, "loss": 0.0367, "step": 6825 }, { "epoch": 6.26, "learning_rate": 2.3890761139199347e-06, "loss": 0.031, "step": 6826 }, { "epoch": 6.26, "learning_rate": 2.3866696292336754e-06, "loss": 0.031, "step": 6827 }, { "epoch": 6.26, "learning_rate": 2.3842641929253897e-06, "loss": 0.0322, "step": 6828 }, { "epoch": 6.26, "learning_rate": 2.3818598053263108e-06, "loss": 0.0299, "step": 6829 }, { "epoch": 6.26, "learning_rate": 2.3794564667675348e-06, "loss": 0.0324, "step": 6830 }, { "epoch": 6.26, "learning_rate": 2.3770541775800073e-06, "loss": 0.0352, "step": 6831 }, { "epoch": 6.26, "learning_rate": 2.3746529380945294e-06, "loss": 0.0306, "step": 6832 }, { "epoch": 6.26, "learning_rate": 2.372252748641757e-06, "loss": 0.03, "step": 6833 }, { "epoch": 6.26, "learning_rate": 2.369853609552207e-06, "loss": 0.0325, "step": 6834 }, { "epoch": 6.26, "learning_rate": 2.3674555211562468e-06, "loss": 0.0307, "step": 6835 }, { "epoch": 6.27, "learning_rate": 2.3650584837840974e-06, "loss": 0.0344, "step": 6836 }, { "epoch": 6.27, "learning_rate": 2.362662497765842e-06, "loss": 0.0297, "step": 6837 }, { "epoch": 6.27, "learning_rate": 2.360267563431413e-06, "loss": 0.03, "step": 6838 }, { "epoch": 6.27, "learning_rate": 2.3578736811106005e-06, "loss": 0.035, "step": 6839 }, { "epoch": 6.27, "learning_rate": 2.3554808511330463e-06, "loss": 0.0331, "step": 6840 }, { "epoch": 6.27, "learning_rate": 2.353089073828255e-06, "loss": 0.0362, "step": 6841 }, { "epoch": 6.27, "learning_rate": 2.3506983495255785e-06, "loss": 0.0313, "step": 6842 }, { "epoch": 6.27, "learning_rate": 2.348308678554224e-06, "loss": 0.0353, "step": 6843 }, { "epoch": 6.27, "learning_rate": 2.3459200612432633e-06, "loss": 0.0329, "step": 6844 }, { "epoch": 6.27, "learning_rate": 2.34353249792161e-06, "loss": 0.0262, "step": 6845 }, { "epoch": 6.27, "learning_rate": 2.341145988918041e-06, "loss": 0.0343, "step": 6846 }, { "epoch": 6.28, "learning_rate": 2.338760534561183e-06, "loss": 0.0304, "step": 6847 }, { "epoch": 6.28, "learning_rate": 2.3363761351795245e-06, "loss": 0.0336, "step": 6848 }, { "epoch": 6.28, "learning_rate": 2.3339927911014015e-06, "loss": 0.028, "step": 6849 }, { "epoch": 6.28, "learning_rate": 2.331610502655005e-06, "loss": 0.0262, "step": 6850 }, { "epoch": 6.28, "learning_rate": 2.3292292701683895e-06, "loss": 0.0282, "step": 6851 }, { "epoch": 6.28, "learning_rate": 2.3268490939694533e-06, "loss": 0.0368, "step": 6852 }, { "epoch": 6.28, "learning_rate": 2.3244699743859523e-06, "loss": 0.0354, "step": 6853 }, { "epoch": 6.28, "learning_rate": 2.322091911745503e-06, "loss": 0.0315, "step": 6854 }, { "epoch": 6.28, "learning_rate": 2.3197149063755677e-06, "loss": 0.0265, "step": 6855 }, { "epoch": 6.28, "learning_rate": 2.3173389586034702e-06, "loss": 0.0283, "step": 6856 }, { "epoch": 6.29, "learning_rate": 2.3149640687563834e-06, "loss": 0.0272, "step": 6857 }, { "epoch": 6.29, "learning_rate": 2.312590237161335e-06, "loss": 0.0319, "step": 6858 }, { "epoch": 6.29, "learning_rate": 2.3102174641452134e-06, "loss": 0.0376, "step": 6859 }, { "epoch": 6.29, "learning_rate": 2.307845750034753e-06, "loss": 0.0379, "step": 6860 }, { "epoch": 6.29, "learning_rate": 2.305475095156544e-06, "loss": 0.0303, "step": 6861 }, { "epoch": 6.29, "learning_rate": 2.3031054998370373e-06, "loss": 0.0351, "step": 6862 }, { "epoch": 6.29, "learning_rate": 2.3007369644025313e-06, "loss": 0.0276, "step": 6863 }, { "epoch": 6.29, "learning_rate": 2.2983694891791797e-06, "loss": 0.0284, "step": 6864 }, { "epoch": 6.29, "learning_rate": 2.2960030744929885e-06, "loss": 0.0343, "step": 6865 }, { "epoch": 6.29, "learning_rate": 2.2936377206698247e-06, "loss": 0.0277, "step": 6866 }, { "epoch": 6.29, "learning_rate": 2.291273428035402e-06, "loss": 0.031, "step": 6867 }, { "epoch": 6.3, "learning_rate": 2.288910196915288e-06, "loss": 0.0349, "step": 6868 }, { "epoch": 6.3, "learning_rate": 2.286548027634913e-06, "loss": 0.0314, "step": 6869 }, { "epoch": 6.3, "learning_rate": 2.284186920519549e-06, "loss": 0.0395, "step": 6870 }, { "epoch": 6.3, "learning_rate": 2.2818268758943297e-06, "loss": 0.0351, "step": 6871 }, { "epoch": 6.3, "learning_rate": 2.2794678940842376e-06, "loss": 0.0302, "step": 6872 }, { "epoch": 6.3, "learning_rate": 2.2771099754141157e-06, "loss": 0.0271, "step": 6873 }, { "epoch": 6.3, "learning_rate": 2.274753120208654e-06, "loss": 0.0266, "step": 6874 }, { "epoch": 6.3, "learning_rate": 2.272397328792396e-06, "loss": 0.0323, "step": 6875 }, { "epoch": 6.3, "learning_rate": 2.270042601489747e-06, "loss": 0.0346, "step": 6876 }, { "epoch": 6.3, "learning_rate": 2.2676889386249555e-06, "loss": 0.03, "step": 6877 }, { "epoch": 6.3, "learning_rate": 2.2653363405221296e-06, "loss": 0.0359, "step": 6878 }, { "epoch": 6.31, "learning_rate": 2.2629848075052253e-06, "loss": 0.0332, "step": 6879 }, { "epoch": 6.31, "learning_rate": 2.26063433989806e-06, "loss": 0.0354, "step": 6880 }, { "epoch": 6.31, "learning_rate": 2.2582849380243e-06, "loss": 0.0332, "step": 6881 }, { "epoch": 6.31, "learning_rate": 2.2559366022074624e-06, "loss": 0.0273, "step": 6882 }, { "epoch": 6.31, "learning_rate": 2.253589332770919e-06, "loss": 0.0356, "step": 6883 }, { "epoch": 6.31, "learning_rate": 2.2512431300379e-06, "loss": 0.0298, "step": 6884 }, { "epoch": 6.31, "learning_rate": 2.2488979943314816e-06, "loss": 0.0279, "step": 6885 }, { "epoch": 6.31, "learning_rate": 2.246553925974594e-06, "loss": 0.0341, "step": 6886 }, { "epoch": 6.31, "learning_rate": 2.244210925290027e-06, "loss": 0.0321, "step": 6887 }, { "epoch": 6.31, "learning_rate": 2.2418689926004166e-06, "loss": 0.0401, "step": 6888 }, { "epoch": 6.31, "learning_rate": 2.2395281282282532e-06, "loss": 0.0289, "step": 6889 }, { "epoch": 6.32, "learning_rate": 2.2371883324958776e-06, "loss": 0.0347, "step": 6890 }, { "epoch": 6.32, "learning_rate": 2.2348496057254943e-06, "loss": 0.0395, "step": 6891 }, { "epoch": 6.32, "learning_rate": 2.2325119482391466e-06, "loss": 0.0309, "step": 6892 }, { "epoch": 6.32, "learning_rate": 2.2301753603587373e-06, "loss": 0.0337, "step": 6893 }, { "epoch": 6.32, "learning_rate": 2.2278398424060256e-06, "loss": 0.0393, "step": 6894 }, { "epoch": 6.32, "learning_rate": 2.2255053947026164e-06, "loss": 0.0322, "step": 6895 }, { "epoch": 6.32, "learning_rate": 2.2231720175699657e-06, "loss": 0.0359, "step": 6896 }, { "epoch": 6.32, "learning_rate": 2.2208397113293943e-06, "loss": 0.0354, "step": 6897 }, { "epoch": 6.32, "learning_rate": 2.218508476302065e-06, "loss": 0.0349, "step": 6898 }, { "epoch": 6.32, "learning_rate": 2.2161783128089907e-06, "loss": 0.0322, "step": 6899 }, { "epoch": 6.32, "learning_rate": 2.2138492211710494e-06, "loss": 0.0327, "step": 6900 }, { "epoch": 6.33, "learning_rate": 2.2115212017089573e-06, "loss": 0.0353, "step": 6901 }, { "epoch": 6.33, "learning_rate": 2.209194254743295e-06, "loss": 0.0329, "step": 6902 }, { "epoch": 6.33, "learning_rate": 2.206868380594488e-06, "loss": 0.0309, "step": 6903 }, { "epoch": 6.33, "learning_rate": 2.204543579582813e-06, "loss": 0.0296, "step": 6904 }, { "epoch": 6.33, "learning_rate": 2.202219852028407e-06, "loss": 0.0311, "step": 6905 }, { "epoch": 6.33, "learning_rate": 2.199897198251251e-06, "loss": 0.0353, "step": 6906 }, { "epoch": 6.33, "learning_rate": 2.1975756185711782e-06, "loss": 0.0335, "step": 6907 }, { "epoch": 6.33, "learning_rate": 2.195255113307884e-06, "loss": 0.0313, "step": 6908 }, { "epoch": 6.33, "learning_rate": 2.1929356827809058e-06, "loss": 0.0342, "step": 6909 }, { "epoch": 6.33, "learning_rate": 2.1906173273096343e-06, "loss": 0.0319, "step": 6910 }, { "epoch": 6.33, "learning_rate": 2.188300047213313e-06, "loss": 0.0292, "step": 6911 }, { "epoch": 6.34, "learning_rate": 2.1859838428110437e-06, "loss": 0.0289, "step": 6912 }, { "epoch": 6.34, "learning_rate": 2.1836687144217705e-06, "loss": 0.0337, "step": 6913 }, { "epoch": 6.34, "learning_rate": 2.1813546623642933e-06, "loss": 0.0335, "step": 6914 }, { "epoch": 6.34, "learning_rate": 2.179041686957263e-06, "loss": 0.0268, "step": 6915 }, { "epoch": 6.34, "learning_rate": 2.176729788519186e-06, "loss": 0.0257, "step": 6916 }, { "epoch": 6.34, "learning_rate": 2.174418967368418e-06, "loss": 0.0337, "step": 6917 }, { "epoch": 6.34, "learning_rate": 2.17210922382316e-06, "loss": 0.032, "step": 6918 }, { "epoch": 6.34, "learning_rate": 2.169800558201477e-06, "loss": 0.0294, "step": 6919 }, { "epoch": 6.34, "learning_rate": 2.1674929708212778e-06, "loss": 0.0337, "step": 6920 }, { "epoch": 6.34, "learning_rate": 2.1651864620003217e-06, "loss": 0.0335, "step": 6921 }, { "epoch": 6.34, "learning_rate": 2.162881032056221e-06, "loss": 0.0314, "step": 6922 }, { "epoch": 6.35, "learning_rate": 2.160576681306443e-06, "loss": 0.0272, "step": 6923 }, { "epoch": 6.35, "learning_rate": 2.1582734100683044e-06, "loss": 0.0341, "step": 6924 }, { "epoch": 6.35, "learning_rate": 2.1559712186589664e-06, "loss": 0.0325, "step": 6925 }, { "epoch": 6.35, "learning_rate": 2.153670107395456e-06, "loss": 0.0337, "step": 6926 }, { "epoch": 6.35, "learning_rate": 2.1513700765946368e-06, "loss": 0.0326, "step": 6927 }, { "epoch": 6.35, "learning_rate": 2.1490711265732333e-06, "loss": 0.0307, "step": 6928 }, { "epoch": 6.35, "learning_rate": 2.146773257647812e-06, "loss": 0.0298, "step": 6929 }, { "epoch": 6.35, "learning_rate": 2.1444764701348044e-06, "loss": 0.0384, "step": 6930 }, { "epoch": 6.35, "learning_rate": 2.142180764350479e-06, "loss": 0.0305, "step": 6931 }, { "epoch": 6.35, "learning_rate": 2.1398861406109616e-06, "loss": 0.0343, "step": 6932 }, { "epoch": 6.35, "learning_rate": 2.1375925992322323e-06, "loss": 0.0283, "step": 6933 }, { "epoch": 6.36, "learning_rate": 2.1353001405301154e-06, "loss": 0.0397, "step": 6934 }, { "epoch": 6.36, "learning_rate": 2.1330087648202915e-06, "loss": 0.0308, "step": 6935 }, { "epoch": 6.36, "learning_rate": 2.130718472418284e-06, "loss": 0.0372, "step": 6936 }, { "epoch": 6.36, "learning_rate": 2.128429263639481e-06, "loss": 0.0328, "step": 6937 }, { "epoch": 6.36, "learning_rate": 2.1261411387991093e-06, "loss": 0.0324, "step": 6938 }, { "epoch": 6.36, "learning_rate": 2.123854098212248e-06, "loss": 0.0281, "step": 6939 }, { "epoch": 6.36, "learning_rate": 2.1215681421938337e-06, "loss": 0.0315, "step": 6940 }, { "epoch": 6.36, "learning_rate": 2.1192832710586487e-06, "loss": 0.0317, "step": 6941 }, { "epoch": 6.36, "learning_rate": 2.1169994851213217e-06, "loss": 0.0294, "step": 6942 }, { "epoch": 6.36, "learning_rate": 2.114716784696342e-06, "loss": 0.0329, "step": 6943 }, { "epoch": 6.36, "learning_rate": 2.1124351700980417e-06, "loss": 0.0303, "step": 6944 }, { "epoch": 6.37, "learning_rate": 2.1101546416406083e-06, "loss": 0.0343, "step": 6945 }, { "epoch": 6.37, "learning_rate": 2.107875199638075e-06, "loss": 0.0283, "step": 6946 }, { "epoch": 6.37, "learning_rate": 2.1055968444043263e-06, "loss": 0.0357, "step": 6947 }, { "epoch": 6.37, "learning_rate": 2.1033195762531033e-06, "loss": 0.0305, "step": 6948 }, { "epoch": 6.37, "learning_rate": 2.101043395497989e-06, "loss": 0.0317, "step": 6949 }, { "epoch": 6.37, "learning_rate": 2.0987683024524174e-06, "loss": 0.0328, "step": 6950 }, { "epoch": 6.37, "learning_rate": 2.096494297429682e-06, "loss": 0.0314, "step": 6951 }, { "epoch": 6.37, "learning_rate": 2.0942213807429168e-06, "loss": 0.0324, "step": 6952 }, { "epoch": 6.37, "learning_rate": 2.091949552705109e-06, "loss": 0.0296, "step": 6953 }, { "epoch": 6.37, "learning_rate": 2.0896788136290937e-06, "loss": 0.0299, "step": 6954 }, { "epoch": 6.37, "learning_rate": 2.087409163827564e-06, "loss": 0.0308, "step": 6955 }, { "epoch": 6.38, "learning_rate": 2.085140603613053e-06, "loss": 0.031, "step": 6956 }, { "epoch": 6.38, "learning_rate": 2.0828731332979466e-06, "loss": 0.0356, "step": 6957 }, { "epoch": 6.38, "learning_rate": 2.0806067531944875e-06, "loss": 0.0301, "step": 6958 }, { "epoch": 6.38, "learning_rate": 2.0783414636147602e-06, "loss": 0.0292, "step": 6959 }, { "epoch": 6.38, "learning_rate": 2.0760772648707016e-06, "loss": 0.0352, "step": 6960 }, { "epoch": 6.38, "learning_rate": 2.0738141572740964e-06, "loss": 0.0293, "step": 6961 }, { "epoch": 6.38, "learning_rate": 2.071552141136586e-06, "loss": 0.0264, "step": 6962 }, { "epoch": 6.38, "learning_rate": 2.0692912167696532e-06, "loss": 0.0321, "step": 6963 }, { "epoch": 6.38, "learning_rate": 2.0670313844846335e-06, "loss": 0.0279, "step": 6964 }, { "epoch": 6.38, "learning_rate": 2.064772644592715e-06, "loss": 0.0316, "step": 6965 }, { "epoch": 6.38, "learning_rate": 2.062514997404932e-06, "loss": 0.029, "step": 6966 }, { "epoch": 6.39, "learning_rate": 2.060258443232168e-06, "loss": 0.0316, "step": 6967 }, { "epoch": 6.39, "learning_rate": 2.0580029823851565e-06, "loss": 0.0354, "step": 6968 }, { "epoch": 6.39, "learning_rate": 2.0557486151744845e-06, "loss": 0.0305, "step": 6969 }, { "epoch": 6.39, "learning_rate": 2.053495341910583e-06, "loss": 0.0301, "step": 6970 }, { "epoch": 6.39, "learning_rate": 2.051243162903734e-06, "loss": 0.031, "step": 6971 }, { "epoch": 6.39, "learning_rate": 2.048992078464067e-06, "loss": 0.0375, "step": 6972 }, { "epoch": 6.39, "learning_rate": 2.0467420889015675e-06, "loss": 0.0358, "step": 6973 }, { "epoch": 6.39, "learning_rate": 2.044493194526064e-06, "loss": 0.0353, "step": 6974 }, { "epoch": 6.39, "learning_rate": 2.0422453956472333e-06, "loss": 0.0311, "step": 6975 }, { "epoch": 6.39, "learning_rate": 2.0399986925746075e-06, "loss": 0.0295, "step": 6976 }, { "epoch": 6.4, "learning_rate": 2.037753085617563e-06, "loss": 0.032, "step": 6977 }, { "epoch": 6.4, "learning_rate": 2.0355085750853277e-06, "loss": 0.0341, "step": 6978 }, { "epoch": 6.4, "learning_rate": 2.033265161286971e-06, "loss": 0.0279, "step": 6979 }, { "epoch": 6.4, "learning_rate": 2.0310228445314263e-06, "loss": 0.0321, "step": 6980 }, { "epoch": 6.4, "learning_rate": 2.0287816251274637e-06, "loss": 0.0261, "step": 6981 }, { "epoch": 6.4, "learning_rate": 2.026541503383702e-06, "loss": 0.0304, "step": 6982 }, { "epoch": 6.4, "learning_rate": 2.0243024796086186e-06, "loss": 0.0291, "step": 6983 }, { "epoch": 6.4, "learning_rate": 2.022064554110531e-06, "loss": 0.034, "step": 6984 }, { "epoch": 6.4, "learning_rate": 2.019827727197605e-06, "loss": 0.0321, "step": 6985 }, { "epoch": 6.4, "learning_rate": 2.017591999177865e-06, "loss": 0.0361, "step": 6986 }, { "epoch": 6.4, "learning_rate": 2.01535737035917e-06, "loss": 0.0371, "step": 6987 }, { "epoch": 6.41, "learning_rate": 2.0131238410492417e-06, "loss": 0.03, "step": 6988 }, { "epoch": 6.41, "learning_rate": 2.01089141155564e-06, "loss": 0.0337, "step": 6989 }, { "epoch": 6.41, "learning_rate": 2.008660082185775e-06, "loss": 0.0431, "step": 6990 }, { "epoch": 6.41, "learning_rate": 2.006429853246913e-06, "loss": 0.0345, "step": 6991 }, { "epoch": 6.41, "learning_rate": 2.0042007250461593e-06, "loss": 0.037, "step": 6992 }, { "epoch": 6.41, "learning_rate": 2.0019726978904697e-06, "loss": 0.0318, "step": 6993 }, { "epoch": 6.41, "learning_rate": 1.9997457720866554e-06, "loss": 0.0277, "step": 6994 }, { "epoch": 6.41, "learning_rate": 1.9975199479413666e-06, "loss": 0.0358, "step": 6995 }, { "epoch": 6.41, "learning_rate": 1.9952952257611045e-06, "loss": 0.033, "step": 6996 }, { "epoch": 6.41, "learning_rate": 1.993071605852226e-06, "loss": 0.0272, "step": 6997 }, { "epoch": 6.41, "learning_rate": 1.990849088520925e-06, "loss": 0.0313, "step": 6998 }, { "epoch": 6.42, "learning_rate": 1.988627674073251e-06, "loss": 0.0285, "step": 6999 }, { "epoch": 6.42, "learning_rate": 1.986407362815096e-06, "loss": 0.0336, "step": 7000 }, { "epoch": 6.42, "learning_rate": 1.9841881550522068e-06, "loss": 0.0325, "step": 7001 }, { "epoch": 6.42, "learning_rate": 1.9819700510901753e-06, "loss": 0.032, "step": 7002 }, { "epoch": 6.42, "learning_rate": 1.9797530512344385e-06, "loss": 0.0285, "step": 7003 }, { "epoch": 6.42, "learning_rate": 1.977537155790282e-06, "loss": 0.0327, "step": 7004 }, { "epoch": 6.42, "learning_rate": 1.9753223650628474e-06, "loss": 0.0302, "step": 7005 }, { "epoch": 6.42, "learning_rate": 1.973108679357113e-06, "loss": 0.027, "step": 7006 }, { "epoch": 6.42, "learning_rate": 1.970896098977909e-06, "loss": 0.0296, "step": 7007 }, { "epoch": 6.42, "learning_rate": 1.9686846242299195e-06, "loss": 0.041, "step": 7008 }, { "epoch": 6.42, "learning_rate": 1.9664742554176673e-06, "loss": 0.031, "step": 7009 }, { "epoch": 6.43, "learning_rate": 1.964264992845527e-06, "loss": 0.0291, "step": 7010 }, { "epoch": 6.43, "learning_rate": 1.9620568368177183e-06, "loss": 0.0333, "step": 7011 }, { "epoch": 6.43, "learning_rate": 1.959849787638317e-06, "loss": 0.035, "step": 7012 }, { "epoch": 6.43, "learning_rate": 1.9576438456112357e-06, "loss": 0.033, "step": 7013 }, { "epoch": 6.43, "learning_rate": 1.9554390110402376e-06, "loss": 0.0294, "step": 7014 }, { "epoch": 6.43, "learning_rate": 1.95323528422894e-06, "loss": 0.0332, "step": 7015 }, { "epoch": 6.43, "learning_rate": 1.9510326654808e-06, "loss": 0.0347, "step": 7016 }, { "epoch": 6.43, "learning_rate": 1.9488311550991235e-06, "loss": 0.0253, "step": 7017 }, { "epoch": 6.43, "learning_rate": 1.946630753387064e-06, "loss": 0.0292, "step": 7018 }, { "epoch": 6.43, "learning_rate": 1.944431460647629e-06, "loss": 0.0348, "step": 7019 }, { "epoch": 6.43, "learning_rate": 1.9422332771836627e-06, "loss": 0.0302, "step": 7020 }, { "epoch": 6.44, "learning_rate": 1.9400362032978593e-06, "loss": 0.0392, "step": 7021 }, { "epoch": 6.44, "learning_rate": 1.9378402392927696e-06, "loss": 0.0238, "step": 7022 }, { "epoch": 6.44, "learning_rate": 1.935645385470779e-06, "loss": 0.0387, "step": 7023 }, { "epoch": 6.44, "learning_rate": 1.9334516421341276e-06, "loss": 0.038, "step": 7024 }, { "epoch": 6.44, "learning_rate": 1.931259009584897e-06, "loss": 0.0316, "step": 7025 }, { "epoch": 6.44, "learning_rate": 1.929067488125024e-06, "loss": 0.0337, "step": 7026 }, { "epoch": 6.44, "learning_rate": 1.926877078056284e-06, "loss": 0.0335, "step": 7027 }, { "epoch": 6.44, "learning_rate": 1.924687779680302e-06, "loss": 0.0334, "step": 7028 }, { "epoch": 6.44, "learning_rate": 1.9224995932985545e-06, "loss": 0.0311, "step": 7029 }, { "epoch": 6.44, "learning_rate": 1.920312519212358e-06, "loss": 0.0285, "step": 7030 }, { "epoch": 6.44, "learning_rate": 1.9181265577228825e-06, "loss": 0.0349, "step": 7031 }, { "epoch": 6.45, "learning_rate": 1.9159417091311395e-06, "loss": 0.0309, "step": 7032 }, { "epoch": 6.45, "learning_rate": 1.9137579737379865e-06, "loss": 0.0286, "step": 7033 }, { "epoch": 6.45, "learning_rate": 1.911575351844135e-06, "loss": 0.035, "step": 7034 }, { "epoch": 6.45, "learning_rate": 1.909393843750136e-06, "loss": 0.0269, "step": 7035 }, { "epoch": 6.45, "learning_rate": 1.9072134497563876e-06, "loss": 0.0339, "step": 7036 }, { "epoch": 6.45, "learning_rate": 1.9050341701631403e-06, "loss": 0.0333, "step": 7037 }, { "epoch": 6.45, "learning_rate": 1.902856005270487e-06, "loss": 0.0346, "step": 7038 }, { "epoch": 6.45, "learning_rate": 1.9006789553783623e-06, "loss": 0.0363, "step": 7039 }, { "epoch": 6.45, "learning_rate": 1.8985030207865585e-06, "loss": 0.0215, "step": 7040 }, { "epoch": 6.45, "learning_rate": 1.896328201794706e-06, "loss": 0.0336, "step": 7041 }, { "epoch": 6.45, "learning_rate": 1.894154498702283e-06, "loss": 0.0317, "step": 7042 }, { "epoch": 6.46, "learning_rate": 1.8919819118086136e-06, "loss": 0.0338, "step": 7043 }, { "epoch": 6.46, "learning_rate": 1.8898104414128737e-06, "loss": 0.0378, "step": 7044 }, { "epoch": 6.46, "learning_rate": 1.8876400878140776e-06, "loss": 0.0287, "step": 7045 }, { "epoch": 6.46, "learning_rate": 1.8854708513110885e-06, "loss": 0.0347, "step": 7046 }, { "epoch": 6.46, "learning_rate": 1.8833027322026209e-06, "loss": 0.0326, "step": 7047 }, { "epoch": 6.46, "learning_rate": 1.8811357307872291e-06, "loss": 0.0342, "step": 7048 }, { "epoch": 6.46, "learning_rate": 1.8789698473633144e-06, "loss": 0.0316, "step": 7049 }, { "epoch": 6.46, "learning_rate": 1.8768050822291239e-06, "loss": 0.0284, "step": 7050 }, { "epoch": 6.46, "learning_rate": 1.874641435682758e-06, "loss": 0.0331, "step": 7051 }, { "epoch": 6.46, "learning_rate": 1.8724789080221516e-06, "loss": 0.0298, "step": 7052 }, { "epoch": 6.46, "learning_rate": 1.8703174995450923e-06, "loss": 0.0281, "step": 7053 }, { "epoch": 6.47, "learning_rate": 1.8681572105492153e-06, "loss": 0.0377, "step": 7054 }, { "epoch": 6.47, "learning_rate": 1.8659980413319956e-06, "loss": 0.0277, "step": 7055 }, { "epoch": 6.47, "learning_rate": 1.863839992190759e-06, "loss": 0.0344, "step": 7056 }, { "epoch": 6.47, "learning_rate": 1.8616830634226713e-06, "loss": 0.0324, "step": 7057 }, { "epoch": 6.47, "learning_rate": 1.8595272553247546e-06, "loss": 0.0363, "step": 7058 }, { "epoch": 6.47, "learning_rate": 1.857372568193866e-06, "loss": 0.0338, "step": 7059 }, { "epoch": 6.47, "learning_rate": 1.8552190023267113e-06, "loss": 0.0351, "step": 7060 }, { "epoch": 6.47, "learning_rate": 1.8530665580198438e-06, "loss": 0.0262, "step": 7061 }, { "epoch": 6.47, "learning_rate": 1.8509152355696625e-06, "loss": 0.0309, "step": 7062 }, { "epoch": 6.47, "learning_rate": 1.8487650352724118e-06, "loss": 0.0282, "step": 7063 }, { "epoch": 6.47, "learning_rate": 1.8466159574241748e-06, "loss": 0.031, "step": 7064 }, { "epoch": 6.48, "learning_rate": 1.8444680023208938e-06, "loss": 0.0354, "step": 7065 }, { "epoch": 6.48, "learning_rate": 1.8423211702583443e-06, "loss": 0.0319, "step": 7066 }, { "epoch": 6.48, "learning_rate": 1.840175461532152e-06, "loss": 0.0338, "step": 7067 }, { "epoch": 6.48, "learning_rate": 1.8380308764377841e-06, "loss": 0.0317, "step": 7068 }, { "epoch": 6.48, "learning_rate": 1.8358874152705619e-06, "loss": 0.0291, "step": 7069 }, { "epoch": 6.48, "learning_rate": 1.8337450783256439e-06, "loss": 0.0374, "step": 7070 }, { "epoch": 6.48, "learning_rate": 1.831603865898034e-06, "loss": 0.0333, "step": 7071 }, { "epoch": 6.48, "learning_rate": 1.8294637782825875e-06, "loss": 0.0303, "step": 7072 }, { "epoch": 6.48, "learning_rate": 1.8273248157739975e-06, "loss": 0.0314, "step": 7073 }, { "epoch": 6.48, "learning_rate": 1.8251869786668052e-06, "loss": 0.0303, "step": 7074 }, { "epoch": 6.48, "learning_rate": 1.823050267255402e-06, "loss": 0.0271, "step": 7075 }, { "epoch": 6.49, "learning_rate": 1.8209146818340117e-06, "loss": 0.0272, "step": 7076 }, { "epoch": 6.49, "learning_rate": 1.8187802226967177e-06, "loss": 0.0261, "step": 7077 }, { "epoch": 6.49, "learning_rate": 1.816646890137439e-06, "loss": 0.0319, "step": 7078 }, { "epoch": 6.49, "learning_rate": 1.8145146844499385e-06, "loss": 0.0278, "step": 7079 }, { "epoch": 6.49, "learning_rate": 1.8123836059278322e-06, "loss": 0.0337, "step": 7080 }, { "epoch": 6.49, "learning_rate": 1.8102536548645734e-06, "loss": 0.0361, "step": 7081 }, { "epoch": 6.49, "learning_rate": 1.8081248315534594e-06, "loss": 0.0317, "step": 7082 }, { "epoch": 6.49, "learning_rate": 1.8059971362876428e-06, "loss": 0.0376, "step": 7083 }, { "epoch": 6.49, "learning_rate": 1.803870569360109e-06, "loss": 0.0319, "step": 7084 }, { "epoch": 6.49, "learning_rate": 1.8017451310636903e-06, "loss": 0.0298, "step": 7085 }, { "epoch": 6.49, "learning_rate": 1.7996208216910704e-06, "loss": 0.0339, "step": 7086 }, { "epoch": 6.5, "learning_rate": 1.7974976415347723e-06, "loss": 0.0349, "step": 7087 }, { "epoch": 6.5, "learning_rate": 1.7953755908871617e-06, "loss": 0.0303, "step": 7088 }, { "epoch": 6.5, "learning_rate": 1.7932546700404506e-06, "loss": 0.0323, "step": 7089 }, { "epoch": 6.5, "learning_rate": 1.7911348792867e-06, "loss": 0.0318, "step": 7090 }, { "epoch": 6.5, "learning_rate": 1.7890162189178094e-06, "loss": 0.0317, "step": 7091 }, { "epoch": 6.5, "learning_rate": 1.7868986892255246e-06, "loss": 0.0318, "step": 7092 }, { "epoch": 6.5, "learning_rate": 1.7847822905014322e-06, "loss": 0.0337, "step": 7093 }, { "epoch": 6.5, "learning_rate": 1.7826670230369736e-06, "loss": 0.0287, "step": 7094 }, { "epoch": 6.5, "learning_rate": 1.7805528871234223e-06, "loss": 0.0296, "step": 7095 }, { "epoch": 6.5, "learning_rate": 1.7784398830519002e-06, "loss": 0.0292, "step": 7096 }, { "epoch": 6.51, "learning_rate": 1.7763280111133774e-06, "loss": 0.0363, "step": 7097 }, { "epoch": 6.51, "learning_rate": 1.774217271598665e-06, "loss": 0.0287, "step": 7098 }, { "epoch": 6.51, "learning_rate": 1.7721076647984158e-06, "loss": 0.034, "step": 7099 }, { "epoch": 6.51, "learning_rate": 1.7699991910031278e-06, "loss": 0.0336, "step": 7100 }, { "epoch": 6.51, "learning_rate": 1.7678918505031483e-06, "loss": 0.0278, "step": 7101 }, { "epoch": 6.51, "learning_rate": 1.7657856435886623e-06, "loss": 0.0283, "step": 7102 }, { "epoch": 6.51, "learning_rate": 1.7636805705496973e-06, "loss": 0.0303, "step": 7103 }, { "epoch": 6.51, "learning_rate": 1.761576631676134e-06, "loss": 0.0355, "step": 7104 }, { "epoch": 6.51, "learning_rate": 1.7594738272576883e-06, "loss": 0.0288, "step": 7105 }, { "epoch": 6.51, "learning_rate": 1.757372157583922e-06, "loss": 0.0312, "step": 7106 }, { "epoch": 6.51, "learning_rate": 1.755271622944238e-06, "loss": 0.0305, "step": 7107 }, { "epoch": 6.52, "learning_rate": 1.7531722236278936e-06, "loss": 0.0317, "step": 7108 }, { "epoch": 6.52, "learning_rate": 1.751073959923978e-06, "loss": 0.0267, "step": 7109 }, { "epoch": 6.52, "learning_rate": 1.748976832121425e-06, "loss": 0.0285, "step": 7110 }, { "epoch": 6.52, "learning_rate": 1.746880840509022e-06, "loss": 0.0292, "step": 7111 }, { "epoch": 6.52, "learning_rate": 1.74478598537539e-06, "loss": 0.0328, "step": 7112 }, { "epoch": 6.52, "learning_rate": 1.7426922670089962e-06, "loss": 0.0268, "step": 7113 }, { "epoch": 6.52, "learning_rate": 1.7405996856981512e-06, "loss": 0.0317, "step": 7114 }, { "epoch": 6.52, "learning_rate": 1.7385082417310118e-06, "loss": 0.0312, "step": 7115 }, { "epoch": 6.52, "learning_rate": 1.7364179353955745e-06, "loss": 0.0341, "step": 7116 }, { "epoch": 6.52, "learning_rate": 1.7343287669796794e-06, "loss": 0.0265, "step": 7117 }, { "epoch": 6.52, "learning_rate": 1.7322407367710158e-06, "loss": 0.0381, "step": 7118 }, { "epoch": 6.53, "learning_rate": 1.7301538450571054e-06, "loss": 0.0322, "step": 7119 }, { "epoch": 6.53, "learning_rate": 1.7280680921253246e-06, "loss": 0.0316, "step": 7120 }, { "epoch": 6.53, "learning_rate": 1.7259834782628859e-06, "loss": 0.0319, "step": 7121 }, { "epoch": 6.53, "learning_rate": 1.7239000037568432e-06, "loss": 0.0302, "step": 7122 }, { "epoch": 6.53, "learning_rate": 1.7218176688941035e-06, "loss": 0.0344, "step": 7123 }, { "epoch": 6.53, "learning_rate": 1.7197364739614075e-06, "loss": 0.0336, "step": 7124 }, { "epoch": 6.53, "learning_rate": 1.7176564192453383e-06, "loss": 0.0272, "step": 7125 }, { "epoch": 6.53, "learning_rate": 1.715577505032332e-06, "loss": 0.0304, "step": 7126 }, { "epoch": 6.53, "learning_rate": 1.7134997316086577e-06, "loss": 0.029, "step": 7127 }, { "epoch": 6.53, "learning_rate": 1.7114230992604286e-06, "loss": 0.0373, "step": 7128 }, { "epoch": 6.53, "learning_rate": 1.7093476082736082e-06, "loss": 0.0383, "step": 7129 }, { "epoch": 6.54, "learning_rate": 1.7072732589339958e-06, "loss": 0.0376, "step": 7130 }, { "epoch": 6.54, "learning_rate": 1.7052000515272349e-06, "loss": 0.0348, "step": 7131 }, { "epoch": 6.54, "learning_rate": 1.7031279863388084e-06, "loss": 0.0307, "step": 7132 }, { "epoch": 6.54, "learning_rate": 1.7010570636540535e-06, "loss": 0.0345, "step": 7133 }, { "epoch": 6.54, "learning_rate": 1.6989872837581378e-06, "loss": 0.0329, "step": 7134 }, { "epoch": 6.54, "learning_rate": 1.6969186469360731e-06, "loss": 0.0319, "step": 7135 }, { "epoch": 6.54, "learning_rate": 1.694851153472724e-06, "loss": 0.0383, "step": 7136 }, { "epoch": 6.54, "learning_rate": 1.6927848036527873e-06, "loss": 0.0257, "step": 7137 }, { "epoch": 6.54, "learning_rate": 1.6907195977608037e-06, "loss": 0.0314, "step": 7138 }, { "epoch": 6.54, "learning_rate": 1.6886555360811575e-06, "loss": 0.0272, "step": 7139 }, { "epoch": 6.54, "learning_rate": 1.6865926188980809e-06, "loss": 0.0339, "step": 7140 }, { "epoch": 6.55, "learning_rate": 1.684530846495639e-06, "loss": 0.0271, "step": 7141 }, { "epoch": 6.55, "learning_rate": 1.6824702191577447e-06, "loss": 0.0306, "step": 7142 }, { "epoch": 6.55, "learning_rate": 1.6804107371681544e-06, "loss": 0.0269, "step": 7143 }, { "epoch": 6.55, "learning_rate": 1.6783524008104646e-06, "loss": 0.0344, "step": 7144 }, { "epoch": 6.55, "learning_rate": 1.676295210368114e-06, "loss": 0.0326, "step": 7145 }, { "epoch": 6.55, "learning_rate": 1.6742391661243796e-06, "loss": 0.0286, "step": 7146 }, { "epoch": 6.55, "learning_rate": 1.6721842683623911e-06, "loss": 0.0327, "step": 7147 }, { "epoch": 6.55, "learning_rate": 1.6701305173651116e-06, "loss": 0.0307, "step": 7148 }, { "epoch": 6.55, "learning_rate": 1.6680779134153479e-06, "loss": 0.0307, "step": 7149 }, { "epoch": 6.55, "learning_rate": 1.6660264567957473e-06, "loss": 0.0327, "step": 7150 }, { "epoch": 6.55, "learning_rate": 1.663976147788806e-06, "loss": 0.0301, "step": 7151 }, { "epoch": 6.56, "learning_rate": 1.6619269866768572e-06, "loss": 0.0297, "step": 7152 }, { "epoch": 6.56, "learning_rate": 1.6598789737420706e-06, "loss": 0.0339, "step": 7153 }, { "epoch": 6.56, "learning_rate": 1.6578321092664718e-06, "loss": 0.0271, "step": 7154 }, { "epoch": 6.56, "learning_rate": 1.6557863935319153e-06, "loss": 0.0372, "step": 7155 }, { "epoch": 6.56, "learning_rate": 1.6537418268201033e-06, "loss": 0.0307, "step": 7156 }, { "epoch": 6.56, "learning_rate": 1.6516984094125754e-06, "loss": 0.0318, "step": 7157 }, { "epoch": 6.56, "learning_rate": 1.649656141590722e-06, "loss": 0.0325, "step": 7158 }, { "epoch": 6.56, "learning_rate": 1.6476150236357656e-06, "loss": 0.0372, "step": 7159 }, { "epoch": 6.56, "learning_rate": 1.645575055828773e-06, "loss": 0.0329, "step": 7160 }, { "epoch": 6.56, "learning_rate": 1.6435362384506581e-06, "loss": 0.0354, "step": 7161 }, { "epoch": 6.56, "learning_rate": 1.6414985717821675e-06, "loss": 0.0292, "step": 7162 }, { "epoch": 6.57, "learning_rate": 1.6394620561038987e-06, "loss": 0.0284, "step": 7163 }, { "epoch": 6.57, "learning_rate": 1.6374266916962832e-06, "loss": 0.04, "step": 7164 }, { "epoch": 6.57, "learning_rate": 1.6353924788395948e-06, "loss": 0.0317, "step": 7165 }, { "epoch": 6.57, "learning_rate": 1.6333594178139544e-06, "loss": 0.0395, "step": 7166 }, { "epoch": 6.57, "learning_rate": 1.6313275088993186e-06, "loss": 0.0259, "step": 7167 }, { "epoch": 6.57, "learning_rate": 1.6292967523754856e-06, "loss": 0.0362, "step": 7168 }, { "epoch": 6.57, "learning_rate": 1.6272671485220992e-06, "loss": 0.0342, "step": 7169 }, { "epoch": 6.57, "learning_rate": 1.6252386976186417e-06, "loss": 0.0282, "step": 7170 }, { "epoch": 6.57, "learning_rate": 1.6232113999444332e-06, "loss": 0.0374, "step": 7171 }, { "epoch": 6.57, "learning_rate": 1.6211852557786446e-06, "loss": 0.0278, "step": 7172 }, { "epoch": 6.57, "learning_rate": 1.619160265400277e-06, "loss": 0.0421, "step": 7173 }, { "epoch": 6.58, "learning_rate": 1.6171364290881808e-06, "loss": 0.0274, "step": 7174 }, { "epoch": 6.58, "learning_rate": 1.6151137471210387e-06, "loss": 0.034, "step": 7175 }, { "epoch": 6.58, "learning_rate": 1.6130922197773868e-06, "loss": 0.0283, "step": 7176 }, { "epoch": 6.58, "learning_rate": 1.6110718473355912e-06, "loss": 0.0317, "step": 7177 }, { "epoch": 6.58, "learning_rate": 1.6090526300738628e-06, "loss": 0.0285, "step": 7178 }, { "epoch": 6.58, "learning_rate": 1.6070345682702582e-06, "loss": 0.0323, "step": 7179 }, { "epoch": 6.58, "learning_rate": 1.6050176622026658e-06, "loss": 0.0311, "step": 7180 }, { "epoch": 6.58, "learning_rate": 1.6030019121488227e-06, "loss": 0.0285, "step": 7181 }, { "epoch": 6.58, "learning_rate": 1.6009873183862989e-06, "loss": 0.0323, "step": 7182 }, { "epoch": 6.58, "learning_rate": 1.5989738811925148e-06, "loss": 0.0308, "step": 7183 }, { "epoch": 6.58, "learning_rate": 1.5969616008447254e-06, "loss": 0.0319, "step": 7184 }, { "epoch": 6.59, "learning_rate": 1.5949504776200254e-06, "loss": 0.0385, "step": 7185 }, { "epoch": 6.59, "learning_rate": 1.5929405117953556e-06, "loss": 0.0302, "step": 7186 }, { "epoch": 6.59, "learning_rate": 1.5909317036474925e-06, "loss": 0.0321, "step": 7187 }, { "epoch": 6.59, "learning_rate": 1.5889240534530547e-06, "loss": 0.0337, "step": 7188 }, { "epoch": 6.59, "learning_rate": 1.5869175614885003e-06, "loss": 0.0304, "step": 7189 }, { "epoch": 6.59, "learning_rate": 1.584912228030132e-06, "loss": 0.0325, "step": 7190 }, { "epoch": 6.59, "learning_rate": 1.5829080533540896e-06, "loss": 0.0322, "step": 7191 }, { "epoch": 6.59, "learning_rate": 1.580905037736351e-06, "loss": 0.0341, "step": 7192 }, { "epoch": 6.59, "learning_rate": 1.578903181452741e-06, "loss": 0.0321, "step": 7193 }, { "epoch": 6.59, "learning_rate": 1.5769024847789205e-06, "loss": 0.0301, "step": 7194 }, { "epoch": 6.59, "learning_rate": 1.5749029479903888e-06, "loss": 0.0324, "step": 7195 }, { "epoch": 6.6, "learning_rate": 1.572904571362488e-06, "loss": 0.0279, "step": 7196 }, { "epoch": 6.6, "learning_rate": 1.5709073551704046e-06, "loss": 0.027, "step": 7197 }, { "epoch": 6.6, "learning_rate": 1.5689112996891576e-06, "loss": 0.0339, "step": 7198 }, { "epoch": 6.6, "learning_rate": 1.5669164051936092e-06, "loss": 0.0242, "step": 7199 }, { "epoch": 6.6, "learning_rate": 1.5649226719584654e-06, "loss": 0.0286, "step": 7200 }, { "epoch": 6.6, "learning_rate": 1.5629301002582676e-06, "loss": 0.0334, "step": 7201 }, { "epoch": 6.6, "learning_rate": 1.5609386903673973e-06, "loss": 0.0276, "step": 7202 }, { "epoch": 6.6, "learning_rate": 1.5589484425600776e-06, "loss": 0.0329, "step": 7203 }, { "epoch": 6.6, "learning_rate": 1.5569593571103747e-06, "loss": 0.0385, "step": 7204 }, { "epoch": 6.6, "learning_rate": 1.5549714342921873e-06, "loss": 0.03, "step": 7205 }, { "epoch": 6.6, "learning_rate": 1.552984674379262e-06, "loss": 0.0328, "step": 7206 }, { "epoch": 6.61, "learning_rate": 1.55099907764518e-06, "loss": 0.0267, "step": 7207 }, { "epoch": 6.61, "learning_rate": 1.5490146443633613e-06, "loss": 0.0291, "step": 7208 }, { "epoch": 6.61, "learning_rate": 1.5470313748070732e-06, "loss": 0.0296, "step": 7209 }, { "epoch": 6.61, "learning_rate": 1.5450492692494145e-06, "loss": 0.0285, "step": 7210 }, { "epoch": 6.61, "learning_rate": 1.5430683279633262e-06, "loss": 0.0323, "step": 7211 }, { "epoch": 6.61, "learning_rate": 1.5410885512215923e-06, "loss": 0.0315, "step": 7212 }, { "epoch": 6.61, "learning_rate": 1.5391099392968333e-06, "loss": 0.0296, "step": 7213 }, { "epoch": 6.61, "learning_rate": 1.537132492461506e-06, "loss": 0.0281, "step": 7214 }, { "epoch": 6.61, "learning_rate": 1.535156210987917e-06, "loss": 0.0326, "step": 7215 }, { "epoch": 6.61, "learning_rate": 1.533181095148203e-06, "loss": 0.0348, "step": 7216 }, { "epoch": 6.62, "learning_rate": 1.5312071452143407e-06, "loss": 0.0261, "step": 7217 }, { "epoch": 6.62, "learning_rate": 1.5292343614581528e-06, "loss": 0.0319, "step": 7218 }, { "epoch": 6.62, "learning_rate": 1.5272627441512965e-06, "loss": 0.0355, "step": 7219 }, { "epoch": 6.62, "learning_rate": 1.5252922935652693e-06, "loss": 0.0313, "step": 7220 }, { "epoch": 6.62, "learning_rate": 1.5233230099714047e-06, "loss": 0.0381, "step": 7221 }, { "epoch": 6.62, "learning_rate": 1.5213548936408828e-06, "loss": 0.0303, "step": 7222 }, { "epoch": 6.62, "learning_rate": 1.5193879448447192e-06, "loss": 0.0301, "step": 7223 }, { "epoch": 6.62, "learning_rate": 1.5174221638537644e-06, "loss": 0.0336, "step": 7224 }, { "epoch": 6.62, "learning_rate": 1.515457550938717e-06, "loss": 0.0306, "step": 7225 }, { "epoch": 6.62, "learning_rate": 1.513494106370108e-06, "loss": 0.0275, "step": 7226 }, { "epoch": 6.62, "learning_rate": 1.5115318304183103e-06, "loss": 0.028, "step": 7227 }, { "epoch": 6.63, "learning_rate": 1.5095707233535307e-06, "loss": 0.0372, "step": 7228 }, { "epoch": 6.63, "learning_rate": 1.5076107854458254e-06, "loss": 0.0251, "step": 7229 }, { "epoch": 6.63, "learning_rate": 1.5056520169650812e-06, "loss": 0.0314, "step": 7230 }, { "epoch": 6.63, "learning_rate": 1.5036944181810243e-06, "loss": 0.0362, "step": 7231 }, { "epoch": 6.63, "learning_rate": 1.5017379893632255e-06, "loss": 0.0318, "step": 7232 }, { "epoch": 6.63, "learning_rate": 1.4997827307810898e-06, "loss": 0.0338, "step": 7233 }, { "epoch": 6.63, "learning_rate": 1.4978286427038602e-06, "loss": 0.0301, "step": 7234 }, { "epoch": 6.63, "learning_rate": 1.4958757254006196e-06, "loss": 0.0332, "step": 7235 }, { "epoch": 6.63, "learning_rate": 1.4939239791402948e-06, "loss": 0.026, "step": 7236 }, { "epoch": 6.63, "learning_rate": 1.4919734041916435e-06, "loss": 0.0347, "step": 7237 }, { "epoch": 6.63, "learning_rate": 1.4900240008232668e-06, "loss": 0.0358, "step": 7238 }, { "epoch": 6.64, "learning_rate": 1.4880757693036019e-06, "loss": 0.0337, "step": 7239 }, { "epoch": 6.64, "learning_rate": 1.486128709900928e-06, "loss": 0.0301, "step": 7240 }, { "epoch": 6.64, "learning_rate": 1.484182822883361e-06, "loss": 0.0344, "step": 7241 }, { "epoch": 6.64, "learning_rate": 1.4822381085188519e-06, "loss": 0.0314, "step": 7242 }, { "epoch": 6.64, "learning_rate": 1.4802945670751978e-06, "loss": 0.038, "step": 7243 }, { "epoch": 6.64, "learning_rate": 1.4783521988200288e-06, "loss": 0.0319, "step": 7244 }, { "epoch": 6.64, "learning_rate": 1.4764110040208136e-06, "loss": 0.0332, "step": 7245 }, { "epoch": 6.64, "learning_rate": 1.474470982944859e-06, "loss": 0.0321, "step": 7246 }, { "epoch": 6.64, "learning_rate": 1.4725321358593159e-06, "loss": 0.0328, "step": 7247 }, { "epoch": 6.64, "learning_rate": 1.4705944630311675e-06, "loss": 0.0287, "step": 7248 }, { "epoch": 6.64, "learning_rate": 1.4686579647272337e-06, "loss": 0.0313, "step": 7249 }, { "epoch": 6.65, "learning_rate": 1.4667226412141811e-06, "loss": 0.0311, "step": 7250 }, { "epoch": 6.65, "learning_rate": 1.4647884927585054e-06, "loss": 0.031, "step": 7251 }, { "epoch": 6.65, "learning_rate": 1.462855519626548e-06, "loss": 0.0227, "step": 7252 }, { "epoch": 6.65, "learning_rate": 1.460923722084483e-06, "loss": 0.0293, "step": 7253 }, { "epoch": 6.65, "learning_rate": 1.4589931003983237e-06, "loss": 0.0315, "step": 7254 }, { "epoch": 6.65, "learning_rate": 1.4570636548339245e-06, "loss": 0.0285, "step": 7255 }, { "epoch": 6.65, "learning_rate": 1.4551353856569727e-06, "loss": 0.0261, "step": 7256 }, { "epoch": 6.65, "learning_rate": 1.4532082931330016e-06, "loss": 0.0315, "step": 7257 }, { "epoch": 6.65, "learning_rate": 1.4512823775273743e-06, "loss": 0.0293, "step": 7258 }, { "epoch": 6.65, "learning_rate": 1.4493576391052966e-06, "loss": 0.0276, "step": 7259 }, { "epoch": 6.65, "learning_rate": 1.4474340781318052e-06, "loss": 0.0265, "step": 7260 }, { "epoch": 6.66, "learning_rate": 1.4455116948717873e-06, "loss": 0.0325, "step": 7261 }, { "epoch": 6.66, "learning_rate": 1.4435904895899578e-06, "loss": 0.031, "step": 7262 }, { "epoch": 6.66, "learning_rate": 1.4416704625508726e-06, "loss": 0.0375, "step": 7263 }, { "epoch": 6.66, "learning_rate": 1.4397516140189216e-06, "loss": 0.0308, "step": 7264 }, { "epoch": 6.66, "learning_rate": 1.4378339442583422e-06, "loss": 0.0357, "step": 7265 }, { "epoch": 6.66, "learning_rate": 1.4359174535331998e-06, "loss": 0.0322, "step": 7266 }, { "epoch": 6.66, "learning_rate": 1.434002142107398e-06, "loss": 0.0312, "step": 7267 }, { "epoch": 6.66, "learning_rate": 1.4320880102446866e-06, "loss": 0.0312, "step": 7268 }, { "epoch": 6.66, "learning_rate": 1.4301750582086448e-06, "loss": 0.0334, "step": 7269 }, { "epoch": 6.66, "learning_rate": 1.4282632862626899e-06, "loss": 0.028, "step": 7270 }, { "epoch": 6.66, "learning_rate": 1.426352694670079e-06, "loss": 0.0295, "step": 7271 }, { "epoch": 6.67, "learning_rate": 1.424443283693908e-06, "loss": 0.0284, "step": 7272 }, { "epoch": 6.67, "learning_rate": 1.4225350535971095e-06, "loss": 0.0342, "step": 7273 }, { "epoch": 6.67, "learning_rate": 1.4206280046424458e-06, "loss": 0.0284, "step": 7274 }, { "epoch": 6.67, "learning_rate": 1.418722137092532e-06, "loss": 0.0306, "step": 7275 }, { "epoch": 6.67, "learning_rate": 1.4168174512098065e-06, "loss": 0.0359, "step": 7276 }, { "epoch": 6.67, "learning_rate": 1.414913947256551e-06, "loss": 0.0348, "step": 7277 }, { "epoch": 6.67, "learning_rate": 1.4130116254948811e-06, "loss": 0.0335, "step": 7278 }, { "epoch": 6.67, "learning_rate": 1.4111104861867564e-06, "loss": 0.0341, "step": 7279 }, { "epoch": 6.67, "learning_rate": 1.4092105295939685e-06, "loss": 0.0413, "step": 7280 }, { "epoch": 6.67, "learning_rate": 1.4073117559781435e-06, "loss": 0.0296, "step": 7281 }, { "epoch": 6.67, "learning_rate": 1.4054141656007536e-06, "loss": 0.0275, "step": 7282 }, { "epoch": 6.68, "learning_rate": 1.4035177587230998e-06, "loss": 0.0248, "step": 7283 }, { "epoch": 6.68, "learning_rate": 1.4016225356063218e-06, "loss": 0.037, "step": 7284 }, { "epoch": 6.68, "learning_rate": 1.3997284965113967e-06, "loss": 0.0311, "step": 7285 }, { "epoch": 6.68, "learning_rate": 1.3978356416991423e-06, "loss": 0.0298, "step": 7286 }, { "epoch": 6.68, "learning_rate": 1.3959439714302104e-06, "loss": 0.0318, "step": 7287 }, { "epoch": 6.68, "learning_rate": 1.3940534859650844e-06, "loss": 0.0342, "step": 7288 }, { "epoch": 6.68, "learning_rate": 1.3921641855640966e-06, "loss": 0.0257, "step": 7289 }, { "epoch": 6.68, "learning_rate": 1.3902760704874063e-06, "loss": 0.0274, "step": 7290 }, { "epoch": 6.68, "learning_rate": 1.3883891409950112e-06, "loss": 0.0318, "step": 7291 }, { "epoch": 6.68, "learning_rate": 1.3865033973467456e-06, "loss": 0.0289, "step": 7292 }, { "epoch": 6.68, "learning_rate": 1.3846188398022875e-06, "loss": 0.0277, "step": 7293 }, { "epoch": 6.69, "learning_rate": 1.3827354686211403e-06, "loss": 0.034, "step": 7294 }, { "epoch": 6.69, "learning_rate": 1.3808532840626543e-06, "loss": 0.0408, "step": 7295 }, { "epoch": 6.69, "learning_rate": 1.378972286386011e-06, "loss": 0.0295, "step": 7296 }, { "epoch": 6.69, "learning_rate": 1.3770924758502246e-06, "loss": 0.0278, "step": 7297 }, { "epoch": 6.69, "learning_rate": 1.3752138527141568e-06, "loss": 0.0355, "step": 7298 }, { "epoch": 6.69, "learning_rate": 1.373336417236495e-06, "loss": 0.0306, "step": 7299 }, { "epoch": 6.69, "learning_rate": 1.3714601696757713e-06, "loss": 0.0327, "step": 7300 }, { "epoch": 6.69, "learning_rate": 1.369585110290349e-06, "loss": 0.0294, "step": 7301 }, { "epoch": 6.69, "learning_rate": 1.3677112393384285e-06, "loss": 0.0374, "step": 7302 }, { "epoch": 6.69, "learning_rate": 1.3658385570780453e-06, "loss": 0.0327, "step": 7303 }, { "epoch": 6.69, "learning_rate": 1.3639670637670788e-06, "loss": 0.0372, "step": 7304 }, { "epoch": 6.7, "learning_rate": 1.3620967596632351e-06, "loss": 0.0301, "step": 7305 }, { "epoch": 6.7, "learning_rate": 1.3602276450240603e-06, "loss": 0.0308, "step": 7306 }, { "epoch": 6.7, "learning_rate": 1.3583597201069398e-06, "loss": 0.0328, "step": 7307 }, { "epoch": 6.7, "learning_rate": 1.3564929851690912e-06, "loss": 0.0261, "step": 7308 }, { "epoch": 6.7, "learning_rate": 1.3546274404675685e-06, "loss": 0.0271, "step": 7309 }, { "epoch": 6.7, "learning_rate": 1.352763086259261e-06, "loss": 0.031, "step": 7310 }, { "epoch": 6.7, "learning_rate": 1.3508999228009001e-06, "loss": 0.0307, "step": 7311 }, { "epoch": 6.7, "learning_rate": 1.349037950349047e-06, "loss": 0.0345, "step": 7312 }, { "epoch": 6.7, "learning_rate": 1.3471771691600976e-06, "loss": 0.03, "step": 7313 }, { "epoch": 6.7, "learning_rate": 1.345317579490293e-06, "loss": 0.0292, "step": 7314 }, { "epoch": 6.7, "learning_rate": 1.3434591815956998e-06, "loss": 0.0282, "step": 7315 }, { "epoch": 6.71, "learning_rate": 1.3416019757322264e-06, "loss": 0.0295, "step": 7316 }, { "epoch": 6.71, "learning_rate": 1.339745962155613e-06, "loss": 0.0355, "step": 7317 }, { "epoch": 6.71, "learning_rate": 1.3378911411214434e-06, "loss": 0.0354, "step": 7318 }, { "epoch": 6.71, "learning_rate": 1.3360375128851277e-06, "loss": 0.033, "step": 7319 }, { "epoch": 6.71, "learning_rate": 1.3341850777019149e-06, "loss": 0.0272, "step": 7320 }, { "epoch": 6.71, "learning_rate": 1.3323338358268955e-06, "loss": 0.0296, "step": 7321 }, { "epoch": 6.71, "learning_rate": 1.330483787514988e-06, "loss": 0.0262, "step": 7322 }, { "epoch": 6.71, "learning_rate": 1.3286349330209504e-06, "loss": 0.0325, "step": 7323 }, { "epoch": 6.71, "learning_rate": 1.3267872725993712e-06, "loss": 0.0288, "step": 7324 }, { "epoch": 6.71, "learning_rate": 1.3249408065046864e-06, "loss": 0.0221, "step": 7325 }, { "epoch": 6.71, "learning_rate": 1.3230955349911545e-06, "loss": 0.034, "step": 7326 }, { "epoch": 6.72, "learning_rate": 1.3212514583128767e-06, "loss": 0.0336, "step": 7327 }, { "epoch": 6.72, "learning_rate": 1.3194085767237842e-06, "loss": 0.0331, "step": 7328 }, { "epoch": 6.72, "learning_rate": 1.317566890477653e-06, "loss": 0.0252, "step": 7329 }, { "epoch": 6.72, "learning_rate": 1.3157263998280845e-06, "loss": 0.0388, "step": 7330 }, { "epoch": 6.72, "learning_rate": 1.3138871050285206e-06, "loss": 0.0276, "step": 7331 }, { "epoch": 6.72, "learning_rate": 1.312049006332239e-06, "loss": 0.0258, "step": 7332 }, { "epoch": 6.72, "learning_rate": 1.310212103992351e-06, "loss": 0.0259, "step": 7333 }, { "epoch": 6.72, "learning_rate": 1.3083763982618026e-06, "loss": 0.0325, "step": 7334 }, { "epoch": 6.72, "learning_rate": 1.3065418893933735e-06, "loss": 0.0279, "step": 7335 }, { "epoch": 6.72, "learning_rate": 1.304708577639685e-06, "loss": 0.0294, "step": 7336 }, { "epoch": 6.73, "learning_rate": 1.3028764632531876e-06, "loss": 0.0394, "step": 7337 }, { "epoch": 6.73, "learning_rate": 1.3010455464861704e-06, "loss": 0.0307, "step": 7338 }, { "epoch": 6.73, "learning_rate": 1.2992158275907541e-06, "loss": 0.0331, "step": 7339 }, { "epoch": 6.73, "learning_rate": 1.2973873068188958e-06, "loss": 0.0395, "step": 7340 }, { "epoch": 6.73, "learning_rate": 1.295559984422392e-06, "loss": 0.0323, "step": 7341 }, { "epoch": 6.73, "learning_rate": 1.2937338606528648e-06, "loss": 0.0292, "step": 7342 }, { "epoch": 6.73, "learning_rate": 1.291908935761783e-06, "loss": 0.0316, "step": 7343 }, { "epoch": 6.73, "learning_rate": 1.2900852100004414e-06, "loss": 0.0325, "step": 7344 }, { "epoch": 6.73, "learning_rate": 1.2882626836199685e-06, "loss": 0.0306, "step": 7345 }, { "epoch": 6.73, "learning_rate": 1.2864413568713385e-06, "loss": 0.0285, "step": 7346 }, { "epoch": 6.73, "learning_rate": 1.2846212300053495e-06, "loss": 0.0343, "step": 7347 }, { "epoch": 6.74, "learning_rate": 1.2828023032726377e-06, "loss": 0.0349, "step": 7348 }, { "epoch": 6.74, "learning_rate": 1.280984576923675e-06, "loss": 0.0352, "step": 7349 }, { "epoch": 6.74, "learning_rate": 1.2791680512087701e-06, "loss": 0.0265, "step": 7350 }, { "epoch": 6.74, "learning_rate": 1.2773527263780626e-06, "loss": 0.0294, "step": 7351 }, { "epoch": 6.74, "learning_rate": 1.2755386026815275e-06, "loss": 0.0253, "step": 7352 }, { "epoch": 6.74, "learning_rate": 1.2737256803689734e-06, "loss": 0.0298, "step": 7353 }, { "epoch": 6.74, "learning_rate": 1.2719139596900487e-06, "loss": 0.0407, "step": 7354 }, { "epoch": 6.74, "learning_rate": 1.2701034408942304e-06, "loss": 0.0327, "step": 7355 }, { "epoch": 6.74, "learning_rate": 1.2682941242308312e-06, "loss": 0.0303, "step": 7356 }, { "epoch": 6.74, "learning_rate": 1.2664860099490028e-06, "loss": 0.0333, "step": 7357 }, { "epoch": 6.74, "learning_rate": 1.2646790982977252e-06, "loss": 0.033, "step": 7358 }, { "epoch": 6.75, "learning_rate": 1.2628733895258172e-06, "loss": 0.0298, "step": 7359 }, { "epoch": 6.75, "learning_rate": 1.2610688838819262e-06, "loss": 0.0305, "step": 7360 }, { "epoch": 6.75, "learning_rate": 1.2592655816145437e-06, "loss": 0.033, "step": 7361 }, { "epoch": 6.75, "learning_rate": 1.2574634829719857e-06, "loss": 0.029, "step": 7362 }, { "epoch": 6.75, "learning_rate": 1.2556625882024065e-06, "loss": 0.033, "step": 7363 }, { "epoch": 6.75, "learning_rate": 1.2538628975537969e-06, "loss": 0.0286, "step": 7364 }, { "epoch": 6.75, "learning_rate": 1.25206441127398e-06, "loss": 0.0319, "step": 7365 }, { "epoch": 6.75, "learning_rate": 1.2502671296106095e-06, "loss": 0.0282, "step": 7366 }, { "epoch": 6.75, "learning_rate": 1.2484710528111766e-06, "loss": 0.0389, "step": 7367 }, { "epoch": 6.75, "learning_rate": 1.2466761811230099e-06, "loss": 0.0327, "step": 7368 }, { "epoch": 6.75, "learning_rate": 1.2448825147932652e-06, "loss": 0.0318, "step": 7369 }, { "epoch": 6.76, "learning_rate": 1.2430900540689361e-06, "loss": 0.0321, "step": 7370 }, { "epoch": 6.76, "learning_rate": 1.2412987991968518e-06, "loss": 0.0295, "step": 7371 }, { "epoch": 6.76, "learning_rate": 1.2395087504236714e-06, "loss": 0.0343, "step": 7372 }, { "epoch": 6.76, "learning_rate": 1.2377199079958913e-06, "loss": 0.0392, "step": 7373 }, { "epoch": 6.76, "learning_rate": 1.2359322721598367e-06, "loss": 0.0289, "step": 7374 }, { "epoch": 6.76, "learning_rate": 1.2341458431616759e-06, "loss": 0.0283, "step": 7375 }, { "epoch": 6.76, "learning_rate": 1.2323606212474014e-06, "loss": 0.0375, "step": 7376 }, { "epoch": 6.76, "learning_rate": 1.2305766066628433e-06, "loss": 0.0273, "step": 7377 }, { "epoch": 6.76, "learning_rate": 1.2287937996536691e-06, "loss": 0.0303, "step": 7378 }, { "epoch": 6.76, "learning_rate": 1.227012200465375e-06, "loss": 0.0349, "step": 7379 }, { "epoch": 6.76, "learning_rate": 1.2252318093432891e-06, "loss": 0.0271, "step": 7380 }, { "epoch": 6.77, "learning_rate": 1.2234526265325818e-06, "loss": 0.0332, "step": 7381 }, { "epoch": 6.77, "learning_rate": 1.2216746522782497e-06, "loss": 0.0369, "step": 7382 }, { "epoch": 6.77, "learning_rate": 1.219897886825122e-06, "loss": 0.0329, "step": 7383 }, { "epoch": 6.77, "learning_rate": 1.2181223304178703e-06, "loss": 0.0356, "step": 7384 }, { "epoch": 6.77, "learning_rate": 1.2163479833009896e-06, "loss": 0.0311, "step": 7385 }, { "epoch": 6.77, "learning_rate": 1.2145748457188122e-06, "loss": 0.0263, "step": 7386 }, { "epoch": 6.77, "learning_rate": 1.2128029179155087e-06, "loss": 0.0254, "step": 7387 }, { "epoch": 6.77, "learning_rate": 1.211032200135075e-06, "loss": 0.035, "step": 7388 }, { "epoch": 6.77, "learning_rate": 1.2092626926213457e-06, "loss": 0.0286, "step": 7389 }, { "epoch": 6.77, "learning_rate": 1.2074943956179884e-06, "loss": 0.0272, "step": 7390 }, { "epoch": 6.77, "learning_rate": 1.2057273093685006e-06, "loss": 0.0251, "step": 7391 }, { "epoch": 6.78, "learning_rate": 1.203961434116213e-06, "loss": 0.0284, "step": 7392 }, { "epoch": 6.78, "learning_rate": 1.202196770104298e-06, "loss": 0.0325, "step": 7393 }, { "epoch": 6.78, "learning_rate": 1.200433317575751e-06, "loss": 0.0289, "step": 7394 }, { "epoch": 6.78, "learning_rate": 1.1986710767734044e-06, "loss": 0.0336, "step": 7395 }, { "epoch": 6.78, "learning_rate": 1.1969100479399254e-06, "loss": 0.0361, "step": 7396 }, { "epoch": 6.78, "learning_rate": 1.1951502313178142e-06, "loss": 0.0327, "step": 7397 }, { "epoch": 6.78, "learning_rate": 1.1933916271493996e-06, "loss": 0.0293, "step": 7398 }, { "epoch": 6.78, "learning_rate": 1.1916342356768462e-06, "loss": 0.0323, "step": 7399 }, { "epoch": 6.78, "learning_rate": 1.1898780571421554e-06, "loss": 0.0295, "step": 7400 }, { "epoch": 6.78, "learning_rate": 1.1881230917871567e-06, "loss": 0.0305, "step": 7401 }, { "epoch": 6.78, "learning_rate": 1.1863693398535115e-06, "loss": 0.0307, "step": 7402 }, { "epoch": 6.79, "learning_rate": 1.1846168015827198e-06, "loss": 0.0377, "step": 7403 }, { "epoch": 6.79, "learning_rate": 1.182865477216112e-06, "loss": 0.0374, "step": 7404 }, { "epoch": 6.79, "learning_rate": 1.1811153669948472e-06, "loss": 0.0262, "step": 7405 }, { "epoch": 6.79, "learning_rate": 1.1793664711599218e-06, "loss": 0.0372, "step": 7406 }, { "epoch": 6.79, "learning_rate": 1.1776187899521652e-06, "loss": 0.0362, "step": 7407 }, { "epoch": 6.79, "learning_rate": 1.1758723236122383e-06, "loss": 0.0317, "step": 7408 }, { "epoch": 6.79, "learning_rate": 1.174127072380632e-06, "loss": 0.038, "step": 7409 }, { "epoch": 6.79, "learning_rate": 1.1723830364976751e-06, "loss": 0.0305, "step": 7410 }, { "epoch": 6.79, "learning_rate": 1.1706402162035268e-06, "loss": 0.0337, "step": 7411 }, { "epoch": 6.79, "learning_rate": 1.1688986117381772e-06, "loss": 0.0242, "step": 7412 }, { "epoch": 6.79, "learning_rate": 1.1671582233414491e-06, "loss": 0.0325, "step": 7413 }, { "epoch": 6.8, "learning_rate": 1.1654190512530016e-06, "loss": 0.0305, "step": 7414 }, { "epoch": 6.8, "learning_rate": 1.1636810957123245e-06, "loss": 0.0322, "step": 7415 }, { "epoch": 6.8, "learning_rate": 1.1619443569587363e-06, "loss": 0.0271, "step": 7416 }, { "epoch": 6.8, "learning_rate": 1.160208835231391e-06, "loss": 0.0267, "step": 7417 }, { "epoch": 6.8, "learning_rate": 1.158474530769279e-06, "loss": 0.029, "step": 7418 }, { "epoch": 6.8, "learning_rate": 1.1567414438112157e-06, "loss": 0.0296, "step": 7419 }, { "epoch": 6.8, "learning_rate": 1.1550095745958523e-06, "loss": 0.0284, "step": 7420 }, { "epoch": 6.8, "learning_rate": 1.1532789233616747e-06, "loss": 0.0312, "step": 7421 }, { "epoch": 6.8, "learning_rate": 1.151549490346997e-06, "loss": 0.0319, "step": 7422 }, { "epoch": 6.8, "learning_rate": 1.1498212757899675e-06, "loss": 0.026, "step": 7423 }, { "epoch": 6.8, "learning_rate": 1.1480942799285643e-06, "loss": 0.0305, "step": 7424 }, { "epoch": 6.81, "learning_rate": 1.1463685030006045e-06, "loss": 0.0243, "step": 7425 }, { "epoch": 6.81, "learning_rate": 1.1446439452437275e-06, "loss": 0.0316, "step": 7426 }, { "epoch": 6.81, "learning_rate": 1.1429206068954135e-06, "loss": 0.0272, "step": 7427 }, { "epoch": 6.81, "learning_rate": 1.1411984881929717e-06, "loss": 0.0417, "step": 7428 }, { "epoch": 6.81, "learning_rate": 1.1394775893735387e-06, "loss": 0.034, "step": 7429 }, { "epoch": 6.81, "learning_rate": 1.137757910674092e-06, "loss": 0.0359, "step": 7430 }, { "epoch": 6.81, "learning_rate": 1.136039452331432e-06, "loss": 0.0351, "step": 7431 }, { "epoch": 6.81, "learning_rate": 1.1343222145822009e-06, "loss": 0.0341, "step": 7432 }, { "epoch": 6.81, "learning_rate": 1.1326061976628644e-06, "loss": 0.0295, "step": 7433 }, { "epoch": 6.81, "learning_rate": 1.1308914018097206e-06, "loss": 0.0298, "step": 7434 }, { "epoch": 6.81, "learning_rate": 1.129177827258906e-06, "loss": 0.0374, "step": 7435 }, { "epoch": 6.82, "learning_rate": 1.1274654742463842e-06, "loss": 0.03, "step": 7436 }, { "epoch": 6.82, "learning_rate": 1.1257543430079498e-06, "loss": 0.0304, "step": 7437 }, { "epoch": 6.82, "learning_rate": 1.124044433779229e-06, "loss": 0.0296, "step": 7438 }, { "epoch": 6.82, "learning_rate": 1.1223357467956853e-06, "loss": 0.0354, "step": 7439 }, { "epoch": 6.82, "learning_rate": 1.1206282822926084e-06, "loss": 0.0352, "step": 7440 }, { "epoch": 6.82, "learning_rate": 1.1189220405051204e-06, "loss": 0.0262, "step": 7441 }, { "epoch": 6.82, "learning_rate": 1.1172170216681755e-06, "loss": 0.0338, "step": 7442 }, { "epoch": 6.82, "learning_rate": 1.1155132260165613e-06, "loss": 0.0348, "step": 7443 }, { "epoch": 6.82, "learning_rate": 1.1138106537848946e-06, "loss": 0.0258, "step": 7444 }, { "epoch": 6.82, "learning_rate": 1.1121093052076238e-06, "loss": 0.0357, "step": 7445 }, { "epoch": 6.82, "learning_rate": 1.110409180519032e-06, "loss": 0.0353, "step": 7446 }, { "epoch": 6.83, "learning_rate": 1.1087102799532301e-06, "loss": 0.0298, "step": 7447 }, { "epoch": 6.83, "learning_rate": 1.1070126037441608e-06, "loss": 0.0269, "step": 7448 }, { "epoch": 6.83, "learning_rate": 1.1053161521255984e-06, "loss": 0.0278, "step": 7449 }, { "epoch": 6.83, "learning_rate": 1.1036209253311525e-06, "loss": 0.0334, "step": 7450 }, { "epoch": 6.83, "learning_rate": 1.1019269235942588e-06, "loss": 0.0362, "step": 7451 }, { "epoch": 6.83, "learning_rate": 1.1002341471481848e-06, "loss": 0.0295, "step": 7452 }, { "epoch": 6.83, "learning_rate": 1.0985425962260342e-06, "loss": 0.0298, "step": 7453 }, { "epoch": 6.83, "learning_rate": 1.0968522710607376e-06, "loss": 0.0284, "step": 7454 }, { "epoch": 6.83, "learning_rate": 1.0951631718850563e-06, "loss": 0.0327, "step": 7455 }, { "epoch": 6.83, "learning_rate": 1.0934752989315833e-06, "loss": 0.0298, "step": 7456 }, { "epoch": 6.84, "learning_rate": 1.091788652432747e-06, "loss": 0.0323, "step": 7457 }, { "epoch": 6.84, "learning_rate": 1.0901032326208017e-06, "loss": 0.0367, "step": 7458 }, { "epoch": 6.84, "learning_rate": 1.0884190397278327e-06, "loss": 0.0337, "step": 7459 }, { "epoch": 6.84, "learning_rate": 1.0867360739857623e-06, "loss": 0.0249, "step": 7460 }, { "epoch": 6.84, "learning_rate": 1.0850543356263388e-06, "loss": 0.0257, "step": 7461 }, { "epoch": 6.84, "learning_rate": 1.0833738248811421e-06, "loss": 0.0352, "step": 7462 }, { "epoch": 6.84, "learning_rate": 1.0816945419815794e-06, "loss": 0.0339, "step": 7463 }, { "epoch": 6.84, "learning_rate": 1.0800164871588992e-06, "loss": 0.0362, "step": 7464 }, { "epoch": 6.84, "learning_rate": 1.0783396606441721e-06, "loss": 0.0314, "step": 7465 }, { "epoch": 6.84, "learning_rate": 1.0766640626683e-06, "loss": 0.0267, "step": 7466 }, { "epoch": 6.84, "learning_rate": 1.0749896934620218e-06, "loss": 0.0327, "step": 7467 }, { "epoch": 6.85, "learning_rate": 1.0733165532558998e-06, "loss": 0.0278, "step": 7468 }, { "epoch": 6.85, "learning_rate": 1.0716446422803306e-06, "loss": 0.0313, "step": 7469 }, { "epoch": 6.85, "learning_rate": 1.0699739607655434e-06, "loss": 0.0393, "step": 7470 }, { "epoch": 6.85, "learning_rate": 1.0683045089415945e-06, "loss": 0.0372, "step": 7471 }, { "epoch": 6.85, "learning_rate": 1.0666362870383694e-06, "loss": 0.027, "step": 7472 }, { "epoch": 6.85, "learning_rate": 1.0649692952855927e-06, "loss": 0.0277, "step": 7473 }, { "epoch": 6.85, "learning_rate": 1.0633035339128106e-06, "loss": 0.0295, "step": 7474 }, { "epoch": 6.85, "learning_rate": 1.061639003149405e-06, "loss": 0.029, "step": 7475 }, { "epoch": 6.85, "learning_rate": 1.059975703224586e-06, "loss": 0.0314, "step": 7476 }, { "epoch": 6.85, "learning_rate": 1.0583136343673927e-06, "loss": 0.0317, "step": 7477 }, { "epoch": 6.85, "learning_rate": 1.056652796806702e-06, "loss": 0.0334, "step": 7478 }, { "epoch": 6.86, "learning_rate": 1.0549931907712118e-06, "loss": 0.026, "step": 7479 }, { "epoch": 6.86, "learning_rate": 1.0533348164894574e-06, "loss": 0.0337, "step": 7480 }, { "epoch": 6.86, "learning_rate": 1.0516776741897971e-06, "loss": 0.0293, "step": 7481 }, { "epoch": 6.86, "learning_rate": 1.0500217641004296e-06, "loss": 0.0302, "step": 7482 }, { "epoch": 6.86, "learning_rate": 1.0483670864493777e-06, "loss": 0.0355, "step": 7483 }, { "epoch": 6.86, "learning_rate": 1.0467136414644918e-06, "loss": 0.0357, "step": 7484 }, { "epoch": 6.86, "learning_rate": 1.0450614293734595e-06, "loss": 0.0232, "step": 7485 }, { "epoch": 6.86, "learning_rate": 1.0434104504037957e-06, "loss": 0.0283, "step": 7486 }, { "epoch": 6.86, "learning_rate": 1.0417607047828426e-06, "loss": 0.0298, "step": 7487 }, { "epoch": 6.86, "learning_rate": 1.0401121927377756e-06, "loss": 0.0264, "step": 7488 }, { "epoch": 6.86, "learning_rate": 1.0384649144956005e-06, "loss": 0.0357, "step": 7489 }, { "epoch": 6.87, "learning_rate": 1.0368188702831528e-06, "loss": 0.0322, "step": 7490 }, { "epoch": 6.87, "learning_rate": 1.0351740603270955e-06, "loss": 0.0326, "step": 7491 }, { "epoch": 6.87, "learning_rate": 1.0335304848539263e-06, "loss": 0.0322, "step": 7492 }, { "epoch": 6.87, "learning_rate": 1.0318881440899687e-06, "loss": 0.0272, "step": 7493 }, { "epoch": 6.87, "learning_rate": 1.0302470382613794e-06, "loss": 0.0334, "step": 7494 }, { "epoch": 6.87, "learning_rate": 1.02860716759414e-06, "loss": 0.0322, "step": 7495 }, { "epoch": 6.87, "learning_rate": 1.02696853231407e-06, "loss": 0.031, "step": 7496 }, { "epoch": 6.87, "learning_rate": 1.0253311326468117e-06, "loss": 0.0258, "step": 7497 }, { "epoch": 6.87, "learning_rate": 1.02369496881784e-06, "loss": 0.0354, "step": 7498 }, { "epoch": 6.87, "learning_rate": 1.0220600410524607e-06, "loss": 0.0258, "step": 7499 }, { "epoch": 6.87, "learning_rate": 1.020426349575807e-06, "loss": 0.0307, "step": 7500 }, { "epoch": 6.88, "learning_rate": 1.018793894612844e-06, "loss": 0.033, "step": 7501 }, { "epoch": 6.88, "learning_rate": 1.0171626763883624e-06, "loss": 0.0305, "step": 7502 }, { "epoch": 6.88, "learning_rate": 1.0155326951269905e-06, "loss": 0.0319, "step": 7503 }, { "epoch": 6.88, "learning_rate": 1.01390395105318e-06, "loss": 0.0326, "step": 7504 }, { "epoch": 6.88, "learning_rate": 1.0122764443912125e-06, "loss": 0.0311, "step": 7505 }, { "epoch": 6.88, "learning_rate": 1.0106501753651988e-06, "loss": 0.0322, "step": 7506 }, { "epoch": 6.88, "learning_rate": 1.009025144199085e-06, "loss": 0.0323, "step": 7507 }, { "epoch": 6.88, "learning_rate": 1.0074013511166403e-06, "loss": 0.033, "step": 7508 }, { "epoch": 6.88, "learning_rate": 1.0057787963414645e-06, "loss": 0.0237, "step": 7509 }, { "epoch": 6.88, "learning_rate": 1.0041574800969922e-06, "loss": 0.0264, "step": 7510 }, { "epoch": 6.88, "learning_rate": 1.0025374026064816e-06, "loss": 0.0353, "step": 7511 }, { "epoch": 6.89, "learning_rate": 1.0009185640930186e-06, "loss": 0.0376, "step": 7512 }, { "epoch": 6.89, "learning_rate": 9.993009647795282e-07, "loss": 0.0331, "step": 7513 }, { "epoch": 6.89, "learning_rate": 9.976846048887545e-07, "loss": 0.024, "step": 7514 }, { "epoch": 6.89, "learning_rate": 9.960694846432738e-07, "loss": 0.0319, "step": 7515 }, { "epoch": 6.89, "learning_rate": 9.944556042654975e-07, "loss": 0.0286, "step": 7516 }, { "epoch": 6.89, "learning_rate": 9.92842963977657e-07, "loss": 0.027, "step": 7517 }, { "epoch": 6.89, "learning_rate": 9.912315640018223e-07, "loss": 0.0302, "step": 7518 }, { "epoch": 6.89, "learning_rate": 9.896214045598851e-07, "loss": 0.035, "step": 7519 }, { "epoch": 6.89, "learning_rate": 9.880124858735686e-07, "loss": 0.0289, "step": 7520 }, { "epoch": 6.89, "learning_rate": 9.86404808164426e-07, "loss": 0.0267, "step": 7521 }, { "epoch": 6.89, "learning_rate": 9.847983716538423e-07, "loss": 0.0341, "step": 7522 }, { "epoch": 6.9, "learning_rate": 9.831931765630221e-07, "loss": 0.0339, "step": 7523 }, { "epoch": 6.9, "learning_rate": 9.81589223113012e-07, "loss": 0.0288, "step": 7524 }, { "epoch": 6.9, "learning_rate": 9.799865115246787e-07, "loss": 0.0247, "step": 7525 }, { "epoch": 6.9, "learning_rate": 9.783850420187203e-07, "loss": 0.0353, "step": 7526 }, { "epoch": 6.9, "learning_rate": 9.76784814815661e-07, "loss": 0.0299, "step": 7527 }, { "epoch": 6.9, "learning_rate": 9.751858301358607e-07, "loss": 0.0325, "step": 7528 }, { "epoch": 6.9, "learning_rate": 9.73588088199504e-07, "loss": 0.0289, "step": 7529 }, { "epoch": 6.9, "learning_rate": 9.71991589226604e-07, "loss": 0.0347, "step": 7530 }, { "epoch": 6.9, "learning_rate": 9.703963334370004e-07, "loss": 0.0274, "step": 7531 }, { "epoch": 6.9, "learning_rate": 9.688023210503684e-07, "loss": 0.0284, "step": 7532 }, { "epoch": 6.9, "learning_rate": 9.67209552286208e-07, "loss": 0.0255, "step": 7533 }, { "epoch": 6.91, "learning_rate": 9.656180273638449e-07, "loss": 0.0247, "step": 7534 }, { "epoch": 6.91, "learning_rate": 9.640277465024405e-07, "loss": 0.0332, "step": 7535 }, { "epoch": 6.91, "learning_rate": 9.624387099209798e-07, "loss": 0.0312, "step": 7536 }, { "epoch": 6.91, "learning_rate": 9.608509178382775e-07, "loss": 0.0291, "step": 7537 }, { "epoch": 6.91, "learning_rate": 9.592643704729754e-07, "loss": 0.0268, "step": 7538 }, { "epoch": 6.91, "learning_rate": 9.5767906804355e-07, "loss": 0.03, "step": 7539 }, { "epoch": 6.91, "learning_rate": 9.560950107683e-07, "loss": 0.0273, "step": 7540 }, { "epoch": 6.91, "learning_rate": 9.545121988653516e-07, "loss": 0.0293, "step": 7541 }, { "epoch": 6.91, "learning_rate": 9.529306325526677e-07, "loss": 0.0285, "step": 7542 }, { "epoch": 6.91, "learning_rate": 9.513503120480338e-07, "loss": 0.0289, "step": 7543 }, { "epoch": 6.91, "learning_rate": 9.497712375690616e-07, "loss": 0.031, "step": 7544 }, { "epoch": 6.92, "learning_rate": 9.481934093331957e-07, "loss": 0.026, "step": 7545 }, { "epoch": 6.92, "learning_rate": 9.466168275577093e-07, "loss": 0.0336, "step": 7546 }, { "epoch": 6.92, "learning_rate": 9.450414924597007e-07, "loss": 0.0298, "step": 7547 }, { "epoch": 6.92, "learning_rate": 9.434674042560976e-07, "loss": 0.0302, "step": 7548 }, { "epoch": 6.92, "learning_rate": 9.41894563163659e-07, "loss": 0.0383, "step": 7549 }, { "epoch": 6.92, "learning_rate": 9.40322969398968e-07, "loss": 0.037, "step": 7550 }, { "epoch": 6.92, "learning_rate": 9.387526231784383e-07, "loss": 0.0333, "step": 7551 }, { "epoch": 6.92, "learning_rate": 9.371835247183092e-07, "loss": 0.0328, "step": 7552 }, { "epoch": 6.92, "learning_rate": 9.356156742346534e-07, "loss": 0.0351, "step": 7553 }, { "epoch": 6.92, "learning_rate": 9.340490719433671e-07, "loss": 0.0295, "step": 7554 }, { "epoch": 6.92, "learning_rate": 9.324837180601743e-07, "loss": 0.0259, "step": 7555 }, { "epoch": 6.93, "learning_rate": 9.309196128006315e-07, "loss": 0.0343, "step": 7556 }, { "epoch": 6.93, "learning_rate": 9.29356756380121e-07, "loss": 0.0345, "step": 7557 }, { "epoch": 6.93, "learning_rate": 9.277951490138482e-07, "loss": 0.0313, "step": 7558 }, { "epoch": 6.93, "learning_rate": 9.262347909168557e-07, "loss": 0.0333, "step": 7559 }, { "epoch": 6.93, "learning_rate": 9.246756823040082e-07, "loss": 0.0368, "step": 7560 }, { "epoch": 6.93, "learning_rate": 9.231178233899973e-07, "loss": 0.0282, "step": 7561 }, { "epoch": 6.93, "learning_rate": 9.215612143893493e-07, "loss": 0.0284, "step": 7562 }, { "epoch": 6.93, "learning_rate": 9.200058555164082e-07, "loss": 0.0299, "step": 7563 }, { "epoch": 6.93, "learning_rate": 9.18451746985356e-07, "loss": 0.0299, "step": 7564 }, { "epoch": 6.93, "learning_rate": 9.168988890101971e-07, "loss": 0.0367, "step": 7565 }, { "epoch": 6.93, "learning_rate": 9.153472818047627e-07, "loss": 0.037, "step": 7566 }, { "epoch": 6.94, "learning_rate": 9.13796925582715e-07, "loss": 0.0396, "step": 7567 }, { "epoch": 6.94, "learning_rate": 9.122478205575436e-07, "loss": 0.0291, "step": 7568 }, { "epoch": 6.94, "learning_rate": 9.106999669425643e-07, "loss": 0.0305, "step": 7569 }, { "epoch": 6.94, "learning_rate": 9.091533649509177e-07, "loss": 0.0342, "step": 7570 }, { "epoch": 6.94, "learning_rate": 9.076080147955801e-07, "loss": 0.0358, "step": 7571 }, { "epoch": 6.94, "learning_rate": 9.060639166893493e-07, "loss": 0.0319, "step": 7572 }, { "epoch": 6.94, "learning_rate": 9.045210708448504e-07, "loss": 0.0313, "step": 7573 }, { "epoch": 6.94, "learning_rate": 9.029794774745404e-07, "loss": 0.0267, "step": 7574 }, { "epoch": 6.94, "learning_rate": 9.014391367906994e-07, "loss": 0.0264, "step": 7575 }, { "epoch": 6.94, "learning_rate": 8.999000490054388e-07, "loss": 0.0296, "step": 7576 }, { "epoch": 6.95, "learning_rate": 8.983622143306914e-07, "loss": 0.0315, "step": 7577 }, { "epoch": 6.95, "learning_rate": 8.968256329782265e-07, "loss": 0.023, "step": 7578 }, { "epoch": 6.95, "learning_rate": 8.952903051596329e-07, "loss": 0.0331, "step": 7579 }, { "epoch": 6.95, "learning_rate": 8.937562310863301e-07, "loss": 0.0298, "step": 7580 }, { "epoch": 6.95, "learning_rate": 8.92223410969566e-07, "loss": 0.032, "step": 7581 }, { "epoch": 6.95, "learning_rate": 8.906918450204139e-07, "loss": 0.0325, "step": 7582 }, { "epoch": 6.95, "learning_rate": 8.891615334497739e-07, "loss": 0.0295, "step": 7583 }, { "epoch": 6.95, "learning_rate": 8.876324764683731e-07, "loss": 0.0334, "step": 7584 }, { "epoch": 6.95, "learning_rate": 8.861046742867718e-07, "loss": 0.0342, "step": 7585 }, { "epoch": 6.95, "learning_rate": 8.845781271153497e-07, "loss": 0.0342, "step": 7586 }, { "epoch": 6.95, "learning_rate": 8.830528351643153e-07, "loss": 0.0268, "step": 7587 }, { "epoch": 6.96, "learning_rate": 8.815287986437094e-07, "loss": 0.0327, "step": 7588 }, { "epoch": 6.96, "learning_rate": 8.800060177633951e-07, "loss": 0.0276, "step": 7589 }, { "epoch": 6.96, "learning_rate": 8.784844927330638e-07, "loss": 0.0316, "step": 7590 }, { "epoch": 6.96, "learning_rate": 8.769642237622311e-07, "loss": 0.0255, "step": 7591 }, { "epoch": 6.96, "learning_rate": 8.754452110602463e-07, "loss": 0.0331, "step": 7592 }, { "epoch": 6.96, "learning_rate": 8.739274548362808e-07, "loss": 0.0308, "step": 7593 }, { "epoch": 6.96, "learning_rate": 8.724109552993343e-07, "loss": 0.0305, "step": 7594 }, { "epoch": 6.96, "learning_rate": 8.708957126582307e-07, "loss": 0.0306, "step": 7595 }, { "epoch": 6.96, "learning_rate": 8.693817271216276e-07, "loss": 0.0349, "step": 7596 }, { "epoch": 6.96, "learning_rate": 8.678689988980027e-07, "loss": 0.0331, "step": 7597 }, { "epoch": 6.96, "learning_rate": 8.663575281956615e-07, "loss": 0.0371, "step": 7598 }, { "epoch": 6.97, "learning_rate": 8.648473152227421e-07, "loss": 0.0299, "step": 7599 }, { "epoch": 6.97, "learning_rate": 8.633383601872036e-07, "loss": 0.0314, "step": 7600 }, { "epoch": 6.97, "learning_rate": 8.618306632968299e-07, "loss": 0.0326, "step": 7601 }, { "epoch": 6.97, "learning_rate": 8.603242247592414e-07, "loss": 0.033, "step": 7602 }, { "epoch": 6.97, "learning_rate": 8.588190447818767e-07, "loss": 0.0302, "step": 7603 }, { "epoch": 6.97, "learning_rate": 8.573151235720001e-07, "loss": 0.0307, "step": 7604 }, { "epoch": 6.97, "learning_rate": 8.558124613367114e-07, "loss": 0.0246, "step": 7605 }, { "epoch": 6.97, "learning_rate": 8.543110582829272e-07, "loss": 0.0367, "step": 7606 }, { "epoch": 6.97, "learning_rate": 8.52810914617398e-07, "loss": 0.0275, "step": 7607 }, { "epoch": 6.97, "learning_rate": 8.513120305466982e-07, "loss": 0.0341, "step": 7608 }, { "epoch": 6.97, "learning_rate": 8.498144062772252e-07, "loss": 0.0336, "step": 7609 }, { "epoch": 6.98, "learning_rate": 8.483180420152093e-07, "loss": 0.0335, "step": 7610 }, { "epoch": 6.98, "learning_rate": 8.468229379667048e-07, "loss": 0.0313, "step": 7611 }, { "epoch": 6.98, "learning_rate": 8.45329094337588e-07, "loss": 0.0357, "step": 7612 }, { "epoch": 6.98, "learning_rate": 8.438365113335689e-07, "loss": 0.0336, "step": 7613 }, { "epoch": 6.98, "learning_rate": 8.423451891601808e-07, "loss": 0.0319, "step": 7614 }, { "epoch": 6.98, "learning_rate": 8.408551280227828e-07, "loss": 0.0265, "step": 7615 }, { "epoch": 6.98, "learning_rate": 8.393663281265574e-07, "loss": 0.0324, "step": 7616 }, { "epoch": 6.98, "learning_rate": 8.378787896765205e-07, "loss": 0.0296, "step": 7617 }, { "epoch": 6.98, "learning_rate": 8.363925128775097e-07, "loss": 0.0361, "step": 7618 }, { "epoch": 6.98, "learning_rate": 8.349074979341898e-07, "loss": 0.0262, "step": 7619 }, { "epoch": 6.98, "learning_rate": 8.334237450510485e-07, "loss": 0.0244, "step": 7620 }, { "epoch": 6.99, "learning_rate": 8.31941254432408e-07, "loss": 0.0323, "step": 7621 }, { "epoch": 6.99, "learning_rate": 8.304600262824092e-07, "loss": 0.028, "step": 7622 }, { "epoch": 6.99, "learning_rate": 8.289800608050202e-07, "loss": 0.0374, "step": 7623 }, { "epoch": 6.99, "learning_rate": 8.275013582040392e-07, "loss": 0.0279, "step": 7624 }, { "epoch": 6.99, "learning_rate": 8.260239186830865e-07, "loss": 0.0352, "step": 7625 }, { "epoch": 6.99, "learning_rate": 8.245477424456116e-07, "loss": 0.0303, "step": 7626 }, { "epoch": 6.99, "learning_rate": 8.230728296948831e-07, "loss": 0.0276, "step": 7627 }, { "epoch": 6.99, "learning_rate": 8.215991806340073e-07, "loss": 0.0337, "step": 7628 }, { "epoch": 6.99, "learning_rate": 8.201267954659076e-07, "loss": 0.0291, "step": 7629 }, { "epoch": 6.99, "learning_rate": 8.186556743933327e-07, "loss": 0.027, "step": 7630 }, { "epoch": 6.99, "learning_rate": 8.171858176188651e-07, "loss": 0.03, "step": 7631 }, { "epoch": 7.0, "learning_rate": 8.157172253449075e-07, "loss": 0.0307, "step": 7632 }, { "epoch": 7.0, "learning_rate": 8.142498977736868e-07, "loss": 0.0282, "step": 7633 }, { "epoch": 7.0, "learning_rate": 8.127838351072581e-07, "loss": 0.0304, "step": 7634 }, { "epoch": 7.0, "learning_rate": 8.113190375475056e-07, "loss": 0.0353, "step": 7635 }, { "epoch": 7.0, "learning_rate": 8.098555052961355e-07, "loss": 0.0286, "step": 7636 }, { "epoch": 7.0, "learning_rate": 8.083932385546777e-07, "loss": 0.0256, "step": 7637 }, { "epoch": 7.0, "learning_rate": 8.069322375244948e-07, "loss": 0.0243, "step": 7638 }, { "epoch": 7.0, "learning_rate": 8.054725024067678e-07, "loss": 0.0253, "step": 7639 }, { "epoch": 7.0, "learning_rate": 8.040140334025082e-07, "loss": 0.0247, "step": 7640 }, { "epoch": 7.0, "learning_rate": 8.025568307125486e-07, "loss": 0.0239, "step": 7641 }, { "epoch": 7.0, "learning_rate": 8.011008945375554e-07, "loss": 0.0226, "step": 7642 }, { "epoch": 7.01, "learning_rate": 7.996462250780112e-07, "loss": 0.0253, "step": 7643 }, { "epoch": 7.01, "learning_rate": 7.981928225342272e-07, "loss": 0.0213, "step": 7644 }, { "epoch": 7.01, "learning_rate": 7.967406871063454e-07, "loss": 0.0201, "step": 7645 }, { "epoch": 7.01, "learning_rate": 7.95289818994327e-07, "loss": 0.0265, "step": 7646 }, { "epoch": 7.01, "learning_rate": 7.938402183979588e-07, "loss": 0.022, "step": 7647 }, { "epoch": 7.01, "learning_rate": 7.923918855168589e-07, "loss": 0.0224, "step": 7648 }, { "epoch": 7.01, "learning_rate": 7.909448205504633e-07, "loss": 0.0207, "step": 7649 }, { "epoch": 7.01, "learning_rate": 7.894990236980405e-07, "loss": 0.0203, "step": 7650 }, { "epoch": 7.01, "learning_rate": 7.880544951586799e-07, "loss": 0.0203, "step": 7651 }, { "epoch": 7.01, "learning_rate": 7.866112351312949e-07, "loss": 0.0183, "step": 7652 }, { "epoch": 7.01, "learning_rate": 7.851692438146308e-07, "loss": 0.0229, "step": 7653 }, { "epoch": 7.02, "learning_rate": 7.837285214072509e-07, "loss": 0.0223, "step": 7654 }, { "epoch": 7.02, "learning_rate": 7.822890681075468e-07, "loss": 0.024, "step": 7655 }, { "epoch": 7.02, "learning_rate": 7.808508841137374e-07, "loss": 0.0184, "step": 7656 }, { "epoch": 7.02, "learning_rate": 7.794139696238645e-07, "loss": 0.0211, "step": 7657 }, { "epoch": 7.02, "learning_rate": 7.779783248357953e-07, "loss": 0.019, "step": 7658 }, { "epoch": 7.02, "learning_rate": 7.765439499472193e-07, "loss": 0.0219, "step": 7659 }, { "epoch": 7.02, "learning_rate": 7.751108451556599e-07, "loss": 0.0184, "step": 7660 }, { "epoch": 7.02, "learning_rate": 7.736790106584558e-07, "loss": 0.0197, "step": 7661 }, { "epoch": 7.02, "learning_rate": 7.722484466527735e-07, "loss": 0.0215, "step": 7662 }, { "epoch": 7.02, "learning_rate": 7.708191533356102e-07, "loss": 0.0235, "step": 7663 }, { "epoch": 7.02, "learning_rate": 7.693911309037816e-07, "loss": 0.0198, "step": 7664 }, { "epoch": 7.03, "learning_rate": 7.679643795539315e-07, "loss": 0.0204, "step": 7665 }, { "epoch": 7.03, "learning_rate": 7.66538899482524e-07, "loss": 0.0195, "step": 7666 }, { "epoch": 7.03, "learning_rate": 7.651146908858576e-07, "loss": 0.022, "step": 7667 }, { "epoch": 7.03, "learning_rate": 7.636917539600463e-07, "loss": 0.0157, "step": 7668 }, { "epoch": 7.03, "learning_rate": 7.622700889010326e-07, "loss": 0.0233, "step": 7669 }, { "epoch": 7.03, "learning_rate": 7.608496959045863e-07, "loss": 0.0212, "step": 7670 }, { "epoch": 7.03, "learning_rate": 7.594305751662978e-07, "loss": 0.0199, "step": 7671 }, { "epoch": 7.03, "learning_rate": 7.580127268815863e-07, "loss": 0.0181, "step": 7672 }, { "epoch": 7.03, "learning_rate": 7.565961512456888e-07, "loss": 0.0209, "step": 7673 }, { "epoch": 7.03, "learning_rate": 7.551808484536782e-07, "loss": 0.0202, "step": 7674 }, { "epoch": 7.03, "learning_rate": 7.537668187004421e-07, "loss": 0.0222, "step": 7675 }, { "epoch": 7.04, "learning_rate": 7.523540621806957e-07, "loss": 0.0195, "step": 7676 }, { "epoch": 7.04, "learning_rate": 7.509425790889835e-07, "loss": 0.0204, "step": 7677 }, { "epoch": 7.04, "learning_rate": 7.495323696196688e-07, "loss": 0.0222, "step": 7678 }, { "epoch": 7.04, "learning_rate": 7.481234339669419e-07, "loss": 0.0198, "step": 7679 }, { "epoch": 7.04, "learning_rate": 7.467157723248152e-07, "loss": 0.0213, "step": 7680 }, { "epoch": 7.04, "learning_rate": 7.453093848871307e-07, "loss": 0.0203, "step": 7681 }, { "epoch": 7.04, "learning_rate": 7.43904271847552e-07, "loss": 0.0177, "step": 7682 }, { "epoch": 7.04, "learning_rate": 7.425004333995655e-07, "loss": 0.0209, "step": 7683 }, { "epoch": 7.04, "learning_rate": 7.410978697364835e-07, "loss": 0.025, "step": 7684 }, { "epoch": 7.04, "learning_rate": 7.396965810514467e-07, "loss": 0.0191, "step": 7685 }, { "epoch": 7.04, "learning_rate": 7.382965675374132e-07, "loss": 0.0138, "step": 7686 }, { "epoch": 7.05, "learning_rate": 7.368978293871686e-07, "loss": 0.0218, "step": 7687 }, { "epoch": 7.05, "learning_rate": 7.35500366793327e-07, "loss": 0.0158, "step": 7688 }, { "epoch": 7.05, "learning_rate": 7.341041799483217e-07, "loss": 0.0198, "step": 7689 }, { "epoch": 7.05, "learning_rate": 7.327092690444082e-07, "loss": 0.0168, "step": 7690 }, { "epoch": 7.05, "learning_rate": 7.313156342736738e-07, "loss": 0.0187, "step": 7691 }, { "epoch": 7.05, "learning_rate": 7.299232758280239e-07, "loss": 0.0172, "step": 7692 }, { "epoch": 7.05, "learning_rate": 7.285321938991929e-07, "loss": 0.0234, "step": 7693 }, { "epoch": 7.05, "learning_rate": 7.271423886787355e-07, "loss": 0.0185, "step": 7694 }, { "epoch": 7.05, "learning_rate": 7.257538603580305e-07, "loss": 0.0188, "step": 7695 }, { "epoch": 7.05, "learning_rate": 7.243666091282842e-07, "loss": 0.0205, "step": 7696 }, { "epoch": 7.05, "learning_rate": 7.229806351805258e-07, "loss": 0.0191, "step": 7697 }, { "epoch": 7.06, "learning_rate": 7.215959387056037e-07, "loss": 0.0226, "step": 7698 }, { "epoch": 7.06, "learning_rate": 7.202125198942001e-07, "loss": 0.0193, "step": 7699 }, { "epoch": 7.06, "learning_rate": 7.188303789368134e-07, "loss": 0.021, "step": 7700 }, { "epoch": 7.06, "learning_rate": 7.174495160237672e-07, "loss": 0.0152, "step": 7701 }, { "epoch": 7.06, "learning_rate": 7.160699313452136e-07, "loss": 0.0202, "step": 7702 }, { "epoch": 7.06, "learning_rate": 7.146916250911229e-07, "loss": 0.024, "step": 7703 }, { "epoch": 7.06, "learning_rate": 7.133145974512934e-07, "loss": 0.0237, "step": 7704 }, { "epoch": 7.06, "learning_rate": 7.119388486153422e-07, "loss": 0.019, "step": 7705 }, { "epoch": 7.06, "learning_rate": 7.10564378772719e-07, "loss": 0.019, "step": 7706 }, { "epoch": 7.06, "learning_rate": 7.091911881126912e-07, "loss": 0.0222, "step": 7707 }, { "epoch": 7.07, "learning_rate": 7.078192768243486e-07, "loss": 0.0168, "step": 7708 }, { "epoch": 7.07, "learning_rate": 7.064486450966079e-07, "loss": 0.0199, "step": 7709 }, { "epoch": 7.07, "learning_rate": 7.050792931182126e-07, "loss": 0.0215, "step": 7710 }, { "epoch": 7.07, "learning_rate": 7.037112210777231e-07, "loss": 0.0251, "step": 7711 }, { "epoch": 7.07, "learning_rate": 7.023444291635273e-07, "loss": 0.0229, "step": 7712 }, { "epoch": 7.07, "learning_rate": 7.009789175638382e-07, "loss": 0.0234, "step": 7713 }, { "epoch": 7.07, "learning_rate": 6.996146864666886e-07, "loss": 0.0214, "step": 7714 }, { "epoch": 7.07, "learning_rate": 6.982517360599394e-07, "loss": 0.0191, "step": 7715 }, { "epoch": 7.07, "learning_rate": 6.968900665312705e-07, "loss": 0.0183, "step": 7716 }, { "epoch": 7.07, "learning_rate": 6.955296780681897e-07, "loss": 0.0212, "step": 7717 }, { "epoch": 7.07, "learning_rate": 6.94170570858026e-07, "loss": 0.0203, "step": 7718 }, { "epoch": 7.08, "learning_rate": 6.928127450879307e-07, "loss": 0.0209, "step": 7719 }, { "epoch": 7.08, "learning_rate": 6.914562009448833e-07, "loss": 0.02, "step": 7720 }, { "epoch": 7.08, "learning_rate": 6.901009386156821e-07, "loss": 0.0195, "step": 7721 }, { "epoch": 7.08, "learning_rate": 6.887469582869522e-07, "loss": 0.0243, "step": 7722 }, { "epoch": 7.08, "learning_rate": 6.873942601451367e-07, "loss": 0.0189, "step": 7723 }, { "epoch": 7.08, "learning_rate": 6.860428443765121e-07, "loss": 0.02, "step": 7724 }, { "epoch": 7.08, "learning_rate": 6.846927111671686e-07, "loss": 0.0219, "step": 7725 }, { "epoch": 7.08, "learning_rate": 6.833438607030218e-07, "loss": 0.0206, "step": 7726 }, { "epoch": 7.08, "learning_rate": 6.819962931698176e-07, "loss": 0.0242, "step": 7727 }, { "epoch": 7.08, "learning_rate": 6.806500087531165e-07, "loss": 0.0179, "step": 7728 }, { "epoch": 7.08, "learning_rate": 6.793050076383067e-07, "loss": 0.0171, "step": 7729 }, { "epoch": 7.09, "learning_rate": 6.779612900105958e-07, "loss": 0.0196, "step": 7730 }, { "epoch": 7.09, "learning_rate": 6.766188560550235e-07, "loss": 0.0183, "step": 7731 }, { "epoch": 7.09, "learning_rate": 6.752777059564431e-07, "loss": 0.0211, "step": 7732 }, { "epoch": 7.09, "learning_rate": 6.739378398995344e-07, "loss": 0.0166, "step": 7733 }, { "epoch": 7.09, "learning_rate": 6.725992580688034e-07, "loss": 0.0184, "step": 7734 }, { "epoch": 7.09, "learning_rate": 6.71261960648576e-07, "loss": 0.0189, "step": 7735 }, { "epoch": 7.09, "learning_rate": 6.699259478230002e-07, "loss": 0.0244, "step": 7736 }, { "epoch": 7.09, "learning_rate": 6.685912197760513e-07, "loss": 0.0193, "step": 7737 }, { "epoch": 7.09, "learning_rate": 6.672577766915222e-07, "loss": 0.0218, "step": 7738 }, { "epoch": 7.09, "learning_rate": 6.65925618753036e-07, "loss": 0.021, "step": 7739 }, { "epoch": 7.09, "learning_rate": 6.645947461440327e-07, "loss": 0.0168, "step": 7740 }, { "epoch": 7.1, "learning_rate": 6.632651590477756e-07, "loss": 0.0227, "step": 7741 }, { "epoch": 7.1, "learning_rate": 6.61936857647355e-07, "loss": 0.0237, "step": 7742 }, { "epoch": 7.1, "learning_rate": 6.606098421256823e-07, "loss": 0.0211, "step": 7743 }, { "epoch": 7.1, "learning_rate": 6.592841126654892e-07, "loss": 0.0186, "step": 7744 }, { "epoch": 7.1, "learning_rate": 6.579596694493351e-07, "loss": 0.0219, "step": 7745 }, { "epoch": 7.1, "learning_rate": 6.566365126595986e-07, "loss": 0.0183, "step": 7746 }, { "epoch": 7.1, "learning_rate": 6.553146424784828e-07, "loss": 0.0207, "step": 7747 }, { "epoch": 7.1, "learning_rate": 6.539940590880101e-07, "loss": 0.0205, "step": 7748 }, { "epoch": 7.1, "learning_rate": 6.526747626700336e-07, "loss": 0.0226, "step": 7749 }, { "epoch": 7.1, "learning_rate": 6.513567534062204e-07, "loss": 0.0214, "step": 7750 }, { "epoch": 7.1, "learning_rate": 6.500400314780653e-07, "loss": 0.0175, "step": 7751 }, { "epoch": 7.11, "learning_rate": 6.487245970668854e-07, "loss": 0.0217, "step": 7752 }, { "epoch": 7.11, "learning_rate": 6.474104503538203e-07, "loss": 0.0171, "step": 7753 }, { "epoch": 7.11, "learning_rate": 6.460975915198319e-07, "loss": 0.0237, "step": 7754 }, { "epoch": 7.11, "learning_rate": 6.447860207457013e-07, "loss": 0.0208, "step": 7755 }, { "epoch": 7.11, "learning_rate": 6.434757382120404e-07, "loss": 0.0217, "step": 7756 }, { "epoch": 7.11, "learning_rate": 6.421667440992774e-07, "loss": 0.0203, "step": 7757 }, { "epoch": 7.11, "learning_rate": 6.408590385876612e-07, "loss": 0.0161, "step": 7758 }, { "epoch": 7.11, "learning_rate": 6.395526218572723e-07, "loss": 0.018, "step": 7759 }, { "epoch": 7.11, "learning_rate": 6.382474940880057e-07, "loss": 0.021, "step": 7760 }, { "epoch": 7.11, "learning_rate": 6.36943655459581e-07, "loss": 0.0228, "step": 7761 }, { "epoch": 7.11, "learning_rate": 6.356411061515377e-07, "loss": 0.017, "step": 7762 }, { "epoch": 7.12, "learning_rate": 6.343398463432471e-07, "loss": 0.0177, "step": 7763 }, { "epoch": 7.12, "learning_rate": 6.330398762138923e-07, "loss": 0.0201, "step": 7764 }, { "epoch": 7.12, "learning_rate": 6.317411959424812e-07, "loss": 0.0192, "step": 7765 }, { "epoch": 7.12, "learning_rate": 6.30443805707851e-07, "loss": 0.0222, "step": 7766 }, { "epoch": 7.12, "learning_rate": 6.291477056886542e-07, "loss": 0.0225, "step": 7767 }, { "epoch": 7.12, "learning_rate": 6.278528960633667e-07, "loss": 0.0192, "step": 7768 }, { "epoch": 7.12, "learning_rate": 6.265593770102862e-07, "loss": 0.0189, "step": 7769 }, { "epoch": 7.12, "learning_rate": 6.252671487075368e-07, "loss": 0.0175, "step": 7770 }, { "epoch": 7.12, "learning_rate": 6.239762113330616e-07, "loss": 0.0196, "step": 7771 }, { "epoch": 7.12, "learning_rate": 6.226865650646263e-07, "loss": 0.0134, "step": 7772 }, { "epoch": 7.12, "learning_rate": 6.213982100798166e-07, "loss": 0.0245, "step": 7773 }, { "epoch": 7.13, "learning_rate": 6.201111465560461e-07, "loss": 0.0169, "step": 7774 }, { "epoch": 7.13, "learning_rate": 6.188253746705464e-07, "loss": 0.0279, "step": 7775 }, { "epoch": 7.13, "learning_rate": 6.175408946003703e-07, "loss": 0.0174, "step": 7776 }, { "epoch": 7.13, "learning_rate": 6.162577065223973e-07, "loss": 0.0239, "step": 7777 }, { "epoch": 7.13, "learning_rate": 6.149758106133241e-07, "loss": 0.0193, "step": 7778 }, { "epoch": 7.13, "learning_rate": 6.136952070496705e-07, "loss": 0.0181, "step": 7779 }, { "epoch": 7.13, "learning_rate": 6.124158960077831e-07, "loss": 0.0215, "step": 7780 }, { "epoch": 7.13, "learning_rate": 6.111378776638232e-07, "loss": 0.0226, "step": 7781 }, { "epoch": 7.13, "learning_rate": 6.098611521937803e-07, "loss": 0.0183, "step": 7782 }, { "epoch": 7.13, "learning_rate": 6.085857197734624e-07, "loss": 0.0197, "step": 7783 }, { "epoch": 7.13, "learning_rate": 6.07311580578499e-07, "loss": 0.0186, "step": 7784 }, { "epoch": 7.14, "learning_rate": 6.060387347843444e-07, "loss": 0.0228, "step": 7785 }, { "epoch": 7.14, "learning_rate": 6.047671825662738e-07, "loss": 0.0187, "step": 7786 }, { "epoch": 7.14, "learning_rate": 6.034969240993793e-07, "loss": 0.0237, "step": 7787 }, { "epoch": 7.14, "learning_rate": 6.022279595585856e-07, "loss": 0.0192, "step": 7788 }, { "epoch": 7.14, "learning_rate": 6.009602891186283e-07, "loss": 0.0211, "step": 7789 }, { "epoch": 7.14, "learning_rate": 5.996939129540702e-07, "loss": 0.0175, "step": 7790 }, { "epoch": 7.14, "learning_rate": 5.984288312392972e-07, "loss": 0.0181, "step": 7791 }, { "epoch": 7.14, "learning_rate": 5.971650441485122e-07, "loss": 0.0205, "step": 7792 }, { "epoch": 7.14, "learning_rate": 5.959025518557438e-07, "loss": 0.0201, "step": 7793 }, { "epoch": 7.14, "learning_rate": 5.946413545348395e-07, "loss": 0.0158, "step": 7794 }, { "epoch": 7.14, "learning_rate": 5.933814523594716e-07, "loss": 0.019, "step": 7795 }, { "epoch": 7.15, "learning_rate": 5.921228455031314e-07, "loss": 0.0219, "step": 7796 }, { "epoch": 7.15, "learning_rate": 5.908655341391334e-07, "loss": 0.0136, "step": 7797 }, { "epoch": 7.15, "learning_rate": 5.896095184406103e-07, "loss": 0.0236, "step": 7798 }, { "epoch": 7.15, "learning_rate": 5.883547985805238e-07, "loss": 0.02, "step": 7799 }, { "epoch": 7.15, "learning_rate": 5.871013747316501e-07, "loss": 0.0236, "step": 7800 }, { "epoch": 7.15, "learning_rate": 5.858492470665878e-07, "loss": 0.0164, "step": 7801 }, { "epoch": 7.15, "learning_rate": 5.845984157577611e-07, "loss": 0.0211, "step": 7802 }, { "epoch": 7.15, "learning_rate": 5.833488809774123e-07, "loss": 0.0224, "step": 7803 }, { "epoch": 7.15, "learning_rate": 5.821006428976061e-07, "loss": 0.0196, "step": 7804 }, { "epoch": 7.15, "learning_rate": 5.808537016902271e-07, "loss": 0.0188, "step": 7805 }, { "epoch": 7.15, "learning_rate": 5.796080575269858e-07, "loss": 0.0207, "step": 7806 }, { "epoch": 7.16, "learning_rate": 5.783637105794082e-07, "loss": 0.0205, "step": 7807 }, { "epoch": 7.16, "learning_rate": 5.771206610188451e-07, "loss": 0.0234, "step": 7808 }, { "epoch": 7.16, "learning_rate": 5.758789090164696e-07, "loss": 0.0208, "step": 7809 }, { "epoch": 7.16, "learning_rate": 5.746384547432738e-07, "loss": 0.0173, "step": 7810 }, { "epoch": 7.16, "learning_rate": 5.733992983700721e-07, "loss": 0.02, "step": 7811 }, { "epoch": 7.16, "learning_rate": 5.721614400674969e-07, "loss": 0.0226, "step": 7812 }, { "epoch": 7.16, "learning_rate": 5.709248800060108e-07, "loss": 0.0169, "step": 7813 }, { "epoch": 7.16, "learning_rate": 5.696896183558875e-07, "loss": 0.0178, "step": 7814 }, { "epoch": 7.16, "learning_rate": 5.684556552872256e-07, "loss": 0.0242, "step": 7815 }, { "epoch": 7.16, "learning_rate": 5.672229909699489e-07, "loss": 0.0217, "step": 7816 }, { "epoch": 7.16, "learning_rate": 5.659916255737973e-07, "loss": 0.0236, "step": 7817 }, { "epoch": 7.17, "learning_rate": 5.64761559268332e-07, "loss": 0.0201, "step": 7818 }, { "epoch": 7.17, "learning_rate": 5.635327922229384e-07, "loss": 0.0252, "step": 7819 }, { "epoch": 7.17, "learning_rate": 5.623053246068211e-07, "loss": 0.0189, "step": 7820 }, { "epoch": 7.17, "learning_rate": 5.610791565890062e-07, "loss": 0.0197, "step": 7821 }, { "epoch": 7.17, "learning_rate": 5.598542883383384e-07, "loss": 0.0221, "step": 7822 }, { "epoch": 7.17, "learning_rate": 5.586307200234886e-07, "loss": 0.0227, "step": 7823 }, { "epoch": 7.17, "learning_rate": 5.574084518129441e-07, "loss": 0.023, "step": 7824 }, { "epoch": 7.17, "learning_rate": 5.561874838750158e-07, "loss": 0.0228, "step": 7825 }, { "epoch": 7.17, "learning_rate": 5.549678163778349e-07, "loss": 0.0184, "step": 7826 }, { "epoch": 7.17, "learning_rate": 5.537494494893514e-07, "loss": 0.0192, "step": 7827 }, { "epoch": 7.18, "learning_rate": 5.525323833773399e-07, "loss": 0.0199, "step": 7828 }, { "epoch": 7.18, "learning_rate": 5.513166182093932e-07, "loss": 0.0181, "step": 7829 }, { "epoch": 7.18, "learning_rate": 5.501021541529239e-07, "loss": 0.0203, "step": 7830 }, { "epoch": 7.18, "learning_rate": 5.488889913751716e-07, "loss": 0.0245, "step": 7831 }, { "epoch": 7.18, "learning_rate": 5.476771300431882e-07, "loss": 0.0233, "step": 7832 }, { "epoch": 7.18, "learning_rate": 5.464665703238514e-07, "loss": 0.0192, "step": 7833 }, { "epoch": 7.18, "learning_rate": 5.452573123838612e-07, "loss": 0.0221, "step": 7834 }, { "epoch": 7.18, "learning_rate": 5.440493563897343e-07, "loss": 0.0232, "step": 7835 }, { "epoch": 7.18, "learning_rate": 5.428427025078109e-07, "loss": 0.0186, "step": 7836 }, { "epoch": 7.18, "learning_rate": 5.416373509042472e-07, "loss": 0.0209, "step": 7837 }, { "epoch": 7.18, "learning_rate": 5.404333017450281e-07, "loss": 0.0239, "step": 7838 }, { "epoch": 7.19, "learning_rate": 5.39230555195952e-07, "loss": 0.0203, "step": 7839 }, { "epoch": 7.19, "learning_rate": 5.380291114226411e-07, "loss": 0.0171, "step": 7840 }, { "epoch": 7.19, "learning_rate": 5.368289705905383e-07, "loss": 0.0194, "step": 7841 }, { "epoch": 7.19, "learning_rate": 5.356301328649072e-07, "loss": 0.0241, "step": 7842 }, { "epoch": 7.19, "learning_rate": 5.3443259841083e-07, "loss": 0.022, "step": 7843 }, { "epoch": 7.19, "learning_rate": 5.332363673932106e-07, "loss": 0.02, "step": 7844 }, { "epoch": 7.19, "learning_rate": 5.320414399767748e-07, "loss": 0.0224, "step": 7845 }, { "epoch": 7.19, "learning_rate": 5.308478163260666e-07, "loss": 0.0176, "step": 7846 }, { "epoch": 7.19, "learning_rate": 5.296554966054512e-07, "loss": 0.0231, "step": 7847 }, { "epoch": 7.19, "learning_rate": 5.28464480979115e-07, "loss": 0.018, "step": 7848 }, { "epoch": 7.19, "learning_rate": 5.272747696110658e-07, "loss": 0.0217, "step": 7849 }, { "epoch": 7.2, "learning_rate": 5.260863626651291e-07, "loss": 0.0212, "step": 7850 }, { "epoch": 7.2, "learning_rate": 5.248992603049497e-07, "loss": 0.0243, "step": 7851 }, { "epoch": 7.2, "learning_rate": 5.237134626939989e-07, "loss": 0.0177, "step": 7852 }, { "epoch": 7.2, "learning_rate": 5.225289699955626e-07, "loss": 0.0228, "step": 7853 }, { "epoch": 7.2, "learning_rate": 5.213457823727485e-07, "loss": 0.0224, "step": 7854 }, { "epoch": 7.2, "learning_rate": 5.201638999884873e-07, "loss": 0.0167, "step": 7855 }, { "epoch": 7.2, "learning_rate": 5.189833230055264e-07, "loss": 0.0182, "step": 7856 }, { "epoch": 7.2, "learning_rate": 5.178040515864335e-07, "loss": 0.0197, "step": 7857 }, { "epoch": 7.2, "learning_rate": 5.166260858935979e-07, "loss": 0.024, "step": 7858 }, { "epoch": 7.2, "learning_rate": 5.154494260892307e-07, "loss": 0.0194, "step": 7859 }, { "epoch": 7.2, "learning_rate": 5.142740723353601e-07, "loss": 0.0185, "step": 7860 }, { "epoch": 7.21, "learning_rate": 5.131000247938367e-07, "loss": 0.0174, "step": 7861 }, { "epoch": 7.21, "learning_rate": 5.11927283626329e-07, "loss": 0.0184, "step": 7862 }, { "epoch": 7.21, "learning_rate": 5.107558489943276e-07, "loss": 0.0195, "step": 7863 }, { "epoch": 7.21, "learning_rate": 5.095857210591437e-07, "loss": 0.0237, "step": 7864 }, { "epoch": 7.21, "learning_rate": 5.08416899981904e-07, "loss": 0.0205, "step": 7865 }, { "epoch": 7.21, "learning_rate": 5.072493859235628e-07, "loss": 0.0193, "step": 7866 }, { "epoch": 7.21, "learning_rate": 5.060831790448872e-07, "loss": 0.0141, "step": 7867 }, { "epoch": 7.21, "learning_rate": 5.049182795064689e-07, "loss": 0.0212, "step": 7868 }, { "epoch": 7.21, "learning_rate": 5.037546874687171e-07, "loss": 0.0164, "step": 7869 }, { "epoch": 7.21, "learning_rate": 5.025924030918616e-07, "loss": 0.0198, "step": 7870 }, { "epoch": 7.21, "learning_rate": 5.014314265359533e-07, "loss": 0.018, "step": 7871 }, { "epoch": 7.22, "learning_rate": 5.002717579608629e-07, "loss": 0.0256, "step": 7872 }, { "epoch": 7.22, "learning_rate": 4.991133975262763e-07, "loss": 0.0198, "step": 7873 }, { "epoch": 7.22, "learning_rate": 4.97956345391708e-07, "loss": 0.0187, "step": 7874 }, { "epoch": 7.22, "learning_rate": 4.968006017164861e-07, "loss": 0.0181, "step": 7875 }, { "epoch": 7.22, "learning_rate": 4.956461666597568e-07, "loss": 0.024, "step": 7876 }, { "epoch": 7.22, "learning_rate": 4.944930403804926e-07, "loss": 0.0198, "step": 7877 }, { "epoch": 7.22, "learning_rate": 4.933412230374812e-07, "loss": 0.0163, "step": 7878 }, { "epoch": 7.22, "learning_rate": 4.921907147893313e-07, "loss": 0.0294, "step": 7879 }, { "epoch": 7.22, "learning_rate": 4.910415157944715e-07, "loss": 0.0214, "step": 7880 }, { "epoch": 7.22, "learning_rate": 4.898936262111498e-07, "loss": 0.0178, "step": 7881 }, { "epoch": 7.22, "learning_rate": 4.887470461974331e-07, "loss": 0.0206, "step": 7882 }, { "epoch": 7.23, "learning_rate": 4.876017759112095e-07, "loss": 0.0204, "step": 7883 }, { "epoch": 7.23, "learning_rate": 4.864578155101873e-07, "loss": 0.0215, "step": 7884 }, { "epoch": 7.23, "learning_rate": 4.853151651518917e-07, "loss": 0.0196, "step": 7885 }, { "epoch": 7.23, "learning_rate": 4.841738249936689e-07, "loss": 0.0203, "step": 7886 }, { "epoch": 7.23, "learning_rate": 4.830337951926845e-07, "loss": 0.0243, "step": 7887 }, { "epoch": 7.23, "learning_rate": 4.81895075905926e-07, "loss": 0.0202, "step": 7888 }, { "epoch": 7.23, "learning_rate": 4.807576672901971e-07, "loss": 0.0217, "step": 7889 }, { "epoch": 7.23, "learning_rate": 4.796215695021212e-07, "loss": 0.0177, "step": 7890 }, { "epoch": 7.23, "learning_rate": 4.784867826981454e-07, "loss": 0.0243, "step": 7891 }, { "epoch": 7.23, "learning_rate": 4.773533070345305e-07, "loss": 0.0205, "step": 7892 }, { "epoch": 7.23, "learning_rate": 4.762211426673602e-07, "loss": 0.019, "step": 7893 }, { "epoch": 7.24, "learning_rate": 4.7509028975253445e-07, "loss": 0.0186, "step": 7894 }, { "epoch": 7.24, "learning_rate": 4.739607484457798e-07, "loss": 0.0166, "step": 7895 }, { "epoch": 7.24, "learning_rate": 4.7283251890263506e-07, "loss": 0.0182, "step": 7896 }, { "epoch": 7.24, "learning_rate": 4.7170560127845934e-07, "loss": 0.0187, "step": 7897 }, { "epoch": 7.24, "learning_rate": 4.7057999572843516e-07, "loss": 0.0235, "step": 7898 }, { "epoch": 7.24, "learning_rate": 4.6945570240755964e-07, "loss": 0.0183, "step": 7899 }, { "epoch": 7.24, "learning_rate": 4.6833272147065347e-07, "loss": 0.0201, "step": 7900 }, { "epoch": 7.24, "learning_rate": 4.672110530723506e-07, "loss": 0.0194, "step": 7901 }, { "epoch": 7.24, "learning_rate": 4.6609069736711197e-07, "loss": 0.0201, "step": 7902 }, { "epoch": 7.24, "learning_rate": 4.64971654509212e-07, "loss": 0.017, "step": 7903 }, { "epoch": 7.24, "learning_rate": 4.6385392465274626e-07, "loss": 0.0202, "step": 7904 }, { "epoch": 7.25, "learning_rate": 4.6273750795163054e-07, "loss": 0.0199, "step": 7905 }, { "epoch": 7.25, "learning_rate": 4.616224045595974e-07, "loss": 0.0213, "step": 7906 }, { "epoch": 7.25, "learning_rate": 4.605086146302007e-07, "loss": 0.016, "step": 7907 }, { "epoch": 7.25, "learning_rate": 4.593961383168111e-07, "loss": 0.0194, "step": 7908 }, { "epoch": 7.25, "learning_rate": 4.582849757726215e-07, "loss": 0.0209, "step": 7909 }, { "epoch": 7.25, "learning_rate": 4.571751271506419e-07, "loss": 0.0219, "step": 7910 }, { "epoch": 7.25, "learning_rate": 4.5606659260370114e-07, "loss": 0.0248, "step": 7911 }, { "epoch": 7.25, "learning_rate": 4.549593722844492e-07, "loss": 0.0193, "step": 7912 }, { "epoch": 7.25, "learning_rate": 4.5385346634534976e-07, "loss": 0.0205, "step": 7913 }, { "epoch": 7.25, "learning_rate": 4.5274887493869437e-07, "loss": 0.0199, "step": 7914 }, { "epoch": 7.25, "learning_rate": 4.516455982165868e-07, "loss": 0.0222, "step": 7915 }, { "epoch": 7.26, "learning_rate": 4.5054363633094897e-07, "loss": 0.021, "step": 7916 }, { "epoch": 7.26, "learning_rate": 4.4944298943352726e-07, "loss": 0.0183, "step": 7917 }, { "epoch": 7.26, "learning_rate": 4.483436576758826e-07, "loss": 0.0224, "step": 7918 }, { "epoch": 7.26, "learning_rate": 4.472456412093962e-07, "loss": 0.0204, "step": 7919 }, { "epoch": 7.26, "learning_rate": 4.461489401852692e-07, "loss": 0.0223, "step": 7920 }, { "epoch": 7.26, "learning_rate": 4.4505355475451986e-07, "loss": 0.0214, "step": 7921 }, { "epoch": 7.26, "learning_rate": 4.4395948506798403e-07, "loss": 0.0215, "step": 7922 }, { "epoch": 7.26, "learning_rate": 4.428667312763224e-07, "loss": 0.0214, "step": 7923 }, { "epoch": 7.26, "learning_rate": 4.4177529353000794e-07, "loss": 0.0205, "step": 7924 }, { "epoch": 7.26, "learning_rate": 4.406851719793348e-07, "loss": 0.0176, "step": 7925 }, { "epoch": 7.26, "learning_rate": 4.395963667744152e-07, "loss": 0.022, "step": 7926 }, { "epoch": 7.27, "learning_rate": 4.385088780651825e-07, "loss": 0.0172, "step": 7927 }, { "epoch": 7.27, "learning_rate": 4.3742270600138694e-07, "loss": 0.0189, "step": 7928 }, { "epoch": 7.27, "learning_rate": 4.363378507325955e-07, "loss": 0.021, "step": 7929 }, { "epoch": 7.27, "learning_rate": 4.352543124081987e-07, "loss": 0.0238, "step": 7930 }, { "epoch": 7.27, "learning_rate": 4.341720911774028e-07, "loss": 0.0211, "step": 7931 }, { "epoch": 7.27, "learning_rate": 4.330911871892307e-07, "loss": 0.0259, "step": 7932 }, { "epoch": 7.27, "learning_rate": 4.320116005925268e-07, "loss": 0.0195, "step": 7933 }, { "epoch": 7.27, "learning_rate": 4.309333315359543e-07, "loss": 0.0206, "step": 7934 }, { "epoch": 7.27, "learning_rate": 4.298563801679934e-07, "loss": 0.0247, "step": 7935 }, { "epoch": 7.27, "learning_rate": 4.28780746636942e-07, "loss": 0.0197, "step": 7936 }, { "epoch": 7.27, "learning_rate": 4.277064310909218e-07, "loss": 0.0218, "step": 7937 }, { "epoch": 7.28, "learning_rate": 4.2663343367786656e-07, "loss": 0.0175, "step": 7938 }, { "epoch": 7.28, "learning_rate": 4.255617545455315e-07, "loss": 0.0178, "step": 7939 }, { "epoch": 7.28, "learning_rate": 4.244913938414885e-07, "loss": 0.02, "step": 7940 }, { "epoch": 7.28, "learning_rate": 4.2342235171313193e-07, "loss": 0.0229, "step": 7941 }, { "epoch": 7.28, "learning_rate": 4.2235462830767183e-07, "loss": 0.0183, "step": 7942 }, { "epoch": 7.28, "learning_rate": 4.2128822377213384e-07, "loss": 0.0171, "step": 7943 }, { "epoch": 7.28, "learning_rate": 4.2022313825336946e-07, "loss": 0.024, "step": 7944 }, { "epoch": 7.28, "learning_rate": 4.191593718980402e-07, "loss": 0.0229, "step": 7945 }, { "epoch": 7.28, "learning_rate": 4.180969248526334e-07, "loss": 0.0228, "step": 7946 }, { "epoch": 7.28, "learning_rate": 4.1703579726344644e-07, "loss": 0.0202, "step": 7947 }, { "epoch": 7.29, "learning_rate": 4.1597598927660465e-07, "loss": 0.0182, "step": 7948 }, { "epoch": 7.29, "learning_rate": 4.1491750103804353e-07, "loss": 0.0243, "step": 7949 }, { "epoch": 7.29, "learning_rate": 4.13860332693522e-07, "loss": 0.0185, "step": 7950 }, { "epoch": 7.29, "learning_rate": 4.128044843886125e-07, "loss": 0.0189, "step": 7951 }, { "epoch": 7.29, "learning_rate": 4.1174995626871217e-07, "loss": 0.0212, "step": 7952 }, { "epoch": 7.29, "learning_rate": 4.1069674847902916e-07, "loss": 0.0211, "step": 7953 }, { "epoch": 7.29, "learning_rate": 4.0964486116459424e-07, "loss": 0.0198, "step": 7954 }, { "epoch": 7.29, "learning_rate": 4.0859429447025813e-07, "loss": 0.021, "step": 7955 }, { "epoch": 7.29, "learning_rate": 4.075450485406829e-07, "loss": 0.0238, "step": 7956 }, { "epoch": 7.29, "learning_rate": 4.064971235203552e-07, "loss": 0.0221, "step": 7957 }, { "epoch": 7.29, "learning_rate": 4.054505195535774e-07, "loss": 0.0212, "step": 7958 }, { "epoch": 7.3, "learning_rate": 4.0440523678446753e-07, "loss": 0.0205, "step": 7959 }, { "epoch": 7.3, "learning_rate": 4.033612753569682e-07, "loss": 0.021, "step": 7960 }, { "epoch": 7.3, "learning_rate": 4.023186354148312e-07, "loss": 0.0182, "step": 7961 }, { "epoch": 7.3, "learning_rate": 4.0127731710163597e-07, "loss": 0.0222, "step": 7962 }, { "epoch": 7.3, "learning_rate": 4.0023732056077235e-07, "loss": 0.0213, "step": 7963 }, { "epoch": 7.3, "learning_rate": 3.991986459354513e-07, "loss": 0.0242, "step": 7964 }, { "epoch": 7.3, "learning_rate": 3.981612933687007e-07, "loss": 0.0183, "step": 7965 }, { "epoch": 7.3, "learning_rate": 3.971252630033684e-07, "loss": 0.0215, "step": 7966 }, { "epoch": 7.3, "learning_rate": 3.960905549821181e-07, "loss": 0.0204, "step": 7967 }, { "epoch": 7.3, "learning_rate": 3.950571694474303e-07, "loss": 0.0178, "step": 7968 }, { "epoch": 7.3, "learning_rate": 3.940251065416101e-07, "loss": 0.0183, "step": 7969 }, { "epoch": 7.31, "learning_rate": 3.9299436640677146e-07, "loss": 0.0182, "step": 7970 }, { "epoch": 7.31, "learning_rate": 3.9196494918485094e-07, "loss": 0.0217, "step": 7971 }, { "epoch": 7.31, "learning_rate": 3.9093685501760294e-07, "loss": 0.0198, "step": 7972 }, { "epoch": 7.31, "learning_rate": 3.899100840465997e-07, "loss": 0.0205, "step": 7973 }, { "epoch": 7.31, "learning_rate": 3.888846364132304e-07, "loss": 0.0216, "step": 7974 }, { "epoch": 7.31, "learning_rate": 3.878605122587009e-07, "loss": 0.0244, "step": 7975 }, { "epoch": 7.31, "learning_rate": 3.868377117240352e-07, "loss": 0.0207, "step": 7976 }, { "epoch": 7.31, "learning_rate": 3.858162349500805e-07, "loss": 0.0232, "step": 7977 }, { "epoch": 7.31, "learning_rate": 3.8479608207749317e-07, "loss": 0.0209, "step": 7978 }, { "epoch": 7.31, "learning_rate": 3.8377725324675097e-07, "loss": 0.0254, "step": 7979 }, { "epoch": 7.31, "learning_rate": 3.827597485981527e-07, "loss": 0.0228, "step": 7980 }, { "epoch": 7.32, "learning_rate": 3.817435682718096e-07, "loss": 0.0239, "step": 7981 }, { "epoch": 7.32, "learning_rate": 3.807287124076542e-07, "loss": 0.0204, "step": 7982 }, { "epoch": 7.32, "learning_rate": 3.797151811454314e-07, "loss": 0.0199, "step": 7983 }, { "epoch": 7.32, "learning_rate": 3.7870297462471286e-07, "loss": 0.0156, "step": 7984 }, { "epoch": 7.32, "learning_rate": 3.776920929848782e-07, "loss": 0.0202, "step": 7985 }, { "epoch": 7.32, "learning_rate": 3.7668253636513054e-07, "loss": 0.022, "step": 7986 }, { "epoch": 7.32, "learning_rate": 3.7567430490448865e-07, "loss": 0.0246, "step": 7987 }, { "epoch": 7.32, "learning_rate": 3.7466739874178925e-07, "loss": 0.0189, "step": 7988 }, { "epoch": 7.32, "learning_rate": 3.736618180156859e-07, "loss": 0.0172, "step": 7989 }, { "epoch": 7.32, "learning_rate": 3.7265756286464784e-07, "loss": 0.023, "step": 7990 }, { "epoch": 7.32, "learning_rate": 3.716546334269677e-07, "loss": 0.0201, "step": 7991 }, { "epoch": 7.33, "learning_rate": 3.706530298407507e-07, "loss": 0.0204, "step": 7992 }, { "epoch": 7.33, "learning_rate": 3.696527522439186e-07, "loss": 0.0214, "step": 7993 }, { "epoch": 7.33, "learning_rate": 3.686538007742158e-07, "loss": 0.0214, "step": 7994 }, { "epoch": 7.33, "learning_rate": 3.676561755691987e-07, "loss": 0.022, "step": 7995 }, { "epoch": 7.33, "learning_rate": 3.6665987676624326e-07, "loss": 0.0214, "step": 7996 }, { "epoch": 7.33, "learning_rate": 3.656649045025429e-07, "loss": 0.0241, "step": 7997 }, { "epoch": 7.33, "learning_rate": 3.6467125891510915e-07, "loss": 0.0204, "step": 7998 }, { "epoch": 7.33, "learning_rate": 3.636789401407681e-07, "loss": 0.0198, "step": 7999 }, { "epoch": 7.33, "learning_rate": 3.6268794831616717e-07, "loss": 0.0205, "step": 8000 }, { "epoch": 7.33, "learning_rate": 3.6169828357776714e-07, "loss": 0.0184, "step": 8001 }, { "epoch": 7.33, "learning_rate": 3.6070994606184793e-07, "loss": 0.0196, "step": 8002 }, { "epoch": 7.34, "learning_rate": 3.597229359045085e-07, "loss": 0.0212, "step": 8003 }, { "epoch": 7.34, "learning_rate": 3.5873725324165996e-07, "loss": 0.0221, "step": 8004 }, { "epoch": 7.34, "learning_rate": 3.5775289820903615e-07, "loss": 0.0199, "step": 8005 }, { "epoch": 7.34, "learning_rate": 3.567698709421852e-07, "loss": 0.0244, "step": 8006 }, { "epoch": 7.34, "learning_rate": 3.5578817157647217e-07, "loss": 0.0173, "step": 8007 }, { "epoch": 7.34, "learning_rate": 3.5480780024707893e-07, "loss": 0.0237, "step": 8008 }, { "epoch": 7.34, "learning_rate": 3.5382875708900863e-07, "loss": 0.0187, "step": 8009 }, { "epoch": 7.34, "learning_rate": 3.528510422370757e-07, "loss": 0.0209, "step": 8010 }, { "epoch": 7.34, "learning_rate": 3.518746558259145e-07, "loss": 0.0177, "step": 8011 }, { "epoch": 7.34, "learning_rate": 3.508995979899787e-07, "loss": 0.02, "step": 8012 }, { "epoch": 7.34, "learning_rate": 3.499258688635343e-07, "loss": 0.0166, "step": 8013 }, { "epoch": 7.35, "learning_rate": 3.4895346858066723e-07, "loss": 0.0197, "step": 8014 }, { "epoch": 7.35, "learning_rate": 3.479823972752794e-07, "loss": 0.0253, "step": 8015 }, { "epoch": 7.35, "learning_rate": 3.470126550810915e-07, "loss": 0.0219, "step": 8016 }, { "epoch": 7.35, "learning_rate": 3.46044242131639e-07, "loss": 0.0233, "step": 8017 }, { "epoch": 7.35, "learning_rate": 3.450771585602741e-07, "loss": 0.0214, "step": 8018 }, { "epoch": 7.35, "learning_rate": 3.4411140450016924e-07, "loss": 0.0244, "step": 8019 }, { "epoch": 7.35, "learning_rate": 3.4314698008431123e-07, "loss": 0.0226, "step": 8020 }, { "epoch": 7.35, "learning_rate": 3.421838854455029e-07, "loss": 0.0237, "step": 8021 }, { "epoch": 7.35, "learning_rate": 3.412221207163646e-07, "loss": 0.0231, "step": 8022 }, { "epoch": 7.35, "learning_rate": 3.402616860293373e-07, "loss": 0.0128, "step": 8023 }, { "epoch": 7.35, "learning_rate": 3.3930258151667286e-07, "loss": 0.0226, "step": 8024 }, { "epoch": 7.36, "learning_rate": 3.383448073104434e-07, "loss": 0.0174, "step": 8025 }, { "epoch": 7.36, "learning_rate": 3.37388363542539e-07, "loss": 0.02, "step": 8026 }, { "epoch": 7.36, "learning_rate": 3.3643325034466214e-07, "loss": 0.0191, "step": 8027 }, { "epoch": 7.36, "learning_rate": 3.354794678483364e-07, "loss": 0.0197, "step": 8028 }, { "epoch": 7.36, "learning_rate": 3.3452701618489903e-07, "loss": 0.0268, "step": 8029 }, { "epoch": 7.36, "learning_rate": 3.335758954855073e-07, "loss": 0.0166, "step": 8030 }, { "epoch": 7.36, "learning_rate": 3.326261058811331e-07, "loss": 0.0181, "step": 8031 }, { "epoch": 7.36, "learning_rate": 3.316776475025629e-07, "loss": 0.0195, "step": 8032 }, { "epoch": 7.36, "learning_rate": 3.3073052048040456e-07, "loss": 0.0228, "step": 8033 }, { "epoch": 7.36, "learning_rate": 3.297847249450803e-07, "loss": 0.0215, "step": 8034 }, { "epoch": 7.36, "learning_rate": 3.288402610268271e-07, "loss": 0.0197, "step": 8035 }, { "epoch": 7.37, "learning_rate": 3.2789712885570093e-07, "loss": 0.0214, "step": 8036 }, { "epoch": 7.37, "learning_rate": 3.2695532856157565e-07, "loss": 0.0186, "step": 8037 }, { "epoch": 7.37, "learning_rate": 3.2601486027413866e-07, "loss": 0.0163, "step": 8038 }, { "epoch": 7.37, "learning_rate": 3.250757241228952e-07, "loss": 0.016, "step": 8039 }, { "epoch": 7.37, "learning_rate": 3.2413792023716527e-07, "loss": 0.0184, "step": 8040 }, { "epoch": 7.37, "learning_rate": 3.2320144874609107e-07, "loss": 0.0175, "step": 8041 }, { "epoch": 7.37, "learning_rate": 3.222663097786249e-07, "loss": 0.0209, "step": 8042 }, { "epoch": 7.37, "learning_rate": 3.213325034635384e-07, "loss": 0.0201, "step": 8043 }, { "epoch": 7.37, "learning_rate": 3.2040002992942075e-07, "loss": 0.0197, "step": 8044 }, { "epoch": 7.37, "learning_rate": 3.194688893046738e-07, "loss": 0.0189, "step": 8045 }, { "epoch": 7.37, "learning_rate": 3.1853908171752155e-07, "loss": 0.0201, "step": 8046 }, { "epoch": 7.38, "learning_rate": 3.1761060729599945e-07, "loss": 0.0207, "step": 8047 }, { "epoch": 7.38, "learning_rate": 3.166834661679596e-07, "loss": 0.0204, "step": 8048 }, { "epoch": 7.38, "learning_rate": 3.1575765846107555e-07, "loss": 0.016, "step": 8049 }, { "epoch": 7.38, "learning_rate": 3.148331843028296e-07, "loss": 0.0191, "step": 8050 }, { "epoch": 7.38, "learning_rate": 3.139100438205289e-07, "loss": 0.0173, "step": 8051 }, { "epoch": 7.38, "learning_rate": 3.1298823714129066e-07, "loss": 0.0192, "step": 8052 }, { "epoch": 7.38, "learning_rate": 3.1206776439205e-07, "loss": 0.0159, "step": 8053 }, { "epoch": 7.38, "learning_rate": 3.1114862569955775e-07, "loss": 0.0183, "step": 8054 }, { "epoch": 7.38, "learning_rate": 3.1023082119038485e-07, "loss": 0.0172, "step": 8055 }, { "epoch": 7.38, "learning_rate": 3.093143509909147e-07, "loss": 0.023, "step": 8056 }, { "epoch": 7.38, "learning_rate": 3.0839921522734517e-07, "loss": 0.0172, "step": 8057 }, { "epoch": 7.39, "learning_rate": 3.0748541402569665e-07, "loss": 0.0197, "step": 8058 }, { "epoch": 7.39, "learning_rate": 3.065729475118007e-07, "loss": 0.0201, "step": 8059 }, { "epoch": 7.39, "learning_rate": 3.056618158113078e-07, "loss": 0.0236, "step": 8060 }, { "epoch": 7.39, "learning_rate": 3.0475201904968e-07, "loss": 0.0209, "step": 8061 }, { "epoch": 7.39, "learning_rate": 3.038435573522036e-07, "loss": 0.0223, "step": 8062 }, { "epoch": 7.39, "learning_rate": 3.0293643084397416e-07, "loss": 0.026, "step": 8063 }, { "epoch": 7.39, "learning_rate": 3.020306396499062e-07, "loss": 0.0202, "step": 8064 }, { "epoch": 7.39, "learning_rate": 3.011261838947277e-07, "loss": 0.0239, "step": 8065 }, { "epoch": 7.39, "learning_rate": 3.002230637029879e-07, "loss": 0.0223, "step": 8066 }, { "epoch": 7.39, "learning_rate": 2.993212791990485e-07, "loss": 0.0286, "step": 8067 }, { "epoch": 7.4, "learning_rate": 2.984208305070857e-07, "loss": 0.0264, "step": 8068 }, { "epoch": 7.4, "learning_rate": 2.9752171775109695e-07, "loss": 0.0213, "step": 8069 }, { "epoch": 7.4, "learning_rate": 2.96623941054891e-07, "loss": 0.0193, "step": 8070 }, { "epoch": 7.4, "learning_rate": 2.957275005420945e-07, "loss": 0.021, "step": 8071 }, { "epoch": 7.4, "learning_rate": 2.948323963361499e-07, "loss": 0.0184, "step": 8072 }, { "epoch": 7.4, "learning_rate": 2.939386285603163e-07, "loss": 0.0239, "step": 8073 }, { "epoch": 7.4, "learning_rate": 2.930461973376675e-07, "loss": 0.0225, "step": 8074 }, { "epoch": 7.4, "learning_rate": 2.9215510279109294e-07, "loss": 0.022, "step": 8075 }, { "epoch": 7.4, "learning_rate": 2.9126534504330116e-07, "loss": 0.0215, "step": 8076 }, { "epoch": 7.4, "learning_rate": 2.9037692421681306e-07, "loss": 0.0212, "step": 8077 }, { "epoch": 7.4, "learning_rate": 2.894898404339674e-07, "loss": 0.0191, "step": 8078 }, { "epoch": 7.41, "learning_rate": 2.886040938169177e-07, "loss": 0.0212, "step": 8079 }, { "epoch": 7.41, "learning_rate": 2.87719684487634e-07, "loss": 0.0237, "step": 8080 }, { "epoch": 7.41, "learning_rate": 2.868366125679023e-07, "loss": 0.0224, "step": 8081 }, { "epoch": 7.41, "learning_rate": 2.8595487817932424e-07, "loss": 0.0207, "step": 8082 }, { "epoch": 7.41, "learning_rate": 2.850744814433171e-07, "loss": 0.0189, "step": 8083 }, { "epoch": 7.41, "learning_rate": 2.8419542248111385e-07, "loss": 0.0199, "step": 8084 }, { "epoch": 7.41, "learning_rate": 2.8331770141376225e-07, "loss": 0.0172, "step": 8085 }, { "epoch": 7.41, "learning_rate": 2.824413183621311e-07, "loss": 0.0182, "step": 8086 }, { "epoch": 7.41, "learning_rate": 2.8156627344689716e-07, "loss": 0.0206, "step": 8087 }, { "epoch": 7.41, "learning_rate": 2.806925667885574e-07, "loss": 0.0184, "step": 8088 }, { "epoch": 7.41, "learning_rate": 2.7982019850742446e-07, "loss": 0.0254, "step": 8089 }, { "epoch": 7.42, "learning_rate": 2.7894916872362675e-07, "loss": 0.02, "step": 8090 }, { "epoch": 7.42, "learning_rate": 2.780794775571061e-07, "loss": 0.0161, "step": 8091 }, { "epoch": 7.42, "learning_rate": 2.7721112512762217e-07, "loss": 0.0271, "step": 8092 }, { "epoch": 7.42, "learning_rate": 2.763441115547483e-07, "loss": 0.0198, "step": 8093 }, { "epoch": 7.42, "learning_rate": 2.754784369578778e-07, "loss": 0.0177, "step": 8094 }, { "epoch": 7.42, "learning_rate": 2.746141014562154e-07, "loss": 0.024, "step": 8095 }, { "epoch": 7.42, "learning_rate": 2.7375110516878133e-07, "loss": 0.0231, "step": 8096 }, { "epoch": 7.42, "learning_rate": 2.7288944821441285e-07, "loss": 0.0235, "step": 8097 }, { "epoch": 7.42, "learning_rate": 2.7202913071176507e-07, "loss": 0.0164, "step": 8098 }, { "epoch": 7.42, "learning_rate": 2.7117015277930317e-07, "loss": 0.0202, "step": 8099 }, { "epoch": 7.42, "learning_rate": 2.703125145353125e-07, "loss": 0.0178, "step": 8100 }, { "epoch": 7.43, "learning_rate": 2.69456216097892e-07, "loss": 0.0219, "step": 8101 }, { "epoch": 7.43, "learning_rate": 2.6860125758495726e-07, "loss": 0.0208, "step": 8102 }, { "epoch": 7.43, "learning_rate": 2.677476391142386e-07, "loss": 0.0225, "step": 8103 }, { "epoch": 7.43, "learning_rate": 2.668953608032798e-07, "loss": 0.0239, "step": 8104 }, { "epoch": 7.43, "learning_rate": 2.6604442276944364e-07, "loss": 0.0172, "step": 8105 }, { "epoch": 7.43, "learning_rate": 2.6519482512990747e-07, "loss": 0.0207, "step": 8106 }, { "epoch": 7.43, "learning_rate": 2.6434656800166103e-07, "loss": 0.0233, "step": 8107 }, { "epoch": 7.43, "learning_rate": 2.6349965150151535e-07, "loss": 0.0219, "step": 8108 }, { "epoch": 7.43, "learning_rate": 2.626540757460905e-07, "loss": 0.0173, "step": 8109 }, { "epoch": 7.43, "learning_rate": 2.618098408518255e-07, "loss": 0.0217, "step": 8110 }, { "epoch": 7.43, "learning_rate": 2.6096694693497294e-07, "loss": 0.0246, "step": 8111 }, { "epoch": 7.44, "learning_rate": 2.601253941116033e-07, "loss": 0.0199, "step": 8112 }, { "epoch": 7.44, "learning_rate": 2.592851824976017e-07, "loss": 0.0209, "step": 8113 }, { "epoch": 7.44, "learning_rate": 2.5844631220866445e-07, "loss": 0.0183, "step": 8114 }, { "epoch": 7.44, "learning_rate": 2.5760878336031026e-07, "loss": 0.019, "step": 8115 }, { "epoch": 7.44, "learning_rate": 2.5677259606786686e-07, "loss": 0.0218, "step": 8116 }, { "epoch": 7.44, "learning_rate": 2.5593775044647995e-07, "loss": 0.0203, "step": 8117 }, { "epoch": 7.44, "learning_rate": 2.5510424661111087e-07, "loss": 0.019, "step": 8118 }, { "epoch": 7.44, "learning_rate": 2.5427208467653565e-07, "loss": 0.0197, "step": 8119 }, { "epoch": 7.44, "learning_rate": 2.534412647573459e-07, "loss": 0.0201, "step": 8120 }, { "epoch": 7.44, "learning_rate": 2.5261178696794563e-07, "loss": 0.0153, "step": 8121 }, { "epoch": 7.44, "learning_rate": 2.51783651422558e-07, "loss": 0.0211, "step": 8122 }, { "epoch": 7.45, "learning_rate": 2.5095685823522063e-07, "loss": 0.0189, "step": 8123 }, { "epoch": 7.45, "learning_rate": 2.5013140751978473e-07, "loss": 0.0174, "step": 8124 }, { "epoch": 7.45, "learning_rate": 2.4930729938991483e-07, "loss": 0.0193, "step": 8125 }, { "epoch": 7.45, "learning_rate": 2.484845339590969e-07, "loss": 0.023, "step": 8126 }, { "epoch": 7.45, "learning_rate": 2.4766311134062693e-07, "loss": 0.0261, "step": 8127 }, { "epoch": 7.45, "learning_rate": 2.468430316476156e-07, "loss": 0.0222, "step": 8128 }, { "epoch": 7.45, "learning_rate": 2.4602429499299143e-07, "loss": 0.0214, "step": 8129 }, { "epoch": 7.45, "learning_rate": 2.452069014894964e-07, "loss": 0.0162, "step": 8130 }, { "epoch": 7.45, "learning_rate": 2.443908512496884e-07, "loss": 0.0177, "step": 8131 }, { "epoch": 7.45, "learning_rate": 2.4357614438594077e-07, "loss": 0.0232, "step": 8132 }, { "epoch": 7.45, "learning_rate": 2.427627810104405e-07, "loss": 0.0191, "step": 8133 }, { "epoch": 7.46, "learning_rate": 2.41950761235189e-07, "loss": 0.0195, "step": 8134 }, { "epoch": 7.46, "learning_rate": 2.411400851720058e-07, "loss": 0.0189, "step": 8135 }, { "epoch": 7.46, "learning_rate": 2.403307529325205e-07, "loss": 0.0201, "step": 8136 }, { "epoch": 7.46, "learning_rate": 2.395227646281828e-07, "loss": 0.0232, "step": 8137 }, { "epoch": 7.46, "learning_rate": 2.3871612037025596e-07, "loss": 0.0207, "step": 8138 }, { "epoch": 7.46, "learning_rate": 2.3791082026981438e-07, "loss": 0.0197, "step": 8139 }, { "epoch": 7.46, "learning_rate": 2.3710686443775167e-07, "loss": 0.0217, "step": 8140 }, { "epoch": 7.46, "learning_rate": 2.363042529847759e-07, "loss": 0.021, "step": 8141 }, { "epoch": 7.46, "learning_rate": 2.3550298602140753e-07, "loss": 0.0204, "step": 8142 }, { "epoch": 7.46, "learning_rate": 2.347030636579839e-07, "loss": 0.0172, "step": 8143 }, { "epoch": 7.46, "learning_rate": 2.3390448600465687e-07, "loss": 0.0182, "step": 8144 }, { "epoch": 7.47, "learning_rate": 2.3310725317139292e-07, "loss": 0.0227, "step": 8145 }, { "epoch": 7.47, "learning_rate": 2.3231136526797205e-07, "loss": 0.0193, "step": 8146 }, { "epoch": 7.47, "learning_rate": 2.315168224039932e-07, "loss": 0.019, "step": 8147 }, { "epoch": 7.47, "learning_rate": 2.3072362468886445e-07, "loss": 0.0208, "step": 8148 }, { "epoch": 7.47, "learning_rate": 2.2993177223181284e-07, "loss": 0.0241, "step": 8149 }, { "epoch": 7.47, "learning_rate": 2.2914126514187784e-07, "loss": 0.0185, "step": 8150 }, { "epoch": 7.47, "learning_rate": 2.2835210352791681e-07, "loss": 0.0215, "step": 8151 }, { "epoch": 7.47, "learning_rate": 2.2756428749859726e-07, "loss": 0.0209, "step": 8152 }, { "epoch": 7.47, "learning_rate": 2.2677781716240466e-07, "loss": 0.0309, "step": 8153 }, { "epoch": 7.47, "learning_rate": 2.259926926276368e-07, "loss": 0.0192, "step": 8154 }, { "epoch": 7.47, "learning_rate": 2.2520891400241052e-07, "loss": 0.0229, "step": 8155 }, { "epoch": 7.48, "learning_rate": 2.2442648139465285e-07, "loss": 0.0183, "step": 8156 }, { "epoch": 7.48, "learning_rate": 2.2364539491210645e-07, "loss": 0.0183, "step": 8157 }, { "epoch": 7.48, "learning_rate": 2.228656546623309e-07, "loss": 0.019, "step": 8158 }, { "epoch": 7.48, "learning_rate": 2.2208726075269805e-07, "loss": 0.0213, "step": 8159 }, { "epoch": 7.48, "learning_rate": 2.2131021329039436e-07, "loss": 0.025, "step": 8160 }, { "epoch": 7.48, "learning_rate": 2.2053451238242096e-07, "loss": 0.021, "step": 8161 }, { "epoch": 7.48, "learning_rate": 2.197601581355957e-07, "loss": 0.0196, "step": 8162 }, { "epoch": 7.48, "learning_rate": 2.1898715065654997e-07, "loss": 0.026, "step": 8163 }, { "epoch": 7.48, "learning_rate": 2.1821549005172637e-07, "loss": 0.0203, "step": 8164 }, { "epoch": 7.48, "learning_rate": 2.174451764273877e-07, "loss": 0.0175, "step": 8165 }, { "epoch": 7.48, "learning_rate": 2.166762098896069e-07, "loss": 0.0205, "step": 8166 }, { "epoch": 7.49, "learning_rate": 2.159085905442737e-07, "loss": 0.0223, "step": 8167 }, { "epoch": 7.49, "learning_rate": 2.1514231849709023e-07, "loss": 0.0203, "step": 8168 }, { "epoch": 7.49, "learning_rate": 2.1437739385357536e-07, "loss": 0.019, "step": 8169 }, { "epoch": 7.49, "learning_rate": 2.1361381671906267e-07, "loss": 0.02, "step": 8170 }, { "epoch": 7.49, "learning_rate": 2.1285158719869581e-07, "loss": 0.0186, "step": 8171 }, { "epoch": 7.49, "learning_rate": 2.1209070539743858e-07, "loss": 0.0216, "step": 8172 }, { "epoch": 7.49, "learning_rate": 2.113311714200672e-07, "loss": 0.0165, "step": 8173 }, { "epoch": 7.49, "learning_rate": 2.105729853711691e-07, "loss": 0.0211, "step": 8174 }, { "epoch": 7.49, "learning_rate": 2.0981614735515076e-07, "loss": 0.0167, "step": 8175 }, { "epoch": 7.49, "learning_rate": 2.0906065747622995e-07, "loss": 0.0208, "step": 8176 }, { "epoch": 7.49, "learning_rate": 2.083065158384401e-07, "loss": 0.0239, "step": 8177 }, { "epoch": 7.5, "learning_rate": 2.0755372254562922e-07, "loss": 0.0183, "step": 8178 }, { "epoch": 7.5, "learning_rate": 2.0680227770145887e-07, "loss": 0.0196, "step": 8179 }, { "epoch": 7.5, "learning_rate": 2.060521814094052e-07, "loss": 0.021, "step": 8180 }, { "epoch": 7.5, "learning_rate": 2.0530343377275886e-07, "loss": 0.022, "step": 8181 }, { "epoch": 7.5, "learning_rate": 2.0455603489462406e-07, "loss": 0.0201, "step": 8182 }, { "epoch": 7.5, "learning_rate": 2.038099848779207e-07, "loss": 0.0197, "step": 8183 }, { "epoch": 7.5, "learning_rate": 2.0306528382538103e-07, "loss": 0.0193, "step": 8184 }, { "epoch": 7.5, "learning_rate": 2.0232193183955306e-07, "loss": 0.0156, "step": 8185 }, { "epoch": 7.5, "learning_rate": 2.0157992902279823e-07, "loss": 0.0226, "step": 8186 }, { "epoch": 7.5, "learning_rate": 2.0083927547729365e-07, "loss": 0.0219, "step": 8187 }, { "epoch": 7.51, "learning_rate": 2.000999713050289e-07, "loss": 0.019, "step": 8188 }, { "epoch": 7.51, "learning_rate": 1.9936201660780695e-07, "loss": 0.0228, "step": 8189 }, { "epoch": 7.51, "learning_rate": 1.9862541148724768e-07, "loss": 0.0182, "step": 8190 }, { "epoch": 7.51, "learning_rate": 1.9789015604478434e-07, "loss": 0.027, "step": 8191 }, { "epoch": 7.51, "learning_rate": 1.9715625038166265e-07, "loss": 0.0174, "step": 8192 }, { "epoch": 7.51, "learning_rate": 1.9642369459894395e-07, "loss": 0.0219, "step": 8193 }, { "epoch": 7.51, "learning_rate": 1.9569248879750315e-07, "loss": 0.0209, "step": 8194 }, { "epoch": 7.51, "learning_rate": 1.9496263307802964e-07, "loss": 0.0188, "step": 8195 }, { "epoch": 7.51, "learning_rate": 1.9423412754102644e-07, "loss": 0.0148, "step": 8196 }, { "epoch": 7.51, "learning_rate": 1.9350697228681104e-07, "loss": 0.0169, "step": 8197 }, { "epoch": 7.51, "learning_rate": 1.9278116741551668e-07, "loss": 0.0199, "step": 8198 }, { "epoch": 7.52, "learning_rate": 1.9205671302708563e-07, "loss": 0.0196, "step": 8199 }, { "epoch": 7.52, "learning_rate": 1.913336092212781e-07, "loss": 0.0182, "step": 8200 }, { "epoch": 7.52, "learning_rate": 1.9061185609766996e-07, "loss": 0.0206, "step": 8201 }, { "epoch": 7.52, "learning_rate": 1.898914537556462e-07, "loss": 0.0208, "step": 8202 }, { "epoch": 7.52, "learning_rate": 1.8917240229440859e-07, "loss": 0.0182, "step": 8203 }, { "epoch": 7.52, "learning_rate": 1.8845470181297455e-07, "loss": 0.0169, "step": 8204 }, { "epoch": 7.52, "learning_rate": 1.877383524101728e-07, "loss": 0.0183, "step": 8205 }, { "epoch": 7.52, "learning_rate": 1.8702335418464557e-07, "loss": 0.0215, "step": 8206 }, { "epoch": 7.52, "learning_rate": 1.8630970723484965e-07, "loss": 0.0149, "step": 8207 }, { "epoch": 7.52, "learning_rate": 1.8559741165905865e-07, "loss": 0.0191, "step": 8208 }, { "epoch": 7.52, "learning_rate": 1.8488646755535745e-07, "loss": 0.0222, "step": 8209 }, { "epoch": 7.53, "learning_rate": 1.8417687502164438e-07, "loss": 0.0177, "step": 8210 }, { "epoch": 7.53, "learning_rate": 1.8346863415563022e-07, "loss": 0.0203, "step": 8211 }, { "epoch": 7.53, "learning_rate": 1.827617450548458e-07, "loss": 0.0211, "step": 8212 }, { "epoch": 7.53, "learning_rate": 1.8205620781663104e-07, "loss": 0.0265, "step": 8213 }, { "epoch": 7.53, "learning_rate": 1.813520225381371e-07, "loss": 0.0203, "step": 8214 }, { "epoch": 7.53, "learning_rate": 1.8064918931633646e-07, "loss": 0.0199, "step": 8215 }, { "epoch": 7.53, "learning_rate": 1.7994770824800944e-07, "loss": 0.0186, "step": 8216 }, { "epoch": 7.53, "learning_rate": 1.79247579429751e-07, "loss": 0.0198, "step": 8217 }, { "epoch": 7.53, "learning_rate": 1.7854880295797406e-07, "loss": 0.0181, "step": 8218 }, { "epoch": 7.53, "learning_rate": 1.7785137892889936e-07, "loss": 0.0177, "step": 8219 }, { "epoch": 7.53, "learning_rate": 1.7715530743856457e-07, "loss": 0.0207, "step": 8220 }, { "epoch": 7.54, "learning_rate": 1.764605885828219e-07, "loss": 0.0187, "step": 8221 }, { "epoch": 7.54, "learning_rate": 1.757672224573359e-07, "loss": 0.0197, "step": 8222 }, { "epoch": 7.54, "learning_rate": 1.750752091575847e-07, "loss": 0.0216, "step": 8223 }, { "epoch": 7.54, "learning_rate": 1.7438454877886091e-07, "loss": 0.0141, "step": 8224 }, { "epoch": 7.54, "learning_rate": 1.736952414162696e-07, "loss": 0.021, "step": 8225 }, { "epoch": 7.54, "learning_rate": 1.7300728716473257e-07, "loss": 0.0206, "step": 8226 }, { "epoch": 7.54, "learning_rate": 1.723206861189819e-07, "loss": 0.0173, "step": 8227 }, { "epoch": 7.54, "learning_rate": 1.71635438373563e-07, "loss": 0.0225, "step": 8228 }, { "epoch": 7.54, "learning_rate": 1.709515440228382e-07, "loss": 0.0198, "step": 8229 }, { "epoch": 7.54, "learning_rate": 1.7026900316098217e-07, "loss": 0.0187, "step": 8230 }, { "epoch": 7.54, "learning_rate": 1.695878158819819e-07, "loss": 0.0176, "step": 8231 }, { "epoch": 7.55, "learning_rate": 1.6890798227963912e-07, "loss": 0.0186, "step": 8232 }, { "epoch": 7.55, "learning_rate": 1.6822950244756886e-07, "loss": 0.0163, "step": 8233 }, { "epoch": 7.55, "learning_rate": 1.6755237647919976e-07, "loss": 0.0195, "step": 8234 }, { "epoch": 7.55, "learning_rate": 1.6687660446777277e-07, "loss": 0.0177, "step": 8235 }, { "epoch": 7.55, "learning_rate": 1.6620218650634677e-07, "loss": 0.0192, "step": 8236 }, { "epoch": 7.55, "learning_rate": 1.6552912268778977e-07, "loss": 0.0209, "step": 8237 }, { "epoch": 7.55, "learning_rate": 1.648574131047831e-07, "loss": 0.0154, "step": 8238 }, { "epoch": 7.55, "learning_rate": 1.6418705784982393e-07, "loss": 0.0216, "step": 8239 }, { "epoch": 7.55, "learning_rate": 1.6351805701522284e-07, "loss": 0.023, "step": 8240 }, { "epoch": 7.55, "learning_rate": 1.628504106931039e-07, "loss": 0.0189, "step": 8241 }, { "epoch": 7.55, "learning_rate": 1.6218411897540253e-07, "loss": 0.0221, "step": 8242 }, { "epoch": 7.56, "learning_rate": 1.6151918195386752e-07, "loss": 0.0222, "step": 8243 }, { "epoch": 7.56, "learning_rate": 1.608555997200667e-07, "loss": 0.0181, "step": 8244 }, { "epoch": 7.56, "learning_rate": 1.601933723653748e-07, "loss": 0.0197, "step": 8245 }, { "epoch": 7.56, "learning_rate": 1.595324999809822e-07, "loss": 0.0184, "step": 8246 }, { "epoch": 7.56, "learning_rate": 1.5887298265789386e-07, "loss": 0.0207, "step": 8247 }, { "epoch": 7.56, "learning_rate": 1.5821482048692716e-07, "loss": 0.0214, "step": 8248 }, { "epoch": 7.56, "learning_rate": 1.5755801355871292e-07, "loss": 0.0237, "step": 8249 }, { "epoch": 7.56, "learning_rate": 1.5690256196369435e-07, "loss": 0.0207, "step": 8250 }, { "epoch": 7.56, "learning_rate": 1.5624846579213148e-07, "loss": 0.019, "step": 8251 }, { "epoch": 7.56, "learning_rate": 1.5559572513409338e-07, "loss": 0.0209, "step": 8252 }, { "epoch": 7.56, "learning_rate": 1.5494434007946368e-07, "loss": 0.0225, "step": 8253 }, { "epoch": 7.57, "learning_rate": 1.5429431071794177e-07, "loss": 0.0191, "step": 8254 }, { "epoch": 7.57, "learning_rate": 1.5364563713903824e-07, "loss": 0.018, "step": 8255 }, { "epoch": 7.57, "learning_rate": 1.5299831943207722e-07, "loss": 0.0206, "step": 8256 }, { "epoch": 7.57, "learning_rate": 1.5235235768619517e-07, "loss": 0.021, "step": 8257 }, { "epoch": 7.57, "learning_rate": 1.5170775199034538e-07, "loss": 0.0234, "step": 8258 }, { "epoch": 7.57, "learning_rate": 1.5106450243328907e-07, "loss": 0.0219, "step": 8259 }, { "epoch": 7.57, "learning_rate": 1.504226091036054e-07, "loss": 0.0196, "step": 8260 }, { "epoch": 7.57, "learning_rate": 1.4978207208968476e-07, "loss": 0.024, "step": 8261 }, { "epoch": 7.57, "learning_rate": 1.4914289147972992e-07, "loss": 0.0253, "step": 8262 }, { "epoch": 7.57, "learning_rate": 1.485050673617583e-07, "loss": 0.0225, "step": 8263 }, { "epoch": 7.57, "learning_rate": 1.478685998236018e-07, "loss": 0.0197, "step": 8264 }, { "epoch": 7.58, "learning_rate": 1.4723348895290257e-07, "loss": 0.0163, "step": 8265 }, { "epoch": 7.58, "learning_rate": 1.465997348371173e-07, "loss": 0.021, "step": 8266 }, { "epoch": 7.58, "learning_rate": 1.4596733756351512e-07, "loss": 0.0227, "step": 8267 }, { "epoch": 7.58, "learning_rate": 1.4533629721917964e-07, "loss": 0.0181, "step": 8268 }, { "epoch": 7.58, "learning_rate": 1.4470661389100804e-07, "loss": 0.017, "step": 8269 }, { "epoch": 7.58, "learning_rate": 1.4407828766570875e-07, "loss": 0.0184, "step": 8270 }, { "epoch": 7.58, "learning_rate": 1.4345131862980255e-07, "loss": 0.0172, "step": 8271 }, { "epoch": 7.58, "learning_rate": 1.4282570686962705e-07, "loss": 0.0261, "step": 8272 }, { "epoch": 7.58, "learning_rate": 1.422014524713311e-07, "loss": 0.0231, "step": 8273 }, { "epoch": 7.58, "learning_rate": 1.4157855552087373e-07, "loss": 0.0212, "step": 8274 }, { "epoch": 7.58, "learning_rate": 1.4095701610403189e-07, "loss": 0.0214, "step": 8275 }, { "epoch": 7.59, "learning_rate": 1.4033683430639268e-07, "loss": 0.0188, "step": 8276 }, { "epoch": 7.59, "learning_rate": 1.3971801021335774e-07, "loss": 0.0209, "step": 8277 }, { "epoch": 7.59, "learning_rate": 1.3910054391014006e-07, "loss": 0.0196, "step": 8278 }, { "epoch": 7.59, "learning_rate": 1.3848443548176606e-07, "loss": 0.0182, "step": 8279 }, { "epoch": 7.59, "learning_rate": 1.3786968501307784e-07, "loss": 0.0177, "step": 8280 }, { "epoch": 7.59, "learning_rate": 1.3725629258872553e-07, "loss": 0.021, "step": 8281 }, { "epoch": 7.59, "learning_rate": 1.3664425829317708e-07, "loss": 0.02, "step": 8282 }, { "epoch": 7.59, "learning_rate": 1.3603358221071062e-07, "loss": 0.0171, "step": 8283 }, { "epoch": 7.59, "learning_rate": 1.354242644254189e-07, "loss": 0.0155, "step": 8284 }, { "epoch": 7.59, "learning_rate": 1.3481630502120478e-07, "loss": 0.0187, "step": 8285 }, { "epoch": 7.59, "learning_rate": 1.3420970408178912e-07, "loss": 0.0218, "step": 8286 }, { "epoch": 7.6, "learning_rate": 1.336044616906995e-07, "loss": 0.0176, "step": 8287 }, { "epoch": 7.6, "learning_rate": 1.3300057793128262e-07, "loss": 0.0255, "step": 8288 }, { "epoch": 7.6, "learning_rate": 1.3239805288669084e-07, "loss": 0.0228, "step": 8289 }, { "epoch": 7.6, "learning_rate": 1.3179688663989886e-07, "loss": 0.0156, "step": 8290 }, { "epoch": 7.6, "learning_rate": 1.3119707927368498e-07, "loss": 0.0228, "step": 8291 }, { "epoch": 7.6, "learning_rate": 1.3059863087064529e-07, "loss": 0.0236, "step": 8292 }, { "epoch": 7.6, "learning_rate": 1.3000154151318945e-07, "loss": 0.0227, "step": 8293 }, { "epoch": 7.6, "learning_rate": 1.2940581128353723e-07, "loss": 0.0255, "step": 8294 }, { "epoch": 7.6, "learning_rate": 1.2881144026372306e-07, "loss": 0.0193, "step": 8295 }, { "epoch": 7.6, "learning_rate": 1.2821842853559252e-07, "loss": 0.0204, "step": 8296 }, { "epoch": 7.6, "learning_rate": 1.2762677618080587e-07, "loss": 0.0212, "step": 8297 }, { "epoch": 7.61, "learning_rate": 1.2703648328083574e-07, "loss": 0.0179, "step": 8298 }, { "epoch": 7.61, "learning_rate": 1.2644754991696595e-07, "loss": 0.0177, "step": 8299 }, { "epoch": 7.61, "learning_rate": 1.25859976170295e-07, "loss": 0.0188, "step": 8300 }, { "epoch": 7.61, "learning_rate": 1.252737621217337e-07, "loss": 0.0206, "step": 8301 }, { "epoch": 7.61, "learning_rate": 1.2468890785200637e-07, "loss": 0.0214, "step": 8302 }, { "epoch": 7.61, "learning_rate": 1.241054134416464e-07, "loss": 0.0234, "step": 8303 }, { "epoch": 7.61, "learning_rate": 1.2352327897100613e-07, "loss": 0.0189, "step": 8304 }, { "epoch": 7.61, "learning_rate": 1.2294250452024482e-07, "loss": 0.0189, "step": 8305 }, { "epoch": 7.61, "learning_rate": 1.2236309016933624e-07, "loss": 0.0205, "step": 8306 }, { "epoch": 7.61, "learning_rate": 1.217850359980699e-07, "loss": 0.0184, "step": 8307 }, { "epoch": 7.62, "learning_rate": 1.2120834208604437e-07, "loss": 0.0207, "step": 8308 }, { "epoch": 7.62, "learning_rate": 1.206330085126717e-07, "loss": 0.0193, "step": 8309 }, { "epoch": 7.62, "learning_rate": 1.2005903535717733e-07, "loss": 0.0201, "step": 8310 }, { "epoch": 7.62, "learning_rate": 1.1948642269859922e-07, "loss": 0.0178, "step": 8311 }, { "epoch": 7.62, "learning_rate": 1.1891517061578762e-07, "loss": 0.0216, "step": 8312 }, { "epoch": 7.62, "learning_rate": 1.1834527918740624e-07, "loss": 0.0188, "step": 8313 }, { "epoch": 7.62, "learning_rate": 1.1777674849192899e-07, "loss": 0.0184, "step": 8314 }, { "epoch": 7.62, "learning_rate": 1.1720957860764659e-07, "loss": 0.0244, "step": 8315 }, { "epoch": 7.62, "learning_rate": 1.1664376961265878e-07, "loss": 0.0165, "step": 8316 }, { "epoch": 7.62, "learning_rate": 1.1607932158487878e-07, "loss": 0.0198, "step": 8317 }, { "epoch": 7.62, "learning_rate": 1.1551623460203443e-07, "loss": 0.018, "step": 8318 }, { "epoch": 7.63, "learning_rate": 1.1495450874166258e-07, "loss": 0.0172, "step": 8319 }, { "epoch": 7.63, "learning_rate": 1.1439414408111471e-07, "loss": 0.0195, "step": 8320 }, { "epoch": 7.63, "learning_rate": 1.1383514069755464e-07, "loss": 0.0201, "step": 8321 }, { "epoch": 7.63, "learning_rate": 1.1327749866796078e-07, "loss": 0.0204, "step": 8322 }, { "epoch": 7.63, "learning_rate": 1.1272121806912062e-07, "loss": 0.021, "step": 8323 }, { "epoch": 7.63, "learning_rate": 1.1216629897763398e-07, "loss": 0.0183, "step": 8324 }, { "epoch": 7.63, "learning_rate": 1.1161274146991752e-07, "loss": 0.0161, "step": 8325 }, { "epoch": 7.63, "learning_rate": 1.1106054562219693e-07, "loss": 0.021, "step": 8326 }, { "epoch": 7.63, "learning_rate": 1.1050971151051138e-07, "loss": 0.0188, "step": 8327 }, { "epoch": 7.63, "learning_rate": 1.099602392107102e-07, "loss": 0.0228, "step": 8328 }, { "epoch": 7.63, "learning_rate": 1.0941212879846064e-07, "loss": 0.0217, "step": 8329 }, { "epoch": 7.64, "learning_rate": 1.0886538034923677e-07, "loss": 0.0212, "step": 8330 }, { "epoch": 7.64, "learning_rate": 1.0831999393832838e-07, "loss": 0.0225, "step": 8331 }, { "epoch": 7.64, "learning_rate": 1.0777596964083648e-07, "loss": 0.0245, "step": 8332 }, { "epoch": 7.64, "learning_rate": 1.072333075316756e-07, "loss": 0.0164, "step": 8333 }, { "epoch": 7.64, "learning_rate": 1.0669200768557153e-07, "loss": 0.0188, "step": 8334 }, { "epoch": 7.64, "learning_rate": 1.061520701770613e-07, "loss": 0.0171, "step": 8335 }, { "epoch": 7.64, "learning_rate": 1.0561349508049879e-07, "loss": 0.0176, "step": 8336 }, { "epoch": 7.64, "learning_rate": 1.0507628247004464e-07, "loss": 0.0199, "step": 8337 }, { "epoch": 7.64, "learning_rate": 1.0454043241967637e-07, "loss": 0.0198, "step": 8338 }, { "epoch": 7.64, "learning_rate": 1.0400594500318051e-07, "loss": 0.0147, "step": 8339 }, { "epoch": 7.64, "learning_rate": 1.0347282029415928e-07, "loss": 0.0246, "step": 8340 }, { "epoch": 7.65, "learning_rate": 1.0294105836602509e-07, "loss": 0.0143, "step": 8341 }, { "epoch": 7.65, "learning_rate": 1.0241065929200156e-07, "loss": 0.0202, "step": 8342 }, { "epoch": 7.65, "learning_rate": 1.0188162314512918e-07, "loss": 0.0199, "step": 8343 }, { "epoch": 7.65, "learning_rate": 1.013539499982552e-07, "loss": 0.0248, "step": 8344 }, { "epoch": 7.65, "learning_rate": 1.0082763992404265e-07, "loss": 0.0213, "step": 8345 }, { "epoch": 7.65, "learning_rate": 1.0030269299496465e-07, "loss": 0.0197, "step": 8346 }, { "epoch": 7.65, "learning_rate": 9.977910928331114e-08, "loss": 0.0193, "step": 8347 }, { "epoch": 7.65, "learning_rate": 9.92568888611789e-08, "loss": 0.0205, "step": 8348 }, { "epoch": 7.65, "learning_rate": 9.873603180047819e-08, "loss": 0.0175, "step": 8349 }, { "epoch": 7.65, "learning_rate": 9.821653817293497e-08, "loss": 0.0226, "step": 8350 }, { "epoch": 7.65, "learning_rate": 9.769840805008424e-08, "loss": 0.016, "step": 8351 }, { "epoch": 7.66, "learning_rate": 9.718164150327225e-08, "loss": 0.0185, "step": 8352 }, { "epoch": 7.66, "learning_rate": 9.66662386036621e-08, "loss": 0.019, "step": 8353 }, { "epoch": 7.66, "learning_rate": 9.615219942222476e-08, "loss": 0.0237, "step": 8354 }, { "epoch": 7.66, "learning_rate": 9.563952402974475e-08, "loss": 0.0235, "step": 8355 }, { "epoch": 7.66, "learning_rate": 9.512821249682003e-08, "loss": 0.0181, "step": 8356 }, { "epoch": 7.66, "learning_rate": 9.461826489385873e-08, "loss": 0.0165, "step": 8357 }, { "epoch": 7.66, "learning_rate": 9.410968129108244e-08, "loss": 0.0198, "step": 8358 }, { "epoch": 7.66, "learning_rate": 9.360246175852516e-08, "loss": 0.0184, "step": 8359 }, { "epoch": 7.66, "learning_rate": 9.309660636603101e-08, "loss": 0.0217, "step": 8360 }, { "epoch": 7.66, "learning_rate": 9.259211518325872e-08, "loss": 0.0255, "step": 8361 }, { "epoch": 7.66, "learning_rate": 9.208898827967938e-08, "loss": 0.0242, "step": 8362 }, { "epoch": 7.67, "learning_rate": 9.158722572457313e-08, "loss": 0.0218, "step": 8363 }, { "epoch": 7.67, "learning_rate": 9.108682758703468e-08, "loss": 0.0222, "step": 8364 }, { "epoch": 7.67, "learning_rate": 9.058779393597116e-08, "loss": 0.0236, "step": 8365 }, { "epoch": 7.67, "learning_rate": 9.009012484010093e-08, "loss": 0.0166, "step": 8366 }, { "epoch": 7.67, "learning_rate": 8.95938203679525e-08, "loss": 0.0155, "step": 8367 }, { "epoch": 7.67, "learning_rate": 8.909888058787009e-08, "loss": 0.0172, "step": 8368 }, { "epoch": 7.67, "learning_rate": 8.860530556800916e-08, "loss": 0.0262, "step": 8369 }, { "epoch": 7.67, "learning_rate": 8.811309537633539e-08, "loss": 0.0204, "step": 8370 }, { "epoch": 7.67, "learning_rate": 8.762225008062675e-08, "loss": 0.0213, "step": 8371 }, { "epoch": 7.67, "learning_rate": 8.713276974847473e-08, "loss": 0.0208, "step": 8372 }, { "epoch": 7.67, "learning_rate": 8.664465444728321e-08, "loss": 0.0205, "step": 8373 }, { "epoch": 7.68, "learning_rate": 8.615790424426618e-08, "loss": 0.0176, "step": 8374 }, { "epoch": 7.68, "learning_rate": 8.567251920645003e-08, "loss": 0.0235, "step": 8375 }, { "epoch": 7.68, "learning_rate": 8.518849940067575e-08, "loss": 0.0168, "step": 8376 }, { "epoch": 7.68, "learning_rate": 8.470584489359224e-08, "loss": 0.0203, "step": 8377 }, { "epoch": 7.68, "learning_rate": 8.422455575166188e-08, "loss": 0.0215, "step": 8378 }, { "epoch": 7.68, "learning_rate": 8.374463204116279e-08, "loss": 0.0173, "step": 8379 }, { "epoch": 7.68, "learning_rate": 8.326607382817875e-08, "loss": 0.0216, "step": 8380 }, { "epoch": 7.68, "learning_rate": 8.278888117861039e-08, "loss": 0.0194, "step": 8381 }, { "epoch": 7.68, "learning_rate": 8.231305415816738e-08, "loss": 0.0184, "step": 8382 }, { "epoch": 7.68, "learning_rate": 8.183859283237394e-08, "loss": 0.02, "step": 8383 }, { "epoch": 7.68, "learning_rate": 8.13654972665634e-08, "loss": 0.0196, "step": 8384 }, { "epoch": 7.69, "learning_rate": 8.089376752588252e-08, "loss": 0.0148, "step": 8385 }, { "epoch": 7.69, "learning_rate": 8.042340367529156e-08, "loss": 0.0238, "step": 8386 }, { "epoch": 7.69, "learning_rate": 7.99544057795587e-08, "loss": 0.0189, "step": 8387 }, { "epoch": 7.69, "learning_rate": 7.948677390326786e-08, "loss": 0.018, "step": 8388 }, { "epoch": 7.69, "learning_rate": 7.902050811081197e-08, "loss": 0.0176, "step": 8389 }, { "epoch": 7.69, "learning_rate": 7.855560846639965e-08, "loss": 0.0255, "step": 8390 }, { "epoch": 7.69, "learning_rate": 7.809207503404636e-08, "loss": 0.0238, "step": 8391 }, { "epoch": 7.69, "learning_rate": 7.762990787758218e-08, "loss": 0.0249, "step": 8392 }, { "epoch": 7.69, "learning_rate": 7.716910706065173e-08, "loss": 0.0143, "step": 8393 }, { "epoch": 7.69, "learning_rate": 7.670967264670537e-08, "loss": 0.0147, "step": 8394 }, { "epoch": 7.69, "learning_rate": 7.625160469900917e-08, "loss": 0.0269, "step": 8395 }, { "epoch": 7.7, "learning_rate": 7.579490328064265e-08, "loss": 0.0172, "step": 8396 }, { "epoch": 7.7, "learning_rate": 7.533956845449331e-08, "loss": 0.0194, "step": 8397 }, { "epoch": 7.7, "learning_rate": 7.488560028326098e-08, "loss": 0.0228, "step": 8398 }, { "epoch": 7.7, "learning_rate": 7.443299882946009e-08, "loss": 0.0197, "step": 8399 }, { "epoch": 7.7, "learning_rate": 7.398176415541413e-08, "loss": 0.0169, "step": 8400 }, { "epoch": 7.7, "learning_rate": 7.353189632326118e-08, "loss": 0.0198, "step": 8401 }, { "epoch": 7.7, "learning_rate": 7.308339539494835e-08, "loss": 0.0259, "step": 8402 }, { "epoch": 7.7, "learning_rate": 7.263626143223401e-08, "loss": 0.0209, "step": 8403 }, { "epoch": 7.7, "learning_rate": 7.219049449669224e-08, "loss": 0.0216, "step": 8404 }, { "epoch": 7.7, "learning_rate": 7.174609464970505e-08, "loss": 0.0179, "step": 8405 }, { "epoch": 7.7, "learning_rate": 7.130306195246794e-08, "loss": 0.0174, "step": 8406 }, { "epoch": 7.71, "learning_rate": 7.086139646598767e-08, "loss": 0.0259, "step": 8407 }, { "epoch": 7.71, "learning_rate": 7.042109825108335e-08, "loss": 0.0145, "step": 8408 }, { "epoch": 7.71, "learning_rate": 6.998216736838537e-08, "loss": 0.0244, "step": 8409 }, { "epoch": 7.71, "learning_rate": 6.954460387833539e-08, "loss": 0.0176, "step": 8410 }, { "epoch": 7.71, "learning_rate": 6.910840784118634e-08, "loss": 0.0219, "step": 8411 }, { "epoch": 7.71, "learning_rate": 6.867357931700569e-08, "loss": 0.0174, "step": 8412 }, { "epoch": 7.71, "learning_rate": 6.82401183656689e-08, "loss": 0.0196, "step": 8413 }, { "epoch": 7.71, "learning_rate": 6.780802504686601e-08, "loss": 0.0175, "step": 8414 }, { "epoch": 7.71, "learning_rate": 6.737729942009718e-08, "loss": 0.0186, "step": 8415 }, { "epoch": 7.71, "learning_rate": 6.694794154467388e-08, "loss": 0.0192, "step": 8416 }, { "epoch": 7.71, "learning_rate": 6.65199514797199e-08, "loss": 0.0185, "step": 8417 }, { "epoch": 7.72, "learning_rate": 6.609332928417256e-08, "loss": 0.0209, "step": 8418 }, { "epoch": 7.72, "learning_rate": 6.566807501677707e-08, "loss": 0.0194, "step": 8419 }, { "epoch": 7.72, "learning_rate": 6.524418873609329e-08, "loss": 0.0192, "step": 8420 }, { "epoch": 7.72, "learning_rate": 6.482167050049004e-08, "loss": 0.0191, "step": 8421 }, { "epoch": 7.72, "learning_rate": 6.440052036815081e-08, "loss": 0.0237, "step": 8422 }, { "epoch": 7.72, "learning_rate": 6.398073839706919e-08, "loss": 0.0235, "step": 8423 }, { "epoch": 7.72, "learning_rate": 6.356232464505007e-08, "loss": 0.0231, "step": 8424 }, { "epoch": 7.72, "learning_rate": 6.31452791697107e-08, "loss": 0.0182, "step": 8425 }, { "epoch": 7.72, "learning_rate": 6.272960202847844e-08, "loss": 0.0218, "step": 8426 }, { "epoch": 7.72, "learning_rate": 6.231529327859421e-08, "loss": 0.0205, "step": 8427 }, { "epoch": 7.73, "learning_rate": 6.1902352977109e-08, "loss": 0.0188, "step": 8428 }, { "epoch": 7.73, "learning_rate": 6.149078118088625e-08, "loss": 0.0205, "step": 8429 }, { "epoch": 7.73, "learning_rate": 6.10805779466006e-08, "loss": 0.0203, "step": 8430 }, { "epoch": 7.73, "learning_rate": 6.067174333073689e-08, "loss": 0.0216, "step": 8431 }, { "epoch": 7.73, "learning_rate": 6.02642773895945e-08, "loss": 0.0204, "step": 8432 }, { "epoch": 7.73, "learning_rate": 5.985818017928302e-08, "loss": 0.0181, "step": 8433 }, { "epoch": 7.73, "learning_rate": 5.9453451755722145e-08, "loss": 0.0217, "step": 8434 }, { "epoch": 7.73, "learning_rate": 5.905009217464397e-08, "loss": 0.0176, "step": 8435 }, { "epoch": 7.73, "learning_rate": 5.864810149159295e-08, "loss": 0.0207, "step": 8436 }, { "epoch": 7.73, "learning_rate": 5.82474797619248e-08, "loss": 0.0173, "step": 8437 }, { "epoch": 7.73, "learning_rate": 5.784822704080428e-08, "loss": 0.0217, "step": 8438 }, { "epoch": 7.74, "learning_rate": 5.745034338321187e-08, "loss": 0.0171, "step": 8439 }, { "epoch": 7.74, "learning_rate": 5.705382884393706e-08, "loss": 0.0159, "step": 8440 }, { "epoch": 7.74, "learning_rate": 5.665868347757952e-08, "loss": 0.0182, "step": 8441 }, { "epoch": 7.74, "learning_rate": 5.626490733855461e-08, "loss": 0.0222, "step": 8442 }, { "epoch": 7.74, "learning_rate": 5.587250048108339e-08, "loss": 0.02, "step": 8443 }, { "epoch": 7.74, "learning_rate": 5.548146295920265e-08, "loss": 0.0218, "step": 8444 }, { "epoch": 7.74, "learning_rate": 5.509179482676041e-08, "loss": 0.0199, "step": 8445 }, { "epoch": 7.74, "learning_rate": 5.470349613741377e-08, "loss": 0.0169, "step": 8446 }, { "epoch": 7.74, "learning_rate": 5.4316566944634386e-08, "loss": 0.0182, "step": 8447 }, { "epoch": 7.74, "learning_rate": 5.393100730170187e-08, "loss": 0.022, "step": 8448 }, { "epoch": 7.74, "learning_rate": 5.354681726170818e-08, "loss": 0.0195, "step": 8449 }, { "epoch": 7.75, "learning_rate": 5.3163996877559906e-08, "loss": 0.0243, "step": 8450 }, { "epoch": 7.75, "learning_rate": 5.278254620197043e-08, "loss": 0.0193, "step": 8451 }, { "epoch": 7.75, "learning_rate": 5.240246528746773e-08, "loss": 0.0189, "step": 8452 }, { "epoch": 7.75, "learning_rate": 5.202375418638994e-08, "loss": 0.0229, "step": 8453 }, { "epoch": 7.75, "learning_rate": 5.164641295088646e-08, "loss": 0.0185, "step": 8454 }, { "epoch": 7.75, "learning_rate": 5.127044163291795e-08, "loss": 0.0251, "step": 8455 }, { "epoch": 7.75, "learning_rate": 5.0895840284257424e-08, "loss": 0.0226, "step": 8456 }, { "epoch": 7.75, "learning_rate": 5.052260895648808e-08, "loss": 0.0171, "step": 8457 }, { "epoch": 7.75, "learning_rate": 5.015074770100548e-08, "loss": 0.0199, "step": 8458 }, { "epoch": 7.75, "learning_rate": 4.9780256569016414e-08, "loss": 0.0211, "step": 8459 }, { "epoch": 7.75, "learning_rate": 4.941113561153676e-08, "loss": 0.0234, "step": 8460 }, { "epoch": 7.76, "learning_rate": 4.904338487939697e-08, "loss": 0.0236, "step": 8461 }, { "epoch": 7.76, "learning_rate": 4.867700442323764e-08, "loss": 0.0224, "step": 8462 }, { "epoch": 7.76, "learning_rate": 4.831199429350952e-08, "loss": 0.0233, "step": 8463 }, { "epoch": 7.76, "learning_rate": 4.794835454047686e-08, "loss": 0.0201, "step": 8464 }, { "epoch": 7.76, "learning_rate": 4.758608521421182e-08, "loss": 0.0183, "step": 8465 }, { "epoch": 7.76, "learning_rate": 4.722518636460227e-08, "loss": 0.0169, "step": 8466 }, { "epoch": 7.76, "learning_rate": 4.686565804134402e-08, "loss": 0.0245, "step": 8467 }, { "epoch": 7.76, "learning_rate": 4.650750029394413e-08, "loss": 0.0168, "step": 8468 }, { "epoch": 7.76, "learning_rate": 4.615071317172426e-08, "loss": 0.0232, "step": 8469 }, { "epoch": 7.76, "learning_rate": 4.579529672381289e-08, "loss": 0.0142, "step": 8470 }, { "epoch": 7.76, "learning_rate": 4.5441250999153086e-08, "loss": 0.0174, "step": 8471 }, { "epoch": 7.77, "learning_rate": 4.5088576046496965e-08, "loss": 0.0207, "step": 8472 }, { "epoch": 7.77, "learning_rate": 4.473727191441124e-08, "loss": 0.0178, "step": 8473 }, { "epoch": 7.77, "learning_rate": 4.438733865126832e-08, "loss": 0.019, "step": 8474 }, { "epoch": 7.77, "learning_rate": 4.4038776305257436e-08, "loss": 0.0183, "step": 8475 }, { "epoch": 7.77, "learning_rate": 4.369158492437575e-08, "loss": 0.0201, "step": 8476 }, { "epoch": 7.77, "learning_rate": 4.33457645564328e-08, "loss": 0.0193, "step": 8477 }, { "epoch": 7.77, "learning_rate": 4.300131524904827e-08, "loss": 0.0167, "step": 8478 }, { "epoch": 7.77, "learning_rate": 4.2658237049655325e-08, "loss": 0.0171, "step": 8479 }, { "epoch": 7.77, "learning_rate": 4.231653000549507e-08, "loss": 0.0171, "step": 8480 }, { "epoch": 7.77, "learning_rate": 4.1976194163623195e-08, "loss": 0.0182, "step": 8481 }, { "epoch": 7.77, "learning_rate": 4.1637229570904436e-08, "loss": 0.0214, "step": 8482 }, { "epoch": 7.78, "learning_rate": 4.129963627401479e-08, "loss": 0.0163, "step": 8483 }, { "epoch": 7.78, "learning_rate": 4.096341431944151e-08, "loss": 0.017, "step": 8484 }, { "epoch": 7.78, "learning_rate": 4.062856375348423e-08, "loss": 0.0164, "step": 8485 }, { "epoch": 7.78, "learning_rate": 4.0295084622252735e-08, "loss": 0.0214, "step": 8486 }, { "epoch": 7.78, "learning_rate": 3.9962976971666954e-08, "loss": 0.0248, "step": 8487 }, { "epoch": 7.78, "learning_rate": 3.9632240847461424e-08, "loss": 0.0168, "step": 8488 }, { "epoch": 7.78, "learning_rate": 3.930287629517748e-08, "loss": 0.021, "step": 8489 }, { "epoch": 7.78, "learning_rate": 3.8974883360169966e-08, "loss": 0.0181, "step": 8490 }, { "epoch": 7.78, "learning_rate": 3.864826208760497e-08, "loss": 0.0197, "step": 8491 }, { "epoch": 7.78, "learning_rate": 3.832301252245874e-08, "loss": 0.0192, "step": 8492 }, { "epoch": 7.78, "learning_rate": 3.799913470951988e-08, "loss": 0.0174, "step": 8493 }, { "epoch": 7.79, "learning_rate": 3.767662869338606e-08, "loss": 0.0186, "step": 8494 }, { "epoch": 7.79, "learning_rate": 3.735549451846843e-08, "loss": 0.0195, "step": 8495 }, { "epoch": 7.79, "learning_rate": 3.7035732228987154e-08, "loss": 0.0208, "step": 8496 }, { "epoch": 7.79, "learning_rate": 3.671734186897591e-08, "loss": 0.0178, "step": 8497 }, { "epoch": 7.79, "learning_rate": 3.640032348227629e-08, "loss": 0.0174, "step": 8498 }, { "epoch": 7.79, "learning_rate": 3.6084677112543377e-08, "loss": 0.024, "step": 8499 }, { "epoch": 7.79, "learning_rate": 3.57704028032424e-08, "loss": 0.0144, "step": 8500 }, { "epoch": 7.79, "learning_rate": 3.545750059764985e-08, "loss": 0.0198, "step": 8501 }, { "epoch": 7.79, "learning_rate": 3.514597053885349e-08, "loss": 0.0223, "step": 8502 }, { "epoch": 7.79, "learning_rate": 3.4835812669753446e-08, "loss": 0.0173, "step": 8503 }, { "epoch": 7.79, "learning_rate": 3.452702703305666e-08, "loss": 0.0233, "step": 8504 }, { "epoch": 7.8, "learning_rate": 3.421961367128468e-08, "loss": 0.0231, "step": 8505 }, { "epoch": 7.8, "learning_rate": 3.3913572626770305e-08, "loss": 0.0231, "step": 8506 }, { "epoch": 7.8, "learning_rate": 3.3608903941655394e-08, "loss": 0.0149, "step": 8507 }, { "epoch": 7.8, "learning_rate": 3.330560765789415e-08, "loss": 0.0172, "step": 8508 }, { "epoch": 7.8, "learning_rate": 3.300368381724983e-08, "loss": 0.0189, "step": 8509 }, { "epoch": 7.8, "learning_rate": 3.2703132461300305e-08, "loss": 0.0226, "step": 8510 }, { "epoch": 7.8, "learning_rate": 3.2403953631431337e-08, "loss": 0.0206, "step": 8511 }, { "epoch": 7.8, "learning_rate": 3.210614736884221e-08, "loss": 0.0229, "step": 8512 }, { "epoch": 7.8, "learning_rate": 3.1809713714539004e-08, "loss": 0.025, "step": 8513 }, { "epoch": 7.8, "learning_rate": 3.151465270934351e-08, "loss": 0.0203, "step": 8514 }, { "epoch": 7.8, "learning_rate": 3.122096439388544e-08, "loss": 0.0221, "step": 8515 }, { "epoch": 7.81, "learning_rate": 3.092864880860802e-08, "loss": 0.0195, "step": 8516 }, { "epoch": 7.81, "learning_rate": 3.063770599376237e-08, "loss": 0.0216, "step": 8517 }, { "epoch": 7.81, "learning_rate": 3.0348135989413106e-08, "loss": 0.0252, "step": 8518 }, { "epoch": 7.81, "learning_rate": 3.00599388354339e-08, "loss": 0.0201, "step": 8519 }, { "epoch": 7.81, "learning_rate": 2.9773114571511885e-08, "loss": 0.0205, "step": 8520 }, { "epoch": 7.81, "learning_rate": 2.9487663237142138e-08, "loss": 0.0177, "step": 8521 }, { "epoch": 7.81, "learning_rate": 2.9203584871633217e-08, "loss": 0.0178, "step": 8522 }, { "epoch": 7.81, "learning_rate": 2.892087951410272e-08, "loss": 0.02, "step": 8523 }, { "epoch": 7.81, "learning_rate": 2.86395472034795e-08, "loss": 0.0231, "step": 8524 }, { "epoch": 7.81, "learning_rate": 2.835958797850591e-08, "loss": 0.0167, "step": 8525 }, { "epoch": 7.81, "learning_rate": 2.8081001877732216e-08, "loss": 0.0157, "step": 8526 }, { "epoch": 7.82, "learning_rate": 2.780378893951885e-08, "loss": 0.0226, "step": 8527 }, { "epoch": 7.82, "learning_rate": 2.7527949202040827e-08, "loss": 0.02, "step": 8528 }, { "epoch": 7.82, "learning_rate": 2.725348270328221e-08, "loss": 0.021, "step": 8529 }, { "epoch": 7.82, "learning_rate": 2.6980389481037206e-08, "loss": 0.0164, "step": 8530 }, { "epoch": 7.82, "learning_rate": 2.6708669572910184e-08, "loss": 0.0154, "step": 8531 }, { "epoch": 7.82, "learning_rate": 2.64383230163201e-08, "loss": 0.0194, "step": 8532 }, { "epoch": 7.82, "learning_rate": 2.616934984849384e-08, "loss": 0.0205, "step": 8533 }, { "epoch": 7.82, "learning_rate": 2.5901750106468447e-08, "loss": 0.0207, "step": 8534 }, { "epoch": 7.82, "learning_rate": 2.5635523827094442e-08, "loss": 0.016, "step": 8535 }, { "epoch": 7.82, "learning_rate": 2.5370671047032504e-08, "loss": 0.0188, "step": 8536 }, { "epoch": 7.82, "learning_rate": 2.510719180275234e-08, "loss": 0.0199, "step": 8537 }, { "epoch": 7.83, "learning_rate": 2.4845086130536045e-08, "loss": 0.0209, "step": 8538 }, { "epoch": 7.83, "learning_rate": 2.4584354066476966e-08, "loss": 0.0203, "step": 8539 }, { "epoch": 7.83, "learning_rate": 2.432499564647861e-08, "loss": 0.0209, "step": 8540 }, { "epoch": 7.83, "learning_rate": 2.406701090625463e-08, "loss": 0.019, "step": 8541 }, { "epoch": 7.83, "learning_rate": 2.3810399881329938e-08, "loss": 0.0207, "step": 8542 }, { "epoch": 7.83, "learning_rate": 2.3555162607042937e-08, "loss": 0.0166, "step": 8543 }, { "epoch": 7.83, "learning_rate": 2.3301299118538844e-08, "loss": 0.0153, "step": 8544 }, { "epoch": 7.83, "learning_rate": 2.3048809450774145e-08, "loss": 0.0168, "step": 8545 }, { "epoch": 7.83, "learning_rate": 2.2797693638521022e-08, "loss": 0.0194, "step": 8546 }, { "epoch": 7.83, "learning_rate": 2.2547951716355155e-08, "loss": 0.0214, "step": 8547 }, { "epoch": 7.84, "learning_rate": 2.2299583718669026e-08, "loss": 0.0165, "step": 8548 }, { "epoch": 7.84, "learning_rate": 2.2052589679663062e-08, "loss": 0.0275, "step": 8549 }, { "epoch": 7.84, "learning_rate": 2.180696963334783e-08, "loss": 0.0208, "step": 8550 }, { "epoch": 7.84, "learning_rate": 2.1562723613548498e-08, "loss": 0.0162, "step": 8551 }, { "epoch": 7.84, "learning_rate": 2.131985165389594e-08, "loss": 0.02, "step": 8552 }, { "epoch": 7.84, "learning_rate": 2.1078353787835626e-08, "loss": 0.0207, "step": 8553 }, { "epoch": 7.84, "learning_rate": 2.0838230048623177e-08, "loss": 0.0228, "step": 8554 }, { "epoch": 7.84, "learning_rate": 2.059948046932214e-08, "loss": 0.0182, "step": 8555 }, { "epoch": 7.84, "learning_rate": 2.0362105082810667e-08, "loss": 0.0236, "step": 8556 }, { "epoch": 7.84, "learning_rate": 2.0126103921775943e-08, "loss": 0.0218, "step": 8557 }, { "epoch": 7.84, "learning_rate": 1.989147701871641e-08, "loss": 0.0201, "step": 8558 }, { "epoch": 7.85, "learning_rate": 1.9658224405939562e-08, "loss": 0.0222, "step": 8559 }, { "epoch": 7.85, "learning_rate": 1.9426346115565264e-08, "loss": 0.0194, "step": 8560 }, { "epoch": 7.85, "learning_rate": 1.919584217952575e-08, "loss": 0.0216, "step": 8561 }, { "epoch": 7.85, "learning_rate": 1.896671262955896e-08, "loss": 0.0204, "step": 8562 }, { "epoch": 7.85, "learning_rate": 1.8738957497219655e-08, "loss": 0.0217, "step": 8563 }, { "epoch": 7.85, "learning_rate": 1.851257681386831e-08, "loss": 0.0174, "step": 8564 }, { "epoch": 7.85, "learning_rate": 1.8287570610678872e-08, "loss": 0.0215, "step": 8565 }, { "epoch": 7.85, "learning_rate": 1.806393891863656e-08, "loss": 0.0163, "step": 8566 }, { "epoch": 7.85, "learning_rate": 1.7841681768533404e-08, "loss": 0.0172, "step": 8567 }, { "epoch": 7.85, "learning_rate": 1.7620799190977145e-08, "loss": 0.0213, "step": 8568 }, { "epoch": 7.85, "learning_rate": 1.7401291216383455e-08, "loss": 0.0212, "step": 8569 }, { "epoch": 7.86, "learning_rate": 1.718315787497926e-08, "loss": 0.0186, "step": 8570 }, { "epoch": 7.86, "learning_rate": 1.6966399196801653e-08, "loss": 0.019, "step": 8571 }, { "epoch": 7.86, "learning_rate": 1.6751015211698975e-08, "loss": 0.0192, "step": 8572 }, { "epoch": 7.86, "learning_rate": 1.653700594933194e-08, "loss": 0.0222, "step": 8573 }, { "epoch": 7.86, "learning_rate": 1.6324371439166986e-08, "loss": 0.0246, "step": 8574 }, { "epoch": 7.86, "learning_rate": 1.611311171048735e-08, "loss": 0.0185, "step": 8575 }, { "epoch": 7.86, "learning_rate": 1.59032267923831e-08, "loss": 0.0221, "step": 8576 }, { "epoch": 7.86, "learning_rate": 1.5694716713755554e-08, "loss": 0.0201, "step": 8577 }, { "epoch": 7.86, "learning_rate": 1.5487581503318417e-08, "loss": 0.0181, "step": 8578 }, { "epoch": 7.86, "learning_rate": 1.5281821189593317e-08, "loss": 0.0239, "step": 8579 }, { "epoch": 7.86, "learning_rate": 1.5077435800915365e-08, "loss": 0.0177, "step": 8580 }, { "epoch": 7.87, "learning_rate": 1.4874425365427602e-08, "loss": 0.0207, "step": 8581 }, { "epoch": 7.87, "learning_rate": 1.467278991108656e-08, "loss": 0.0197, "step": 8582 }, { "epoch": 7.87, "learning_rate": 1.4472529465657802e-08, "loss": 0.022, "step": 8583 }, { "epoch": 7.87, "learning_rate": 1.4273644056717051e-08, "loss": 0.0226, "step": 8584 }, { "epoch": 7.87, "learning_rate": 1.4076133711652395e-08, "loss": 0.0239, "step": 8585 }, { "epoch": 7.87, "learning_rate": 1.387999845765986e-08, "loss": 0.0204, "step": 8586 }, { "epoch": 7.87, "learning_rate": 1.368523832175006e-08, "loss": 0.0186, "step": 8587 }, { "epoch": 7.87, "learning_rate": 1.3491853330740434e-08, "loss": 0.018, "step": 8588 }, { "epoch": 7.87, "learning_rate": 1.329984351126079e-08, "loss": 0.0229, "step": 8589 }, { "epoch": 7.87, "learning_rate": 1.31092088897522e-08, "loss": 0.019, "step": 8590 }, { "epoch": 7.87, "learning_rate": 1.2919949492465888e-08, "loss": 0.0203, "step": 8591 }, { "epoch": 7.88, "learning_rate": 1.2732065345462118e-08, "loss": 0.0191, "step": 8592 }, { "epoch": 7.88, "learning_rate": 1.2545556474614639e-08, "loss": 0.0193, "step": 8593 }, { "epoch": 7.88, "learning_rate": 1.236042290560402e-08, "loss": 0.0167, "step": 8594 }, { "epoch": 7.88, "learning_rate": 1.2176664663926529e-08, "loss": 0.0223, "step": 8595 }, { "epoch": 7.88, "learning_rate": 1.1994281774884153e-08, "loss": 0.0161, "step": 8596 }, { "epoch": 7.88, "learning_rate": 1.1813274263592356e-08, "loss": 0.0221, "step": 8597 }, { "epoch": 7.88, "learning_rate": 1.1633642154975645e-08, "loss": 0.0229, "step": 8598 }, { "epoch": 7.88, "learning_rate": 1.1455385473770896e-08, "loss": 0.0164, "step": 8599 }, { "epoch": 7.88, "learning_rate": 1.1278504244524036e-08, "loss": 0.0157, "step": 8600 }, { "epoch": 7.88, "learning_rate": 1.1102998491592243e-08, "loss": 0.0199, "step": 8601 }, { "epoch": 7.88, "learning_rate": 1.0928868239142854e-08, "loss": 0.0217, "step": 8602 }, { "epoch": 7.89, "learning_rate": 1.0756113511155575e-08, "loss": 0.022, "step": 8603 }, { "epoch": 7.89, "learning_rate": 1.0584734331416935e-08, "loss": 0.0196, "step": 8604 }, { "epoch": 7.89, "learning_rate": 1.0414730723528055e-08, "loss": 0.0203, "step": 8605 }, { "epoch": 7.89, "learning_rate": 1.0246102710897986e-08, "loss": 0.025, "step": 8606 }, { "epoch": 7.89, "learning_rate": 1.0078850316748157e-08, "loss": 0.0131, "step": 8607 }, { "epoch": 7.89, "learning_rate": 9.91297356411014e-09, "loss": 0.0206, "step": 8608 }, { "epoch": 7.89, "learning_rate": 9.748472475823445e-09, "loss": 0.0191, "step": 8609 }, { "epoch": 7.89, "learning_rate": 9.585347074542173e-09, "loss": 0.0161, "step": 8610 }, { "epoch": 7.89, "learning_rate": 9.423597382728356e-09, "loss": 0.0201, "step": 8611 }, { "epoch": 7.89, "learning_rate": 9.263223422656398e-09, "loss": 0.0187, "step": 8612 }, { "epoch": 7.89, "learning_rate": 9.104225216408635e-09, "loss": 0.0161, "step": 8613 }, { "epoch": 7.9, "learning_rate": 8.946602785880887e-09, "loss": 0.0207, "step": 8614 }, { "epoch": 7.9, "learning_rate": 8.790356152778013e-09, "loss": 0.0205, "step": 8615 }, { "epoch": 7.9, "learning_rate": 8.635485338616135e-09, "loss": 0.0218, "step": 8616 }, { "epoch": 7.9, "learning_rate": 8.481990364720416e-09, "loss": 0.0221, "step": 8617 }, { "epoch": 7.9, "learning_rate": 8.329871252227283e-09, "loss": 0.019, "step": 8618 }, { "epoch": 7.9, "learning_rate": 8.179128022085536e-09, "loss": 0.0161, "step": 8619 }, { "epoch": 7.9, "learning_rate": 8.029760695051903e-09, "loss": 0.0226, "step": 8620 }, { "epoch": 7.9, "learning_rate": 7.881769291694374e-09, "loss": 0.0235, "step": 8621 }, { "epoch": 7.9, "learning_rate": 7.735153832392206e-09, "loss": 0.021, "step": 8622 }, { "epoch": 7.9, "learning_rate": 7.589914337335913e-09, "loss": 0.0167, "step": 8623 }, { "epoch": 7.9, "learning_rate": 7.446050826522832e-09, "loss": 0.023, "step": 8624 }, { "epoch": 7.91, "learning_rate": 7.303563319766005e-09, "loss": 0.0189, "step": 8625 }, { "epoch": 7.91, "learning_rate": 7.162451836685291e-09, "loss": 0.0188, "step": 8626 }, { "epoch": 7.91, "learning_rate": 7.022716396711815e-09, "loss": 0.0184, "step": 8627 }, { "epoch": 7.91, "learning_rate": 6.8843570190890714e-09, "loss": 0.0213, "step": 8628 }, { "epoch": 7.91, "learning_rate": 6.747373722867378e-09, "loss": 0.0218, "step": 8629 }, { "epoch": 7.91, "learning_rate": 6.6117665269105344e-09, "loss": 0.0226, "step": 8630 }, { "epoch": 7.91, "learning_rate": 6.4775354498924915e-09, "loss": 0.0197, "step": 8631 }, { "epoch": 7.91, "learning_rate": 6.344680510297352e-09, "loss": 0.0197, "step": 8632 }, { "epoch": 7.91, "learning_rate": 6.213201726418261e-09, "loss": 0.0216, "step": 8633 }, { "epoch": 7.91, "learning_rate": 6.083099116362956e-09, "loss": 0.0184, "step": 8634 }, { "epoch": 7.91, "learning_rate": 5.954372698043775e-09, "loss": 0.0225, "step": 8635 }, { "epoch": 7.92, "learning_rate": 5.8270224891887605e-09, "loss": 0.0197, "step": 8636 }, { "epoch": 7.92, "learning_rate": 5.701048507333884e-09, "loss": 0.0185, "step": 8637 }, { "epoch": 7.92, "learning_rate": 5.576450769826381e-09, "loss": 0.0218, "step": 8638 }, { "epoch": 7.92, "learning_rate": 5.453229293823636e-09, "loss": 0.019, "step": 8639 }, { "epoch": 7.92, "learning_rate": 5.331384096293191e-09, "loss": 0.0164, "step": 8640 }, { "epoch": 7.92, "learning_rate": 5.210915194013843e-09, "loss": 0.0219, "step": 8641 }, { "epoch": 7.92, "learning_rate": 5.091822603574548e-09, "loss": 0.019, "step": 8642 }, { "epoch": 7.92, "learning_rate": 4.9741063413744075e-09, "loss": 0.0207, "step": 8643 }, { "epoch": 7.92, "learning_rate": 4.857766423622678e-09, "loss": 0.0207, "step": 8644 }, { "epoch": 7.92, "learning_rate": 4.74280286634099e-09, "loss": 0.0221, "step": 8645 }, { "epoch": 7.92, "learning_rate": 4.629215685360011e-09, "loss": 0.0185, "step": 8646 }, { "epoch": 7.93, "learning_rate": 4.517004896320565e-09, "loss": 0.018, "step": 8647 }, { "epoch": 7.93, "learning_rate": 4.406170514674735e-09, "loss": 0.0208, "step": 8648 }, { "epoch": 7.93, "learning_rate": 4.296712555683646e-09, "loss": 0.0228, "step": 8649 }, { "epoch": 7.93, "learning_rate": 4.188631034421909e-09, "loss": 0.0194, "step": 8650 }, { "epoch": 7.93, "learning_rate": 4.08192596577095e-09, "loss": 0.0178, "step": 8651 }, { "epoch": 7.93, "learning_rate": 3.976597364425683e-09, "loss": 0.019, "step": 8652 }, { "epoch": 7.93, "learning_rate": 3.872645244888951e-09, "loss": 0.0206, "step": 8653 }, { "epoch": 7.93, "learning_rate": 3.770069621475969e-09, "loss": 0.0231, "step": 8654 }, { "epoch": 7.93, "learning_rate": 3.6688705083121057e-09, "loss": 0.0198, "step": 8655 }, { "epoch": 7.93, "learning_rate": 3.5690479193317696e-09, "loss": 0.0223, "step": 8656 }, { "epoch": 7.93, "learning_rate": 3.4706018682817422e-09, "loss": 0.0205, "step": 8657 }, { "epoch": 7.94, "learning_rate": 3.373532368717847e-09, "loss": 0.0249, "step": 8658 }, { "epoch": 7.94, "learning_rate": 3.277839434006058e-09, "loss": 0.0233, "step": 8659 }, { "epoch": 7.94, "learning_rate": 3.183523077324724e-09, "loss": 0.0212, "step": 8660 }, { "epoch": 7.94, "learning_rate": 3.0905833116612326e-09, "loss": 0.0207, "step": 8661 }, { "epoch": 7.94, "learning_rate": 2.999020149814236e-09, "loss": 0.0217, "step": 8662 }, { "epoch": 7.94, "learning_rate": 2.9088336043903154e-09, "loss": 0.0154, "step": 8663 }, { "epoch": 7.94, "learning_rate": 2.820023687810647e-09, "loss": 0.0237, "step": 8664 }, { "epoch": 7.94, "learning_rate": 2.732590412304337e-09, "loss": 0.0209, "step": 8665 }, { "epoch": 7.94, "learning_rate": 2.6465337899095334e-09, "loss": 0.0184, "step": 8666 }, { "epoch": 7.94, "learning_rate": 2.5618538324778676e-09, "loss": 0.0178, "step": 8667 }, { "epoch": 7.95, "learning_rate": 2.478550551668901e-09, "loss": 0.0176, "step": 8668 }, { "epoch": 7.95, "learning_rate": 2.3966239589556793e-09, "loss": 0.015, "step": 8669 }, { "epoch": 7.95, "learning_rate": 2.316074065618068e-09, "loss": 0.0194, "step": 8670 }, { "epoch": 7.95, "learning_rate": 2.2369008827483053e-09, "loss": 0.0212, "step": 8671 }, { "epoch": 7.95, "learning_rate": 2.159104421248781e-09, "loss": 0.0186, "step": 8672 }, { "epoch": 7.95, "learning_rate": 2.0826846918331477e-09, "loss": 0.019, "step": 8673 }, { "epoch": 7.95, "learning_rate": 2.0076417050240994e-09, "loss": 0.0206, "step": 8674 }, { "epoch": 7.95, "learning_rate": 1.933975471154481e-09, "loss": 0.0206, "step": 8675 }, { "epoch": 7.95, "learning_rate": 1.8616860003695115e-09, "loss": 0.0227, "step": 8676 }, { "epoch": 7.95, "learning_rate": 1.7907733026223396e-09, "loss": 0.0159, "step": 8677 }, { "epoch": 7.95, "learning_rate": 1.7212373876795974e-09, "loss": 0.0164, "step": 8678 }, { "epoch": 7.96, "learning_rate": 1.653078265114738e-09, "loss": 0.0163, "step": 8679 }, { "epoch": 7.96, "learning_rate": 1.5862959443158077e-09, "loss": 0.0152, "step": 8680 }, { "epoch": 7.96, "learning_rate": 1.5208904344765629e-09, "loss": 0.0227, "step": 8681 }, { "epoch": 7.96, "learning_rate": 1.4568617446042432e-09, "loss": 0.0203, "step": 8682 }, { "epoch": 7.96, "learning_rate": 1.39420988351624e-09, "loss": 0.0167, "step": 8683 }, { "epoch": 7.96, "learning_rate": 1.3329348598400959e-09, "loss": 0.0178, "step": 8684 }, { "epoch": 7.96, "learning_rate": 1.2730366820135066e-09, "loss": 0.0163, "step": 8685 }, { "epoch": 7.96, "learning_rate": 1.214515358283208e-09, "loss": 0.0175, "step": 8686 }, { "epoch": 7.96, "learning_rate": 1.1573708967094199e-09, "loss": 0.0196, "step": 8687 }, { "epoch": 7.96, "learning_rate": 1.1016033051602927e-09, "loss": 0.0219, "step": 8688 }, { "epoch": 7.96, "learning_rate": 1.0472125913163489e-09, "loss": 0.0244, "step": 8689 }, { "epoch": 7.97, "learning_rate": 9.94198762664933e-10, "loss": 0.0208, "step": 8690 }, { "epoch": 7.97, "learning_rate": 9.425618265090918e-10, "loss": 0.0182, "step": 8691 }, { "epoch": 7.97, "learning_rate": 8.923017899575836e-10, "loss": 0.0205, "step": 8692 }, { "epoch": 7.97, "learning_rate": 8.434186599315386e-10, "loss": 0.0186, "step": 8693 }, { "epoch": 7.97, "learning_rate": 7.959124431622389e-10, "loss": 0.021, "step": 8694 }, { "epoch": 7.97, "learning_rate": 7.497831461922289e-10, "loss": 0.0162, "step": 8695 }, { "epoch": 7.97, "learning_rate": 7.050307753719843e-10, "loss": 0.0218, "step": 8696 }, { "epoch": 7.97, "learning_rate": 6.616553368665734e-10, "loss": 0.0141, "step": 8697 }, { "epoch": 7.97, "learning_rate": 6.196568366467758e-10, "loss": 0.0218, "step": 8698 }, { "epoch": 7.97, "learning_rate": 5.790352804968535e-10, "loss": 0.018, "step": 8699 }, { "epoch": 7.97, "learning_rate": 5.397906740112202e-10, "loss": 0.0217, "step": 8700 }, { "epoch": 7.98, "learning_rate": 5.019230225922212e-10, "loss": 0.0182, "step": 8701 }, { "epoch": 7.98, "learning_rate": 4.654323314545739e-10, "loss": 0.02, "step": 8702 }, { "epoch": 7.98, "learning_rate": 4.303186056253683e-10, "loss": 0.0281, "step": 8703 }, { "epoch": 7.98, "learning_rate": 3.9658184993740524e-10, "loss": 0.0199, "step": 8704 }, { "epoch": 7.98, "learning_rate": 3.64222069036968e-10, "loss": 0.0254, "step": 8705 }, { "epoch": 7.98, "learning_rate": 3.332392673804918e-10, "loss": 0.0169, "step": 8706 }, { "epoch": 7.98, "learning_rate": 3.0363344923456385e-10, "loss": 0.0248, "step": 8707 }, { "epoch": 7.98, "learning_rate": 2.7540461867592295e-10, "loss": 0.0166, "step": 8708 }, { "epoch": 7.98, "learning_rate": 2.485527795903497e-10, "loss": 0.0205, "step": 8709 }, { "epoch": 7.98, "learning_rate": 2.2307793567710733e-10, "loss": 0.0216, "step": 8710 }, { "epoch": 7.98, "learning_rate": 1.989800904445005e-10, "loss": 0.0181, "step": 8711 }, { "epoch": 7.99, "learning_rate": 1.7625924720876543e-10, "loss": 0.0218, "step": 8712 }, { "epoch": 7.99, "learning_rate": 1.54915409100731e-10, "loss": 0.0196, "step": 8713 }, { "epoch": 7.99, "learning_rate": 1.349485790591576e-10, "loss": 0.0203, "step": 8714 }, { "epoch": 7.99, "learning_rate": 1.1635875983184719e-10, "loss": 0.0211, "step": 8715 }, { "epoch": 7.99, "learning_rate": 9.914595398008431e-11, "loss": 0.0156, "step": 8716 }, { "epoch": 7.99, "learning_rate": 8.331016387419511e-11, "loss": 0.0245, "step": 8717 }, { "epoch": 7.99, "learning_rate": 6.88513916946576e-11, "loss": 0.0214, "step": 8718 }, { "epoch": 7.99, "learning_rate": 5.5769639432101675e-11, "loss": 0.0188, "step": 8719 }, { "epoch": 7.99, "learning_rate": 4.40649088884193e-11, "loss": 0.0194, "step": 8720 }, { "epoch": 7.99, "learning_rate": 3.373720167454409e-11, "loss": 0.0158, "step": 8721 }, { "epoch": 7.99, "learning_rate": 2.4786519213781947e-11, "loss": 0.0223, "step": 8722 }, { "epoch": 8.0, "learning_rate": 1.7212862737370216e-11, "loss": 0.0264, "step": 8723 }, { "epoch": 8.0, "learning_rate": 1.1016233290028766e-11, "loss": 0.0203, "step": 8724 }, { "epoch": 8.0, "learning_rate": 6.196631723298652e-12, "loss": 0.0163, "step": 8725 }, { "epoch": 8.0, "learning_rate": 2.754058702203466e-12, "loss": 0.0167, "step": 8726 }, { "epoch": 8.0, "learning_rate": 6.885146985879943e-13, "loss": 0.0209, "step": 8727 }, { "epoch": 8.0, "learning_rate": 0.0, "loss": 0.0151, "step": 8728 }, { "epoch": 8.0, "step": 8728, "total_flos": 2998912226918400.0, "train_loss": 0.02180936751933299, "train_runtime": 13747.5254, "train_samples_per_second": 81.229, "train_steps_per_second": 0.635 } ], "logging_steps": 1.0, "max_steps": 8728, "num_input_tokens_seen": 0, "num_train_epochs": 8, "save_steps": 500, "total_flos": 2998912226918400.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }