{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 5773, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.1494252873563219e-07, "loss": 0.9484, "step": 1 }, { "epoch": 0.0, "learning_rate": 2.2988505747126437e-07, "loss": 0.95, "step": 2 }, { "epoch": 0.0, "learning_rate": 3.4482758620689656e-07, "loss": 0.9177, "step": 3 }, { "epoch": 0.0, "learning_rate": 4.5977011494252875e-07, "loss": 0.9504, "step": 4 }, { "epoch": 0.0, "learning_rate": 5.747126436781609e-07, "loss": 0.9311, "step": 5 }, { "epoch": 0.0, "learning_rate": 6.896551724137931e-07, "loss": 0.9279, "step": 6 }, { "epoch": 0.0, "learning_rate": 8.045977011494253e-07, "loss": 0.9171, "step": 7 }, { "epoch": 0.0, "learning_rate": 9.195402298850575e-07, "loss": 0.922, "step": 8 }, { "epoch": 0.0, "learning_rate": 1.0344827586206898e-06, "loss": 0.9144, "step": 9 }, { "epoch": 0.0, "learning_rate": 1.1494252873563219e-06, "loss": 0.8822, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.2643678160919542e-06, "loss": 0.8678, "step": 11 }, { "epoch": 0.0, "learning_rate": 1.3793103448275862e-06, "loss": 0.8998, "step": 12 }, { "epoch": 0.0, "learning_rate": 1.4942528735632185e-06, "loss": 0.8748, "step": 13 }, { "epoch": 0.0, "learning_rate": 1.6091954022988506e-06, "loss": 0.8285, "step": 14 }, { "epoch": 0.0, "learning_rate": 1.724137931034483e-06, "loss": 0.8115, "step": 15 }, { "epoch": 0.0, "learning_rate": 1.839080459770115e-06, "loss": 0.8366, "step": 16 }, { "epoch": 0.0, "learning_rate": 1.9540229885057475e-06, "loss": 0.8117, "step": 17 }, { "epoch": 0.0, "learning_rate": 2.0689655172413796e-06, "loss": 0.8034, "step": 18 }, { "epoch": 0.0, "learning_rate": 2.1839080459770117e-06, "loss": 0.7927, "step": 19 }, { "epoch": 0.0, "learning_rate": 2.2988505747126437e-06, "loss": 0.7821, "step": 20 }, { "epoch": 0.0, "learning_rate": 2.4137931034482762e-06, "loss": 0.7701, "step": 21 }, { "epoch": 0.0, "learning_rate": 2.5287356321839083e-06, "loss": 0.7644, "step": 22 }, { "epoch": 0.0, "learning_rate": 2.6436781609195404e-06, "loss": 0.7802, "step": 23 }, { "epoch": 0.0, "learning_rate": 2.7586206896551725e-06, "loss": 0.7547, "step": 24 }, { "epoch": 0.0, "learning_rate": 2.8735632183908046e-06, "loss": 0.7396, "step": 25 }, { "epoch": 0.0, "learning_rate": 2.988505747126437e-06, "loss": 0.7298, "step": 26 }, { "epoch": 0.0, "learning_rate": 3.103448275862069e-06, "loss": 0.7564, "step": 27 }, { "epoch": 0.0, "learning_rate": 3.2183908045977012e-06, "loss": 0.75, "step": 28 }, { "epoch": 0.01, "learning_rate": 3.3333333333333333e-06, "loss": 0.7496, "step": 29 }, { "epoch": 0.01, "learning_rate": 3.448275862068966e-06, "loss": 0.7511, "step": 30 }, { "epoch": 0.01, "learning_rate": 3.563218390804598e-06, "loss": 0.7417, "step": 31 }, { "epoch": 0.01, "learning_rate": 3.67816091954023e-06, "loss": 0.7239, "step": 32 }, { "epoch": 0.01, "learning_rate": 3.793103448275862e-06, "loss": 0.7558, "step": 33 }, { "epoch": 0.01, "learning_rate": 3.908045977011495e-06, "loss": 0.7159, "step": 34 }, { "epoch": 0.01, "learning_rate": 4.022988505747127e-06, "loss": 0.704, "step": 35 }, { "epoch": 0.01, "learning_rate": 4.137931034482759e-06, "loss": 0.7202, "step": 36 }, { "epoch": 0.01, "learning_rate": 4.252873563218391e-06, "loss": 0.7301, "step": 37 }, { "epoch": 0.01, "learning_rate": 4.367816091954023e-06, "loss": 0.712, "step": 38 }, { "epoch": 0.01, "learning_rate": 4.482758620689656e-06, "loss": 0.7307, "step": 39 }, { "epoch": 0.01, "learning_rate": 4.5977011494252875e-06, "loss": 0.7201, "step": 40 }, { "epoch": 0.01, "learning_rate": 4.71264367816092e-06, "loss": 0.7153, "step": 41 }, { "epoch": 0.01, "learning_rate": 4.8275862068965525e-06, "loss": 0.7191, "step": 42 }, { "epoch": 0.01, "learning_rate": 4.942528735632184e-06, "loss": 0.7077, "step": 43 }, { "epoch": 0.01, "learning_rate": 5.057471264367817e-06, "loss": 0.7397, "step": 44 }, { "epoch": 0.01, "learning_rate": 5.172413793103449e-06, "loss": 0.7121, "step": 45 }, { "epoch": 0.01, "learning_rate": 5.287356321839081e-06, "loss": 0.7071, "step": 46 }, { "epoch": 0.01, "learning_rate": 5.402298850574713e-06, "loss": 0.7157, "step": 47 }, { "epoch": 0.01, "learning_rate": 5.517241379310345e-06, "loss": 0.7052, "step": 48 }, { "epoch": 0.01, "learning_rate": 5.6321839080459775e-06, "loss": 0.7093, "step": 49 }, { "epoch": 0.01, "learning_rate": 5.747126436781609e-06, "loss": 0.7005, "step": 50 }, { "epoch": 0.01, "learning_rate": 5.862068965517242e-06, "loss": 0.6826, "step": 51 }, { "epoch": 0.01, "learning_rate": 5.977011494252874e-06, "loss": 0.6902, "step": 52 }, { "epoch": 0.01, "learning_rate": 6.091954022988507e-06, "loss": 0.6901, "step": 53 }, { "epoch": 0.01, "learning_rate": 6.206896551724138e-06, "loss": 0.6878, "step": 54 }, { "epoch": 0.01, "learning_rate": 6.321839080459771e-06, "loss": 0.6942, "step": 55 }, { "epoch": 0.01, "learning_rate": 6.4367816091954025e-06, "loss": 0.6913, "step": 56 }, { "epoch": 0.01, "learning_rate": 6.551724137931035e-06, "loss": 0.6963, "step": 57 }, { "epoch": 0.01, "learning_rate": 6.666666666666667e-06, "loss": 0.7162, "step": 58 }, { "epoch": 0.01, "learning_rate": 6.781609195402299e-06, "loss": 0.6866, "step": 59 }, { "epoch": 0.01, "learning_rate": 6.896551724137932e-06, "loss": 0.6854, "step": 60 }, { "epoch": 0.01, "learning_rate": 7.011494252873564e-06, "loss": 0.6834, "step": 61 }, { "epoch": 0.01, "learning_rate": 7.126436781609196e-06, "loss": 0.6805, "step": 62 }, { "epoch": 0.01, "learning_rate": 7.241379310344828e-06, "loss": 0.7041, "step": 63 }, { "epoch": 0.01, "learning_rate": 7.35632183908046e-06, "loss": 0.662, "step": 64 }, { "epoch": 0.01, "learning_rate": 7.4712643678160925e-06, "loss": 0.6952, "step": 65 }, { "epoch": 0.01, "learning_rate": 7.586206896551724e-06, "loss": 0.6859, "step": 66 }, { "epoch": 0.01, "learning_rate": 7.701149425287356e-06, "loss": 0.6869, "step": 67 }, { "epoch": 0.01, "learning_rate": 7.81609195402299e-06, "loss": 0.6638, "step": 68 }, { "epoch": 0.01, "learning_rate": 7.93103448275862e-06, "loss": 0.6986, "step": 69 }, { "epoch": 0.01, "learning_rate": 8.045977011494253e-06, "loss": 0.6896, "step": 70 }, { "epoch": 0.01, "learning_rate": 8.160919540229886e-06, "loss": 0.6787, "step": 71 }, { "epoch": 0.01, "learning_rate": 8.275862068965518e-06, "loss": 0.6859, "step": 72 }, { "epoch": 0.01, "learning_rate": 8.390804597701149e-06, "loss": 0.6779, "step": 73 }, { "epoch": 0.01, "learning_rate": 8.505747126436782e-06, "loss": 0.6737, "step": 74 }, { "epoch": 0.01, "learning_rate": 8.620689655172414e-06, "loss": 0.6723, "step": 75 }, { "epoch": 0.01, "learning_rate": 8.735632183908047e-06, "loss": 0.6783, "step": 76 }, { "epoch": 0.01, "learning_rate": 8.85057471264368e-06, "loss": 0.6898, "step": 77 }, { "epoch": 0.01, "learning_rate": 8.965517241379312e-06, "loss": 0.6489, "step": 78 }, { "epoch": 0.01, "learning_rate": 9.080459770114942e-06, "loss": 0.6737, "step": 79 }, { "epoch": 0.01, "learning_rate": 9.195402298850575e-06, "loss": 0.681, "step": 80 }, { "epoch": 0.01, "learning_rate": 9.310344827586207e-06, "loss": 0.6784, "step": 81 }, { "epoch": 0.01, "learning_rate": 9.42528735632184e-06, "loss": 0.6903, "step": 82 }, { "epoch": 0.01, "learning_rate": 9.54022988505747e-06, "loss": 0.6695, "step": 83 }, { "epoch": 0.01, "learning_rate": 9.655172413793105e-06, "loss": 0.6616, "step": 84 }, { "epoch": 0.01, "learning_rate": 9.770114942528738e-06, "loss": 0.6745, "step": 85 }, { "epoch": 0.01, "learning_rate": 9.885057471264368e-06, "loss": 0.6705, "step": 86 }, { "epoch": 0.02, "learning_rate": 1e-05, "loss": 0.6864, "step": 87 }, { "epoch": 0.02, "learning_rate": 1.0114942528735633e-05, "loss": 0.692, "step": 88 }, { "epoch": 0.02, "learning_rate": 1.0229885057471264e-05, "loss": 0.682, "step": 89 }, { "epoch": 0.02, "learning_rate": 1.0344827586206898e-05, "loss": 0.6716, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.0459770114942529e-05, "loss": 0.647, "step": 91 }, { "epoch": 0.02, "learning_rate": 1.0574712643678162e-05, "loss": 0.6862, "step": 92 }, { "epoch": 0.02, "learning_rate": 1.0689655172413792e-05, "loss": 0.6731, "step": 93 }, { "epoch": 0.02, "learning_rate": 1.0804597701149427e-05, "loss": 0.6585, "step": 94 }, { "epoch": 0.02, "learning_rate": 1.091954022988506e-05, "loss": 0.6959, "step": 95 }, { "epoch": 0.02, "learning_rate": 1.103448275862069e-05, "loss": 0.6723, "step": 96 }, { "epoch": 0.02, "learning_rate": 1.1149425287356324e-05, "loss": 0.6741, "step": 97 }, { "epoch": 0.02, "learning_rate": 1.1264367816091955e-05, "loss": 0.6569, "step": 98 }, { "epoch": 0.02, "learning_rate": 1.1379310344827587e-05, "loss": 0.6839, "step": 99 }, { "epoch": 0.02, "learning_rate": 1.1494252873563218e-05, "loss": 0.6599, "step": 100 }, { "epoch": 0.02, "learning_rate": 1.1609195402298852e-05, "loss": 0.6688, "step": 101 }, { "epoch": 0.02, "learning_rate": 1.1724137931034483e-05, "loss": 0.6554, "step": 102 }, { "epoch": 0.02, "learning_rate": 1.1839080459770116e-05, "loss": 0.6618, "step": 103 }, { "epoch": 0.02, "learning_rate": 1.1954022988505748e-05, "loss": 0.6612, "step": 104 }, { "epoch": 0.02, "learning_rate": 1.206896551724138e-05, "loss": 0.6625, "step": 105 }, { "epoch": 0.02, "learning_rate": 1.2183908045977013e-05, "loss": 0.6691, "step": 106 }, { "epoch": 0.02, "learning_rate": 1.2298850574712644e-05, "loss": 0.6362, "step": 107 }, { "epoch": 0.02, "learning_rate": 1.2413793103448277e-05, "loss": 0.643, "step": 108 }, { "epoch": 0.02, "learning_rate": 1.2528735632183907e-05, "loss": 0.6652, "step": 109 }, { "epoch": 0.02, "learning_rate": 1.2643678160919542e-05, "loss": 0.6515, "step": 110 }, { "epoch": 0.02, "learning_rate": 1.2758620689655174e-05, "loss": 0.6379, "step": 111 }, { "epoch": 0.02, "learning_rate": 1.2873563218390805e-05, "loss": 0.653, "step": 112 }, { "epoch": 0.02, "learning_rate": 1.298850574712644e-05, "loss": 0.6801, "step": 113 }, { "epoch": 0.02, "learning_rate": 1.310344827586207e-05, "loss": 0.6515, "step": 114 }, { "epoch": 0.02, "learning_rate": 1.3218390804597702e-05, "loss": 0.6486, "step": 115 }, { "epoch": 0.02, "learning_rate": 1.3333333333333333e-05, "loss": 0.638, "step": 116 }, { "epoch": 0.02, "learning_rate": 1.3448275862068967e-05, "loss": 0.6468, "step": 117 }, { "epoch": 0.02, "learning_rate": 1.3563218390804598e-05, "loss": 0.6605, "step": 118 }, { "epoch": 0.02, "learning_rate": 1.367816091954023e-05, "loss": 0.6476, "step": 119 }, { "epoch": 0.02, "learning_rate": 1.3793103448275863e-05, "loss": 0.6396, "step": 120 }, { "epoch": 0.02, "learning_rate": 1.3908045977011496e-05, "loss": 0.649, "step": 121 }, { "epoch": 0.02, "learning_rate": 1.4022988505747128e-05, "loss": 0.6456, "step": 122 }, { "epoch": 0.02, "learning_rate": 1.4137931034482759e-05, "loss": 0.6465, "step": 123 }, { "epoch": 0.02, "learning_rate": 1.4252873563218392e-05, "loss": 0.6495, "step": 124 }, { "epoch": 0.02, "learning_rate": 1.4367816091954022e-05, "loss": 0.6575, "step": 125 }, { "epoch": 0.02, "learning_rate": 1.4482758620689657e-05, "loss": 0.6484, "step": 126 }, { "epoch": 0.02, "learning_rate": 1.459770114942529e-05, "loss": 0.6599, "step": 127 }, { "epoch": 0.02, "learning_rate": 1.471264367816092e-05, "loss": 0.6501, "step": 128 }, { "epoch": 0.02, "learning_rate": 1.4827586206896554e-05, "loss": 0.6442, "step": 129 }, { "epoch": 0.02, "learning_rate": 1.4942528735632185e-05, "loss": 0.6614, "step": 130 }, { "epoch": 0.02, "learning_rate": 1.5057471264367817e-05, "loss": 0.6597, "step": 131 }, { "epoch": 0.02, "learning_rate": 1.5172413793103448e-05, "loss": 0.6476, "step": 132 }, { "epoch": 0.02, "learning_rate": 1.528735632183908e-05, "loss": 0.6354, "step": 133 }, { "epoch": 0.02, "learning_rate": 1.540229885057471e-05, "loss": 0.6382, "step": 134 }, { "epoch": 0.02, "learning_rate": 1.5517241379310346e-05, "loss": 0.6321, "step": 135 }, { "epoch": 0.02, "learning_rate": 1.563218390804598e-05, "loss": 0.6458, "step": 136 }, { "epoch": 0.02, "learning_rate": 1.574712643678161e-05, "loss": 0.6523, "step": 137 }, { "epoch": 0.02, "learning_rate": 1.586206896551724e-05, "loss": 0.6389, "step": 138 }, { "epoch": 0.02, "learning_rate": 1.5977011494252876e-05, "loss": 0.6269, "step": 139 }, { "epoch": 0.02, "learning_rate": 1.6091954022988507e-05, "loss": 0.6324, "step": 140 }, { "epoch": 0.02, "learning_rate": 1.6206896551724137e-05, "loss": 0.6405, "step": 141 }, { "epoch": 0.02, "learning_rate": 1.632183908045977e-05, "loss": 0.6517, "step": 142 }, { "epoch": 0.02, "learning_rate": 1.6436781609195406e-05, "loss": 0.6412, "step": 143 }, { "epoch": 0.02, "learning_rate": 1.6551724137931037e-05, "loss": 0.6369, "step": 144 }, { "epoch": 0.03, "learning_rate": 1.6666666666666667e-05, "loss": 0.6206, "step": 145 }, { "epoch": 0.03, "learning_rate": 1.6781609195402298e-05, "loss": 0.6327, "step": 146 }, { "epoch": 0.03, "learning_rate": 1.6896551724137932e-05, "loss": 0.6432, "step": 147 }, { "epoch": 0.03, "learning_rate": 1.7011494252873563e-05, "loss": 0.6379, "step": 148 }, { "epoch": 0.03, "learning_rate": 1.7126436781609197e-05, "loss": 0.6332, "step": 149 }, { "epoch": 0.03, "learning_rate": 1.7241379310344828e-05, "loss": 0.632, "step": 150 }, { "epoch": 0.03, "learning_rate": 1.7356321839080462e-05, "loss": 0.6294, "step": 151 }, { "epoch": 0.03, "learning_rate": 1.7471264367816093e-05, "loss": 0.6389, "step": 152 }, { "epoch": 0.03, "learning_rate": 1.7586206896551724e-05, "loss": 0.625, "step": 153 }, { "epoch": 0.03, "learning_rate": 1.770114942528736e-05, "loss": 0.6336, "step": 154 }, { "epoch": 0.03, "learning_rate": 1.781609195402299e-05, "loss": 0.6247, "step": 155 }, { "epoch": 0.03, "learning_rate": 1.7931034482758623e-05, "loss": 0.632, "step": 156 }, { "epoch": 0.03, "learning_rate": 1.8045977011494254e-05, "loss": 0.6346, "step": 157 }, { "epoch": 0.03, "learning_rate": 1.8160919540229885e-05, "loss": 0.6394, "step": 158 }, { "epoch": 0.03, "learning_rate": 1.827586206896552e-05, "loss": 0.6375, "step": 159 }, { "epoch": 0.03, "learning_rate": 1.839080459770115e-05, "loss": 0.6443, "step": 160 }, { "epoch": 0.03, "learning_rate": 1.8505747126436784e-05, "loss": 0.6303, "step": 161 }, { "epoch": 0.03, "learning_rate": 1.8620689655172415e-05, "loss": 0.6395, "step": 162 }, { "epoch": 0.03, "learning_rate": 1.873563218390805e-05, "loss": 0.6484, "step": 163 }, { "epoch": 0.03, "learning_rate": 1.885057471264368e-05, "loss": 0.6404, "step": 164 }, { "epoch": 0.03, "learning_rate": 1.896551724137931e-05, "loss": 0.6204, "step": 165 }, { "epoch": 0.03, "learning_rate": 1.908045977011494e-05, "loss": 0.64, "step": 166 }, { "epoch": 0.03, "learning_rate": 1.9195402298850576e-05, "loss": 0.6064, "step": 167 }, { "epoch": 0.03, "learning_rate": 1.931034482758621e-05, "loss": 0.6252, "step": 168 }, { "epoch": 0.03, "learning_rate": 1.942528735632184e-05, "loss": 0.6307, "step": 169 }, { "epoch": 0.03, "learning_rate": 1.9540229885057475e-05, "loss": 0.6399, "step": 170 }, { "epoch": 0.03, "learning_rate": 1.9655172413793106e-05, "loss": 0.6291, "step": 171 }, { "epoch": 0.03, "learning_rate": 1.9770114942528737e-05, "loss": 0.6556, "step": 172 }, { "epoch": 0.03, "learning_rate": 1.9885057471264367e-05, "loss": 0.633, "step": 173 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 0.6308, "step": 174 }, { "epoch": 0.03, "learning_rate": 1.9999998425840254e-05, "loss": 0.6344, "step": 175 }, { "epoch": 0.03, "learning_rate": 1.99999937033615e-05, "loss": 0.6327, "step": 176 }, { "epoch": 0.03, "learning_rate": 1.9999985832565235e-05, "loss": 0.6402, "step": 177 }, { "epoch": 0.03, "learning_rate": 1.999997481345393e-05, "loss": 0.6353, "step": 178 }, { "epoch": 0.03, "learning_rate": 1.999996064603106e-05, "loss": 0.6332, "step": 179 }, { "epoch": 0.03, "learning_rate": 1.9999943330301076e-05, "loss": 0.6242, "step": 180 }, { "epoch": 0.03, "learning_rate": 1.9999922866269443e-05, "loss": 0.6324, "step": 181 }, { "epoch": 0.03, "learning_rate": 1.999989925394259e-05, "loss": 0.6229, "step": 182 }, { "epoch": 0.03, "learning_rate": 1.999987249332796e-05, "loss": 0.6267, "step": 183 }, { "epoch": 0.03, "learning_rate": 1.9999842584433976e-05, "loss": 0.6369, "step": 184 }, { "epoch": 0.03, "learning_rate": 1.9999809527270053e-05, "loss": 0.6204, "step": 185 }, { "epoch": 0.03, "learning_rate": 1.9999773321846598e-05, "loss": 0.6399, "step": 186 }, { "epoch": 0.03, "learning_rate": 1.9999733968175014e-05, "loss": 0.6158, "step": 187 }, { "epoch": 0.03, "learning_rate": 1.9999691466267683e-05, "loss": 0.6203, "step": 188 }, { "epoch": 0.03, "learning_rate": 1.9999645816137994e-05, "loss": 0.6336, "step": 189 }, { "epoch": 0.03, "learning_rate": 1.9999597017800315e-05, "loss": 0.6087, "step": 190 }, { "epoch": 0.03, "learning_rate": 1.9999545071270006e-05, "loss": 0.6241, "step": 191 }, { "epoch": 0.03, "learning_rate": 1.999948997656343e-05, "loss": 0.641, "step": 192 }, { "epoch": 0.03, "learning_rate": 1.9999431733697928e-05, "loss": 0.6503, "step": 193 }, { "epoch": 0.03, "learning_rate": 1.9999370342691836e-05, "loss": 0.6344, "step": 194 }, { "epoch": 0.03, "learning_rate": 1.999930580356448e-05, "loss": 0.6391, "step": 195 }, { "epoch": 0.03, "learning_rate": 1.999923811633618e-05, "loss": 0.6198, "step": 196 }, { "epoch": 0.03, "learning_rate": 1.9999167281028252e-05, "loss": 0.6342, "step": 197 }, { "epoch": 0.03, "learning_rate": 1.999909329766299e-05, "loss": 0.6203, "step": 198 }, { "epoch": 0.03, "learning_rate": 1.999901616626369e-05, "loss": 0.6187, "step": 199 }, { "epoch": 0.03, "learning_rate": 1.9998935886854634e-05, "loss": 0.6222, "step": 200 }, { "epoch": 0.03, "learning_rate": 1.99988524594611e-05, "loss": 0.6203, "step": 201 }, { "epoch": 0.03, "learning_rate": 1.9998765884109348e-05, "loss": 0.6475, "step": 202 }, { "epoch": 0.04, "learning_rate": 1.9998676160826637e-05, "loss": 0.617, "step": 203 }, { "epoch": 0.04, "learning_rate": 1.9998583289641214e-05, "loss": 0.6321, "step": 204 }, { "epoch": 0.04, "learning_rate": 1.999848727058232e-05, "loss": 0.6227, "step": 205 }, { "epoch": 0.04, "learning_rate": 1.9998388103680186e-05, "loss": 0.6272, "step": 206 }, { "epoch": 0.04, "learning_rate": 1.9998285788966027e-05, "loss": 0.6111, "step": 207 }, { "epoch": 0.04, "learning_rate": 1.999818032647206e-05, "loss": 0.6265, "step": 208 }, { "epoch": 0.04, "learning_rate": 1.999807171623149e-05, "loss": 0.6103, "step": 209 }, { "epoch": 0.04, "learning_rate": 1.9997959958278502e-05, "loss": 0.6242, "step": 210 }, { "epoch": 0.04, "learning_rate": 1.9997845052648288e-05, "loss": 0.6296, "step": 211 }, { "epoch": 0.04, "learning_rate": 1.999772699937702e-05, "loss": 0.6518, "step": 212 }, { "epoch": 0.04, "learning_rate": 1.9997605798501873e-05, "loss": 0.6436, "step": 213 }, { "epoch": 0.04, "learning_rate": 1.9997481450061e-05, "loss": 0.6357, "step": 214 }, { "epoch": 0.04, "learning_rate": 1.9997353954093545e-05, "loss": 0.6404, "step": 215 }, { "epoch": 0.04, "learning_rate": 1.9997223310639652e-05, "loss": 0.6359, "step": 216 }, { "epoch": 0.04, "learning_rate": 1.9997089519740455e-05, "loss": 0.6312, "step": 217 }, { "epoch": 0.04, "learning_rate": 1.999695258143807e-05, "loss": 0.6377, "step": 218 }, { "epoch": 0.04, "learning_rate": 1.9996812495775612e-05, "loss": 0.6157, "step": 219 }, { "epoch": 0.04, "learning_rate": 1.9996669262797183e-05, "loss": 0.6222, "step": 220 }, { "epoch": 0.04, "learning_rate": 1.999652288254788e-05, "loss": 0.6137, "step": 221 }, { "epoch": 0.04, "learning_rate": 1.999637335507379e-05, "loss": 0.6331, "step": 222 }, { "epoch": 0.04, "learning_rate": 1.999622068042198e-05, "loss": 0.6109, "step": 223 }, { "epoch": 0.04, "learning_rate": 1.9996064858640526e-05, "loss": 0.6034, "step": 224 }, { "epoch": 0.04, "learning_rate": 1.999590588977848e-05, "loss": 0.6281, "step": 225 }, { "epoch": 0.04, "learning_rate": 1.9995743773885896e-05, "loss": 0.6236, "step": 226 }, { "epoch": 0.04, "learning_rate": 1.999557851101381e-05, "loss": 0.6178, "step": 227 }, { "epoch": 0.04, "learning_rate": 1.999541010121425e-05, "loss": 0.6191, "step": 228 }, { "epoch": 0.04, "learning_rate": 1.999523854454024e-05, "loss": 0.6195, "step": 229 }, { "epoch": 0.04, "learning_rate": 1.9995063841045793e-05, "loss": 0.6062, "step": 230 }, { "epoch": 0.04, "learning_rate": 1.9994885990785903e-05, "loss": 0.6265, "step": 231 }, { "epoch": 0.04, "learning_rate": 1.9994704993816575e-05, "loss": 0.6169, "step": 232 }, { "epoch": 0.04, "learning_rate": 1.999452085019478e-05, "loss": 0.63, "step": 233 }, { "epoch": 0.04, "learning_rate": 1.9994333559978502e-05, "loss": 0.6159, "step": 234 }, { "epoch": 0.04, "learning_rate": 1.9994143123226702e-05, "loss": 0.6054, "step": 235 }, { "epoch": 0.04, "learning_rate": 1.999394953999934e-05, "loss": 0.6233, "step": 236 }, { "epoch": 0.04, "learning_rate": 1.9993752810357353e-05, "loss": 0.6193, "step": 237 }, { "epoch": 0.04, "learning_rate": 1.9993552934362684e-05, "loss": 0.6157, "step": 238 }, { "epoch": 0.04, "learning_rate": 1.9993349912078265e-05, "loss": 0.6402, "step": 239 }, { "epoch": 0.04, "learning_rate": 1.9993143743568e-05, "loss": 0.6184, "step": 240 }, { "epoch": 0.04, "learning_rate": 1.999293442889681e-05, "loss": 0.6102, "step": 241 }, { "epoch": 0.04, "learning_rate": 1.9992721968130588e-05, "loss": 0.5927, "step": 242 }, { "epoch": 0.04, "learning_rate": 1.9992506361336227e-05, "loss": 0.6344, "step": 243 }, { "epoch": 0.04, "learning_rate": 1.9992287608581605e-05, "loss": 0.6077, "step": 244 }, { "epoch": 0.04, "learning_rate": 1.999206570993559e-05, "loss": 0.6181, "step": 245 }, { "epoch": 0.04, "learning_rate": 1.9991840665468047e-05, "loss": 0.6215, "step": 246 }, { "epoch": 0.04, "learning_rate": 1.9991612475249827e-05, "loss": 0.6308, "step": 247 }, { "epoch": 0.04, "learning_rate": 1.9991381139352767e-05, "loss": 0.6446, "step": 248 }, { "epoch": 0.04, "learning_rate": 1.9991146657849705e-05, "loss": 0.6091, "step": 249 }, { "epoch": 0.04, "learning_rate": 1.999090903081446e-05, "loss": 0.6265, "step": 250 }, { "epoch": 0.04, "learning_rate": 1.999066825832184e-05, "loss": 0.6095, "step": 251 }, { "epoch": 0.04, "learning_rate": 1.9990424340447658e-05, "loss": 0.6304, "step": 252 }, { "epoch": 0.04, "learning_rate": 1.9990177277268698e-05, "loss": 0.6213, "step": 253 }, { "epoch": 0.04, "learning_rate": 1.998992706886275e-05, "loss": 0.6093, "step": 254 }, { "epoch": 0.04, "learning_rate": 1.9989673715308582e-05, "loss": 0.6306, "step": 255 }, { "epoch": 0.04, "learning_rate": 1.998941721668596e-05, "loss": 0.6159, "step": 256 }, { "epoch": 0.04, "learning_rate": 1.9989157573075642e-05, "loss": 0.6099, "step": 257 }, { "epoch": 0.04, "learning_rate": 1.9988894784559366e-05, "loss": 0.622, "step": 258 }, { "epoch": 0.04, "learning_rate": 1.9988628851219872e-05, "loss": 0.6336, "step": 259 }, { "epoch": 0.05, "learning_rate": 1.9988359773140877e-05, "loss": 0.6199, "step": 260 }, { "epoch": 0.05, "learning_rate": 1.9988087550407102e-05, "loss": 0.6125, "step": 261 }, { "epoch": 0.05, "learning_rate": 1.998781218310425e-05, "loss": 0.6242, "step": 262 }, { "epoch": 0.05, "learning_rate": 1.9987533671319013e-05, "loss": 0.6283, "step": 263 }, { "epoch": 0.05, "learning_rate": 1.9987252015139074e-05, "loss": 0.6259, "step": 264 }, { "epoch": 0.05, "learning_rate": 1.9986967214653114e-05, "loss": 0.6323, "step": 265 }, { "epoch": 0.05, "learning_rate": 1.998667926995079e-05, "loss": 0.622, "step": 266 }, { "epoch": 0.05, "learning_rate": 1.9986388181122762e-05, "loss": 0.6234, "step": 267 }, { "epoch": 0.05, "learning_rate": 1.998609394826067e-05, "loss": 0.611, "step": 268 }, { "epoch": 0.05, "learning_rate": 1.998579657145715e-05, "loss": 0.6196, "step": 269 }, { "epoch": 0.05, "learning_rate": 1.998549605080583e-05, "loss": 0.6275, "step": 270 }, { "epoch": 0.05, "learning_rate": 1.9985192386401314e-05, "loss": 0.6154, "step": 271 }, { "epoch": 0.05, "learning_rate": 1.998488557833921e-05, "loss": 0.6367, "step": 272 }, { "epoch": 0.05, "learning_rate": 1.9984575626716113e-05, "loss": 0.6135, "step": 273 }, { "epoch": 0.05, "learning_rate": 1.99842625316296e-05, "loss": 0.6162, "step": 274 }, { "epoch": 0.05, "learning_rate": 1.998394629317825e-05, "loss": 0.6284, "step": 275 }, { "epoch": 0.05, "learning_rate": 1.9983626911461625e-05, "loss": 0.6147, "step": 276 }, { "epoch": 0.05, "learning_rate": 1.9983304386580267e-05, "loss": 0.6059, "step": 277 }, { "epoch": 0.05, "learning_rate": 1.9982978718635727e-05, "loss": 0.6096, "step": 278 }, { "epoch": 0.05, "learning_rate": 1.9982649907730534e-05, "loss": 0.603, "step": 279 }, { "epoch": 0.05, "learning_rate": 1.9982317953968204e-05, "loss": 0.6226, "step": 280 }, { "epoch": 0.05, "learning_rate": 1.998198285745325e-05, "loss": 0.631, "step": 281 }, { "epoch": 0.05, "learning_rate": 1.998164461829117e-05, "loss": 0.6162, "step": 282 }, { "epoch": 0.05, "learning_rate": 1.9981303236588455e-05, "loss": 0.6211, "step": 283 }, { "epoch": 0.05, "learning_rate": 1.9980958712452577e-05, "loss": 0.6206, "step": 284 }, { "epoch": 0.05, "learning_rate": 1.998061104599201e-05, "loss": 0.6216, "step": 285 }, { "epoch": 0.05, "learning_rate": 1.9980260237316205e-05, "loss": 0.6063, "step": 286 }, { "epoch": 0.05, "learning_rate": 1.997990628653561e-05, "loss": 0.6263, "step": 287 }, { "epoch": 0.05, "learning_rate": 1.997954919376166e-05, "loss": 0.62, "step": 288 }, { "epoch": 0.05, "learning_rate": 1.9979188959106784e-05, "loss": 0.6097, "step": 289 }, { "epoch": 0.05, "learning_rate": 1.9978825582684384e-05, "loss": 0.6127, "step": 290 }, { "epoch": 0.05, "learning_rate": 1.9978459064608874e-05, "loss": 0.6249, "step": 291 }, { "epoch": 0.05, "learning_rate": 1.9978089404995642e-05, "loss": 0.6246, "step": 292 }, { "epoch": 0.05, "learning_rate": 1.9977716603961065e-05, "loss": 0.612, "step": 293 }, { "epoch": 0.05, "learning_rate": 1.9977340661622513e-05, "loss": 0.5941, "step": 294 }, { "epoch": 0.05, "learning_rate": 1.9976961578098352e-05, "loss": 0.6133, "step": 295 }, { "epoch": 0.05, "learning_rate": 1.997657935350792e-05, "loss": 0.6039, "step": 296 }, { "epoch": 0.05, "learning_rate": 1.997619398797156e-05, "loss": 0.6123, "step": 297 }, { "epoch": 0.05, "learning_rate": 1.9975805481610594e-05, "loss": 0.6188, "step": 298 }, { "epoch": 0.05, "learning_rate": 1.997541383454734e-05, "loss": 0.6135, "step": 299 }, { "epoch": 0.05, "learning_rate": 1.9975019046905096e-05, "loss": 0.6042, "step": 300 }, { "epoch": 0.05, "learning_rate": 1.997462111880816e-05, "loss": 0.599, "step": 301 }, { "epoch": 0.05, "learning_rate": 1.9974220050381805e-05, "loss": 0.6022, "step": 302 }, { "epoch": 0.05, "learning_rate": 1.9973815841752303e-05, "loss": 0.6387, "step": 303 }, { "epoch": 0.05, "learning_rate": 1.9973408493046917e-05, "loss": 0.6156, "step": 304 }, { "epoch": 0.05, "learning_rate": 1.9972998004393886e-05, "loss": 0.6115, "step": 305 }, { "epoch": 0.05, "learning_rate": 1.9972584375922453e-05, "loss": 0.6183, "step": 306 }, { "epoch": 0.05, "learning_rate": 1.997216760776283e-05, "loss": 0.6074, "step": 307 }, { "epoch": 0.05, "learning_rate": 1.997174770004624e-05, "loss": 0.6039, "step": 308 }, { "epoch": 0.05, "learning_rate": 1.9971324652904877e-05, "loss": 0.6359, "step": 309 }, { "epoch": 0.05, "learning_rate": 1.9970898466471934e-05, "loss": 0.6073, "step": 310 }, { "epoch": 0.05, "learning_rate": 1.9970469140881582e-05, "loss": 0.6202, "step": 311 }, { "epoch": 0.05, "learning_rate": 1.997003667626899e-05, "loss": 0.6165, "step": 312 }, { "epoch": 0.05, "learning_rate": 1.9969601072770314e-05, "loss": 0.6435, "step": 313 }, { "epoch": 0.05, "learning_rate": 1.9969162330522693e-05, "loss": 0.6184, "step": 314 }, { "epoch": 0.05, "learning_rate": 1.996872044966426e-05, "loss": 0.616, "step": 315 }, { "epoch": 0.05, "learning_rate": 1.9968275430334128e-05, "loss": 0.6168, "step": 316 }, { "epoch": 0.05, "learning_rate": 1.9967827272672407e-05, "loss": 0.6143, "step": 317 }, { "epoch": 0.06, "learning_rate": 1.996737597682019e-05, "loss": 0.6193, "step": 318 }, { "epoch": 0.06, "learning_rate": 1.9966921542919565e-05, "loss": 0.6214, "step": 319 }, { "epoch": 0.06, "learning_rate": 1.9966463971113592e-05, "loss": 0.6383, "step": 320 }, { "epoch": 0.06, "learning_rate": 1.9966003261546335e-05, "loss": 0.6318, "step": 321 }, { "epoch": 0.06, "learning_rate": 1.9965539414362843e-05, "loss": 0.6191, "step": 322 }, { "epoch": 0.06, "learning_rate": 1.996507242970914e-05, "loss": 0.6317, "step": 323 }, { "epoch": 0.06, "learning_rate": 1.996460230773226e-05, "loss": 0.6119, "step": 324 }, { "epoch": 0.06, "learning_rate": 1.9964129048580206e-05, "loss": 0.6235, "step": 325 }, { "epoch": 0.06, "learning_rate": 1.9963652652401976e-05, "loss": 0.5966, "step": 326 }, { "epoch": 0.06, "learning_rate": 1.996317311934755e-05, "loss": 0.6118, "step": 327 }, { "epoch": 0.06, "learning_rate": 1.996269044956791e-05, "loss": 0.6134, "step": 328 }, { "epoch": 0.06, "learning_rate": 1.9962204643215012e-05, "loss": 0.6047, "step": 329 }, { "epoch": 0.06, "learning_rate": 1.9961715700441793e-05, "loss": 0.6179, "step": 330 }, { "epoch": 0.06, "learning_rate": 1.9961223621402206e-05, "loss": 0.5981, "step": 331 }, { "epoch": 0.06, "learning_rate": 1.996072840625116e-05, "loss": 0.6077, "step": 332 }, { "epoch": 0.06, "learning_rate": 1.996023005514457e-05, "loss": 0.607, "step": 333 }, { "epoch": 0.06, "learning_rate": 1.995972856823933e-05, "loss": 0.6184, "step": 334 }, { "epoch": 0.06, "learning_rate": 1.9959223945693325e-05, "loss": 0.6179, "step": 335 }, { "epoch": 0.06, "learning_rate": 1.995871618766543e-05, "loss": 0.6067, "step": 336 }, { "epoch": 0.06, "learning_rate": 1.99582052943155e-05, "loss": 0.6283, "step": 337 }, { "epoch": 0.06, "learning_rate": 1.995769126580438e-05, "loss": 0.6102, "step": 338 }, { "epoch": 0.06, "learning_rate": 1.99571741022939e-05, "loss": 0.6031, "step": 339 }, { "epoch": 0.06, "learning_rate": 1.9956653803946888e-05, "loss": 0.6226, "step": 340 }, { "epoch": 0.06, "learning_rate": 1.9956130370927145e-05, "loss": 0.618, "step": 341 }, { "epoch": 0.06, "learning_rate": 1.9955603803399462e-05, "loss": 0.6083, "step": 342 }, { "epoch": 0.06, "learning_rate": 1.9955074101529625e-05, "loss": 0.6137, "step": 343 }, { "epoch": 0.06, "learning_rate": 1.9954541265484394e-05, "loss": 0.6179, "step": 344 }, { "epoch": 0.06, "learning_rate": 1.9954005295431532e-05, "loss": 0.6029, "step": 345 }, { "epoch": 0.06, "learning_rate": 1.9953466191539775e-05, "loss": 0.6218, "step": 346 }, { "epoch": 0.06, "learning_rate": 1.995292395397885e-05, "loss": 0.6241, "step": 347 }, { "epoch": 0.06, "learning_rate": 1.9952378582919464e-05, "loss": 0.6216, "step": 348 }, { "epoch": 0.06, "learning_rate": 1.995183007853333e-05, "loss": 0.6077, "step": 349 }, { "epoch": 0.06, "learning_rate": 1.9951278440993128e-05, "loss": 0.6268, "step": 350 }, { "epoch": 0.06, "learning_rate": 1.9950723670472533e-05, "loss": 0.6047, "step": 351 }, { "epoch": 0.06, "learning_rate": 1.99501657671462e-05, "loss": 0.6095, "step": 352 }, { "epoch": 0.06, "learning_rate": 1.9949604731189778e-05, "loss": 0.6173, "step": 353 }, { "epoch": 0.06, "learning_rate": 1.99490405627799e-05, "loss": 0.6109, "step": 354 }, { "epoch": 0.06, "learning_rate": 1.994847326209418e-05, "loss": 0.5984, "step": 355 }, { "epoch": 0.06, "learning_rate": 1.9947902829311227e-05, "loss": 0.6131, "step": 356 }, { "epoch": 0.06, "learning_rate": 1.994732926461063e-05, "loss": 0.6298, "step": 357 }, { "epoch": 0.06, "learning_rate": 1.994675256817297e-05, "loss": 0.602, "step": 358 }, { "epoch": 0.06, "learning_rate": 1.99461727401798e-05, "loss": 0.6151, "step": 359 }, { "epoch": 0.06, "learning_rate": 1.9945589780813677e-05, "loss": 0.6279, "step": 360 }, { "epoch": 0.06, "learning_rate": 1.9945003690258127e-05, "loss": 0.6264, "step": 361 }, { "epoch": 0.06, "learning_rate": 1.994441446869768e-05, "loss": 0.6115, "step": 362 }, { "epoch": 0.06, "learning_rate": 1.994382211631783e-05, "loss": 0.5953, "step": 363 }, { "epoch": 0.06, "learning_rate": 1.9943226633305078e-05, "loss": 0.6248, "step": 364 }, { "epoch": 0.06, "learning_rate": 1.9942628019846898e-05, "loss": 0.597, "step": 365 }, { "epoch": 0.06, "learning_rate": 1.9942026276131752e-05, "loss": 0.5972, "step": 366 }, { "epoch": 0.06, "learning_rate": 1.9941421402349092e-05, "loss": 0.6029, "step": 367 }, { "epoch": 0.06, "learning_rate": 1.9940813398689344e-05, "loss": 0.6114, "step": 368 }, { "epoch": 0.06, "learning_rate": 1.9940202265343935e-05, "loss": 0.5961, "step": 369 }, { "epoch": 0.06, "learning_rate": 1.9939588002505264e-05, "loss": 0.6229, "step": 370 }, { "epoch": 0.06, "learning_rate": 1.9938970610366722e-05, "loss": 0.5918, "step": 371 }, { "epoch": 0.06, "learning_rate": 1.993835008912268e-05, "loss": 0.6277, "step": 372 }, { "epoch": 0.06, "learning_rate": 1.993772643896851e-05, "loss": 0.6041, "step": 373 }, { "epoch": 0.06, "learning_rate": 1.993709966010054e-05, "loss": 0.6054, "step": 374 }, { "epoch": 0.06, "learning_rate": 1.9936469752716116e-05, "loss": 0.6097, "step": 375 }, { "epoch": 0.07, "learning_rate": 1.9935836717013544e-05, "loss": 0.6088, "step": 376 }, { "epoch": 0.07, "learning_rate": 1.9935200553192124e-05, "loss": 0.5998, "step": 377 }, { "epoch": 0.07, "learning_rate": 1.9934561261452142e-05, "loss": 0.6098, "step": 378 }, { "epoch": 0.07, "learning_rate": 1.993391884199487e-05, "loss": 0.6132, "step": 379 }, { "epoch": 0.07, "learning_rate": 1.993327329502256e-05, "loss": 0.6082, "step": 380 }, { "epoch": 0.07, "learning_rate": 1.9932624620738448e-05, "loss": 0.6104, "step": 381 }, { "epoch": 0.07, "learning_rate": 1.9931972819346762e-05, "loss": 0.6089, "step": 382 }, { "epoch": 0.07, "learning_rate": 1.9931317891052707e-05, "loss": 0.6008, "step": 383 }, { "epoch": 0.07, "learning_rate": 1.9930659836062478e-05, "loss": 0.6191, "step": 384 }, { "epoch": 0.07, "learning_rate": 1.992999865458325e-05, "loss": 0.6038, "step": 385 }, { "epoch": 0.07, "learning_rate": 1.9929334346823185e-05, "loss": 0.6262, "step": 386 }, { "epoch": 0.07, "learning_rate": 1.9928666912991425e-05, "loss": 0.6122, "step": 387 }, { "epoch": 0.07, "learning_rate": 1.9927996353298105e-05, "loss": 0.5922, "step": 388 }, { "epoch": 0.07, "learning_rate": 1.992732266795433e-05, "loss": 0.6182, "step": 389 }, { "epoch": 0.07, "learning_rate": 1.9926645857172213e-05, "loss": 0.6177, "step": 390 }, { "epoch": 0.07, "learning_rate": 1.992596592116482e-05, "loss": 0.6056, "step": 391 }, { "epoch": 0.07, "learning_rate": 1.992528286014622e-05, "loss": 0.6314, "step": 392 }, { "epoch": 0.07, "learning_rate": 1.992459667433147e-05, "loss": 0.604, "step": 393 }, { "epoch": 0.07, "learning_rate": 1.9923907363936593e-05, "loss": 0.621, "step": 394 }, { "epoch": 0.07, "learning_rate": 1.9923214929178617e-05, "loss": 0.6169, "step": 395 }, { "epoch": 0.07, "learning_rate": 1.9922519370275536e-05, "loss": 0.6061, "step": 396 }, { "epoch": 0.07, "learning_rate": 1.992182068744633e-05, "loss": 0.6069, "step": 397 }, { "epoch": 0.07, "learning_rate": 1.9921118880910976e-05, "loss": 0.6186, "step": 398 }, { "epoch": 0.07, "learning_rate": 1.992041395089042e-05, "loss": 0.6042, "step": 399 }, { "epoch": 0.07, "learning_rate": 1.9919705897606596e-05, "loss": 0.6093, "step": 400 }, { "epoch": 0.07, "learning_rate": 1.9918994721282423e-05, "loss": 0.5962, "step": 401 }, { "epoch": 0.07, "learning_rate": 1.99182804221418e-05, "loss": 0.6078, "step": 402 }, { "epoch": 0.07, "learning_rate": 1.9917563000409618e-05, "loss": 0.5974, "step": 403 }, { "epoch": 0.07, "learning_rate": 1.9916842456311736e-05, "loss": 0.6103, "step": 404 }, { "epoch": 0.07, "learning_rate": 1.991611879007501e-05, "loss": 0.5955, "step": 405 }, { "epoch": 0.07, "learning_rate": 1.991539200192727e-05, "loss": 0.6116, "step": 406 }, { "epoch": 0.07, "learning_rate": 1.9914662092097334e-05, "loss": 0.6101, "step": 407 }, { "epoch": 0.07, "learning_rate": 1.9913929060815e-05, "loss": 0.6209, "step": 408 }, { "epoch": 0.07, "learning_rate": 1.991319290831105e-05, "loss": 0.5969, "step": 409 }, { "epoch": 0.07, "learning_rate": 1.9912453634817246e-05, "loss": 0.608, "step": 410 }, { "epoch": 0.07, "learning_rate": 1.991171124056634e-05, "loss": 0.6242, "step": 411 }, { "epoch": 0.07, "learning_rate": 1.9910965725792057e-05, "loss": 0.6174, "step": 412 }, { "epoch": 0.07, "learning_rate": 1.991021709072911e-05, "loss": 0.5988, "step": 413 }, { "epoch": 0.07, "learning_rate": 1.9909465335613194e-05, "loss": 0.6049, "step": 414 }, { "epoch": 0.07, "learning_rate": 1.990871046068098e-05, "loss": 0.6252, "step": 415 }, { "epoch": 0.07, "learning_rate": 1.990795246617014e-05, "loss": 0.6116, "step": 416 }, { "epoch": 0.07, "learning_rate": 1.9907191352319302e-05, "loss": 0.6236, "step": 417 }, { "epoch": 0.07, "learning_rate": 1.9906427119368098e-05, "loss": 0.6215, "step": 418 }, { "epoch": 0.07, "learning_rate": 1.9905659767557126e-05, "loss": 0.5936, "step": 419 }, { "epoch": 0.07, "learning_rate": 1.9904889297127973e-05, "loss": 0.6029, "step": 420 }, { "epoch": 0.07, "learning_rate": 1.9904115708323214e-05, "loss": 0.614, "step": 421 }, { "epoch": 0.07, "learning_rate": 1.9903339001386396e-05, "loss": 0.6128, "step": 422 }, { "epoch": 0.07, "learning_rate": 1.9902559176562045e-05, "loss": 0.6009, "step": 423 }, { "epoch": 0.07, "learning_rate": 1.9901776234095686e-05, "loss": 0.5944, "step": 424 }, { "epoch": 0.07, "learning_rate": 1.9900990174233807e-05, "loss": 0.5916, "step": 425 }, { "epoch": 0.07, "learning_rate": 1.9900200997223886e-05, "loss": 0.6182, "step": 426 }, { "epoch": 0.07, "learning_rate": 1.9899408703314383e-05, "loss": 0.61, "step": 427 }, { "epoch": 0.07, "learning_rate": 1.9898613292754738e-05, "loss": 0.6146, "step": 428 }, { "epoch": 0.07, "learning_rate": 1.9897814765795365e-05, "loss": 0.5921, "step": 429 }, { "epoch": 0.07, "learning_rate": 1.9897013122687676e-05, "loss": 0.5978, "step": 430 }, { "epoch": 0.07, "learning_rate": 1.9896208363684048e-05, "loss": 0.6121, "step": 431 }, { "epoch": 0.07, "learning_rate": 1.9895400489037843e-05, "loss": 0.6173, "step": 432 }, { "epoch": 0.08, "learning_rate": 1.989458949900341e-05, "loss": 0.6051, "step": 433 }, { "epoch": 0.08, "learning_rate": 1.989377539383607e-05, "loss": 0.5981, "step": 434 }, { "epoch": 0.08, "learning_rate": 1.9892958173792136e-05, "loss": 0.6088, "step": 435 }, { "epoch": 0.08, "learning_rate": 1.989213783912889e-05, "loss": 0.6112, "step": 436 }, { "epoch": 0.08, "learning_rate": 1.9891314390104597e-05, "loss": 0.5997, "step": 437 }, { "epoch": 0.08, "learning_rate": 1.989048782697851e-05, "loss": 0.6162, "step": 438 }, { "epoch": 0.08, "learning_rate": 1.988965815001086e-05, "loss": 0.606, "step": 439 }, { "epoch": 0.08, "learning_rate": 1.9888825359462846e-05, "loss": 0.597, "step": 440 }, { "epoch": 0.08, "learning_rate": 1.9887989455596667e-05, "loss": 0.5985, "step": 441 }, { "epoch": 0.08, "learning_rate": 1.988715043867549e-05, "loss": 0.5891, "step": 442 }, { "epoch": 0.08, "learning_rate": 1.9886308308963458e-05, "loss": 0.6059, "step": 443 }, { "epoch": 0.08, "learning_rate": 1.9885463066725708e-05, "loss": 0.5968, "step": 444 }, { "epoch": 0.08, "learning_rate": 1.9884614712228346e-05, "loss": 0.614, "step": 445 }, { "epoch": 0.08, "learning_rate": 1.9883763245738457e-05, "loss": 0.6141, "step": 446 }, { "epoch": 0.08, "learning_rate": 1.988290866752412e-05, "loss": 0.6001, "step": 447 }, { "epoch": 0.08, "learning_rate": 1.9882050977854374e-05, "loss": 0.6214, "step": 448 }, { "epoch": 0.08, "learning_rate": 1.9881190176999255e-05, "loss": 0.6109, "step": 449 }, { "epoch": 0.08, "learning_rate": 1.9880326265229764e-05, "loss": 0.6052, "step": 450 }, { "epoch": 0.08, "learning_rate": 1.9879459242817888e-05, "loss": 0.6109, "step": 451 }, { "epoch": 0.08, "learning_rate": 1.98785891100366e-05, "loss": 0.6089, "step": 452 }, { "epoch": 0.08, "learning_rate": 1.9877715867159838e-05, "loss": 0.6117, "step": 453 }, { "epoch": 0.08, "learning_rate": 1.9876839514462532e-05, "loss": 0.6079, "step": 454 }, { "epoch": 0.08, "learning_rate": 1.9875960052220582e-05, "loss": 0.6146, "step": 455 }, { "epoch": 0.08, "learning_rate": 1.9875077480710875e-05, "loss": 0.589, "step": 456 }, { "epoch": 0.08, "learning_rate": 1.987419180021127e-05, "loss": 0.6036, "step": 457 }, { "epoch": 0.08, "learning_rate": 1.9873303011000606e-05, "loss": 0.617, "step": 458 }, { "epoch": 0.08, "learning_rate": 1.9872411113358707e-05, "loss": 0.6215, "step": 459 }, { "epoch": 0.08, "learning_rate": 1.9871516107566366e-05, "loss": 0.6044, "step": 460 }, { "epoch": 0.08, "learning_rate": 1.9870617993905363e-05, "loss": 0.6287, "step": 461 }, { "epoch": 0.08, "learning_rate": 1.986971677265845e-05, "loss": 0.5913, "step": 462 }, { "epoch": 0.08, "learning_rate": 1.9868812444109364e-05, "loss": 0.6002, "step": 463 }, { "epoch": 0.08, "learning_rate": 1.9867905008542812e-05, "loss": 0.6056, "step": 464 }, { "epoch": 0.08, "learning_rate": 1.986699446624449e-05, "loss": 0.614, "step": 465 }, { "epoch": 0.08, "learning_rate": 1.9866080817501057e-05, "loss": 0.6014, "step": 466 }, { "epoch": 0.08, "learning_rate": 1.9865164062600165e-05, "loss": 0.5943, "step": 467 }, { "epoch": 0.08, "learning_rate": 1.9864244201830437e-05, "loss": 0.5931, "step": 468 }, { "epoch": 0.08, "learning_rate": 1.9863321235481472e-05, "loss": 0.6144, "step": 469 }, { "epoch": 0.08, "learning_rate": 1.9862395163843854e-05, "loss": 0.6133, "step": 470 }, { "epoch": 0.08, "learning_rate": 1.986146598720913e-05, "loss": 0.6078, "step": 471 }, { "epoch": 0.08, "learning_rate": 1.986053370586985e-05, "loss": 0.6064, "step": 472 }, { "epoch": 0.08, "learning_rate": 1.9859598320119514e-05, "loss": 0.5851, "step": 473 }, { "epoch": 0.08, "learning_rate": 1.9858659830252616e-05, "loss": 0.6073, "step": 474 }, { "epoch": 0.08, "learning_rate": 1.985771823656462e-05, "loss": 0.6012, "step": 475 }, { "epoch": 0.08, "learning_rate": 1.9856773539351972e-05, "loss": 0.5959, "step": 476 }, { "epoch": 0.08, "learning_rate": 1.9855825738912094e-05, "loss": 0.6199, "step": 477 }, { "epoch": 0.08, "learning_rate": 1.9854874835543376e-05, "loss": 0.6013, "step": 478 }, { "epoch": 0.08, "learning_rate": 1.9853920829545204e-05, "loss": 0.606, "step": 479 }, { "epoch": 0.08, "learning_rate": 1.9852963721217922e-05, "loss": 0.6081, "step": 480 }, { "epoch": 0.08, "learning_rate": 1.9852003510862864e-05, "loss": 0.6038, "step": 481 }, { "epoch": 0.08, "learning_rate": 1.985104019878233e-05, "loss": 0.6033, "step": 482 }, { "epoch": 0.08, "learning_rate": 1.9850073785279598e-05, "loss": 0.6023, "step": 483 }, { "epoch": 0.08, "learning_rate": 1.9849104270658933e-05, "loss": 0.6237, "step": 484 }, { "epoch": 0.08, "learning_rate": 1.9848131655225567e-05, "loss": 0.6063, "step": 485 }, { "epoch": 0.08, "learning_rate": 1.9847155939285713e-05, "loss": 0.5956, "step": 486 }, { "epoch": 0.08, "learning_rate": 1.9846177123146553e-05, "loss": 0.6237, "step": 487 }, { "epoch": 0.08, "learning_rate": 1.9845195207116253e-05, "loss": 0.6095, "step": 488 }, { "epoch": 0.08, "learning_rate": 1.9844210191503944e-05, "loss": 0.6018, "step": 489 }, { "epoch": 0.08, "learning_rate": 1.9843222076619755e-05, "loss": 0.6045, "step": 490 }, { "epoch": 0.09, "learning_rate": 1.984223086277476e-05, "loss": 0.6302, "step": 491 }, { "epoch": 0.09, "learning_rate": 1.9841236550281037e-05, "loss": 0.6164, "step": 492 }, { "epoch": 0.09, "learning_rate": 1.9840239139451622e-05, "loss": 0.6149, "step": 493 }, { "epoch": 0.09, "learning_rate": 1.983923863060053e-05, "loss": 0.6118, "step": 494 }, { "epoch": 0.09, "learning_rate": 1.9838235024042757e-05, "loss": 0.6115, "step": 495 }, { "epoch": 0.09, "learning_rate": 1.983722832009427e-05, "loss": 0.6039, "step": 496 }, { "epoch": 0.09, "learning_rate": 1.983621851907201e-05, "loss": 0.6038, "step": 497 }, { "epoch": 0.09, "learning_rate": 1.9835205621293893e-05, "loss": 0.599, "step": 498 }, { "epoch": 0.09, "learning_rate": 1.9834189627078817e-05, "loss": 0.6251, "step": 499 }, { "epoch": 0.09, "learning_rate": 1.9833170536746645e-05, "loss": 0.5853, "step": 500 }, { "epoch": 0.09, "learning_rate": 1.983214835061822e-05, "loss": 0.6136, "step": 501 }, { "epoch": 0.09, "learning_rate": 1.983112306901536e-05, "loss": 0.5988, "step": 502 }, { "epoch": 0.09, "learning_rate": 1.9830094692260853e-05, "loss": 0.6188, "step": 503 }, { "epoch": 0.09, "learning_rate": 1.982906322067847e-05, "loss": 0.6121, "step": 504 }, { "epoch": 0.09, "learning_rate": 1.9828028654592947e-05, "loss": 0.6202, "step": 505 }, { "epoch": 0.09, "learning_rate": 1.9826990994330003e-05, "loss": 0.6086, "step": 506 }, { "epoch": 0.09, "learning_rate": 1.9825950240216324e-05, "loss": 0.6041, "step": 507 }, { "epoch": 0.09, "learning_rate": 1.9824906392579568e-05, "loss": 0.6091, "step": 508 }, { "epoch": 0.09, "learning_rate": 1.982385945174838e-05, "loss": 0.604, "step": 509 }, { "epoch": 0.09, "learning_rate": 1.9822809418052363e-05, "loss": 0.6076, "step": 510 }, { "epoch": 0.09, "learning_rate": 1.982175629182211e-05, "loss": 0.5983, "step": 511 }, { "epoch": 0.09, "learning_rate": 1.9820700073389165e-05, "loss": 0.5774, "step": 512 }, { "epoch": 0.09, "learning_rate": 1.9819640763086075e-05, "loss": 0.6021, "step": 513 }, { "epoch": 0.09, "learning_rate": 1.9818578361246335e-05, "loss": 0.6193, "step": 514 }, { "epoch": 0.09, "learning_rate": 1.9817512868204425e-05, "loss": 0.6317, "step": 515 }, { "epoch": 0.09, "learning_rate": 1.9816444284295796e-05, "loss": 0.6134, "step": 516 }, { "epoch": 0.09, "learning_rate": 1.9815372609856875e-05, "loss": 0.5902, "step": 517 }, { "epoch": 0.09, "learning_rate": 1.9814297845225057e-05, "loss": 0.6075, "step": 518 }, { "epoch": 0.09, "learning_rate": 1.981321999073871e-05, "loss": 0.6012, "step": 519 }, { "epoch": 0.09, "learning_rate": 1.9812139046737184e-05, "loss": 0.597, "step": 520 }, { "epoch": 0.09, "learning_rate": 1.981105501356079e-05, "loss": 0.6076, "step": 521 }, { "epoch": 0.09, "learning_rate": 1.9809967891550812e-05, "loss": 0.6027, "step": 522 }, { "epoch": 0.09, "learning_rate": 1.980887768104952e-05, "loss": 0.6027, "step": 523 }, { "epoch": 0.09, "learning_rate": 1.980778438240014e-05, "loss": 0.5832, "step": 524 }, { "epoch": 0.09, "learning_rate": 1.980668799594688e-05, "loss": 0.6114, "step": 525 }, { "epoch": 0.09, "learning_rate": 1.980558852203492e-05, "loss": 0.6256, "step": 526 }, { "epoch": 0.09, "learning_rate": 1.98044859610104e-05, "loss": 0.6069, "step": 527 }, { "epoch": 0.09, "learning_rate": 1.9803380313220457e-05, "loss": 0.6081, "step": 528 }, { "epoch": 0.09, "learning_rate": 1.9802271579013172e-05, "loss": 0.6018, "step": 529 }, { "epoch": 0.09, "learning_rate": 1.980115975873761e-05, "loss": 0.5994, "step": 530 }, { "epoch": 0.09, "learning_rate": 1.9800044852743817e-05, "loss": 0.6088, "step": 531 }, { "epoch": 0.09, "learning_rate": 1.9798926861382792e-05, "loss": 0.6008, "step": 532 }, { "epoch": 0.09, "learning_rate": 1.9797805785006516e-05, "loss": 0.5962, "step": 533 }, { "epoch": 0.09, "learning_rate": 1.9796681623967943e-05, "loss": 0.6001, "step": 534 }, { "epoch": 0.09, "learning_rate": 1.9795554378620995e-05, "loss": 0.6219, "step": 535 }, { "epoch": 0.09, "learning_rate": 1.9794424049320557e-05, "loss": 0.6074, "step": 536 }, { "epoch": 0.09, "learning_rate": 1.9793290636422503e-05, "loss": 0.6006, "step": 537 }, { "epoch": 0.09, "learning_rate": 1.9792154140283663e-05, "loss": 0.6061, "step": 538 }, { "epoch": 0.09, "learning_rate": 1.979101456126184e-05, "loss": 0.624, "step": 539 }, { "epoch": 0.09, "learning_rate": 1.9789871899715817e-05, "loss": 0.6041, "step": 540 }, { "epoch": 0.09, "learning_rate": 1.9788726156005332e-05, "loss": 0.5929, "step": 541 }, { "epoch": 0.09, "learning_rate": 1.9787577330491104e-05, "loss": 0.5962, "step": 542 }, { "epoch": 0.09, "learning_rate": 1.9786425423534826e-05, "loss": 0.616, "step": 543 }, { "epoch": 0.09, "learning_rate": 1.9785270435499147e-05, "loss": 0.6177, "step": 544 }, { "epoch": 0.09, "learning_rate": 1.97841123667477e-05, "loss": 0.5935, "step": 545 }, { "epoch": 0.09, "learning_rate": 1.9782951217645078e-05, "loss": 0.5896, "step": 546 }, { "epoch": 0.09, "learning_rate": 1.978178698855685e-05, "loss": 0.5974, "step": 547 }, { "epoch": 0.09, "learning_rate": 1.9780619679849552e-05, "loss": 0.6032, "step": 548 }, { "epoch": 0.1, "learning_rate": 1.9779449291890687e-05, "loss": 0.6117, "step": 549 }, { "epoch": 0.1, "learning_rate": 1.9778275825048737e-05, "loss": 0.598, "step": 550 }, { "epoch": 0.1, "learning_rate": 1.9777099279693147e-05, "loss": 0.6053, "step": 551 }, { "epoch": 0.1, "learning_rate": 1.9775919656194324e-05, "loss": 0.6015, "step": 552 }, { "epoch": 0.1, "learning_rate": 1.9774736954923654e-05, "loss": 0.6051, "step": 553 }, { "epoch": 0.1, "learning_rate": 1.9773551176253488e-05, "loss": 0.6125, "step": 554 }, { "epoch": 0.1, "learning_rate": 1.9772362320557152e-05, "loss": 0.6097, "step": 555 }, { "epoch": 0.1, "learning_rate": 1.9771170388208935e-05, "loss": 0.5941, "step": 556 }, { "epoch": 0.1, "learning_rate": 1.9769975379584087e-05, "loss": 0.5953, "step": 557 }, { "epoch": 0.1, "learning_rate": 1.9768777295058845e-05, "loss": 0.6236, "step": 558 }, { "epoch": 0.1, "learning_rate": 1.97675761350104e-05, "loss": 0.6261, "step": 559 }, { "epoch": 0.1, "learning_rate": 1.9766371899816915e-05, "loss": 0.6027, "step": 560 }, { "epoch": 0.1, "learning_rate": 1.9765164589857522e-05, "loss": 0.5988, "step": 561 }, { "epoch": 0.1, "learning_rate": 1.976395420551232e-05, "loss": 0.5868, "step": 562 }, { "epoch": 0.1, "learning_rate": 1.976274074716238e-05, "loss": 0.5934, "step": 563 }, { "epoch": 0.1, "learning_rate": 1.9761524215189736e-05, "loss": 0.6056, "step": 564 }, { "epoch": 0.1, "learning_rate": 1.976030460997739e-05, "loss": 0.5929, "step": 565 }, { "epoch": 0.1, "learning_rate": 1.975908193190931e-05, "loss": 0.5968, "step": 566 }, { "epoch": 0.1, "learning_rate": 1.975785618137044e-05, "loss": 0.6033, "step": 567 }, { "epoch": 0.1, "learning_rate": 1.9756627358746683e-05, "loss": 0.6032, "step": 568 }, { "epoch": 0.1, "learning_rate": 1.9755395464424913e-05, "loss": 0.5939, "step": 569 }, { "epoch": 0.1, "learning_rate": 1.9754160498792964e-05, "loss": 0.5951, "step": 570 }, { "epoch": 0.1, "learning_rate": 1.975292246223965e-05, "loss": 0.5868, "step": 571 }, { "epoch": 0.1, "learning_rate": 1.975168135515474e-05, "loss": 0.608, "step": 572 }, { "epoch": 0.1, "learning_rate": 1.9750437177928975e-05, "loss": 0.6115, "step": 573 }, { "epoch": 0.1, "learning_rate": 1.9749189930954062e-05, "loss": 0.6017, "step": 574 }, { "epoch": 0.1, "learning_rate": 1.9747939614622673e-05, "loss": 0.6064, "step": 575 }, { "epoch": 0.1, "learning_rate": 1.974668622932845e-05, "loss": 0.6007, "step": 576 }, { "epoch": 0.1, "learning_rate": 1.9745429775465997e-05, "loss": 0.5994, "step": 577 }, { "epoch": 0.1, "learning_rate": 1.974417025343089e-05, "loss": 0.6094, "step": 578 }, { "epoch": 0.1, "learning_rate": 1.9742907663619658e-05, "loss": 0.6046, "step": 579 }, { "epoch": 0.1, "learning_rate": 1.9741642006429813e-05, "loss": 0.6151, "step": 580 }, { "epoch": 0.1, "learning_rate": 1.974037328225982e-05, "loss": 0.592, "step": 581 }, { "epoch": 0.1, "learning_rate": 1.9739101491509114e-05, "loss": 0.6019, "step": 582 }, { "epoch": 0.1, "learning_rate": 1.9737826634578097e-05, "loss": 0.5802, "step": 583 }, { "epoch": 0.1, "learning_rate": 1.9736548711868137e-05, "loss": 0.5994, "step": 584 }, { "epoch": 0.1, "learning_rate": 1.9735267723781556e-05, "loss": 0.6105, "step": 585 }, { "epoch": 0.1, "learning_rate": 1.9733983670721658e-05, "loss": 0.6154, "step": 586 }, { "epoch": 0.1, "learning_rate": 1.97326965530927e-05, "loss": 0.6158, "step": 587 }, { "epoch": 0.1, "learning_rate": 1.9731406371299917e-05, "loss": 0.6222, "step": 588 }, { "epoch": 0.1, "learning_rate": 1.973011312574949e-05, "loss": 0.6039, "step": 589 }, { "epoch": 0.1, "learning_rate": 1.972881681684857e-05, "loss": 0.6224, "step": 590 }, { "epoch": 0.1, "learning_rate": 1.9727517445005286e-05, "loss": 0.6005, "step": 591 }, { "epoch": 0.1, "learning_rate": 1.9726215010628717e-05, "loss": 0.6008, "step": 592 }, { "epoch": 0.1, "learning_rate": 1.9724909514128914e-05, "loss": 0.6044, "step": 593 }, { "epoch": 0.1, "learning_rate": 1.9723600955916887e-05, "loss": 0.5839, "step": 594 }, { "epoch": 0.1, "learning_rate": 1.972228933640461e-05, "loss": 0.5957, "step": 595 }, { "epoch": 0.1, "learning_rate": 1.9720974656005027e-05, "loss": 0.6019, "step": 596 }, { "epoch": 0.1, "learning_rate": 1.9719656915132038e-05, "loss": 0.6063, "step": 597 }, { "epoch": 0.1, "learning_rate": 1.9718336114200515e-05, "loss": 0.602, "step": 598 }, { "epoch": 0.1, "learning_rate": 1.971701225362628e-05, "loss": 0.6031, "step": 599 }, { "epoch": 0.1, "learning_rate": 1.9715685333826134e-05, "loss": 0.598, "step": 600 }, { "epoch": 0.1, "learning_rate": 1.9714355355217828e-05, "loss": 0.6178, "step": 601 }, { "epoch": 0.1, "learning_rate": 1.9713022318220086e-05, "loss": 0.5961, "step": 602 }, { "epoch": 0.1, "learning_rate": 1.971168622325259e-05, "loss": 0.606, "step": 603 }, { "epoch": 0.1, "learning_rate": 1.9710347070735984e-05, "loss": 0.6057, "step": 604 }, { "epoch": 0.1, "learning_rate": 1.9709004861091876e-05, "loss": 0.6, "step": 605 }, { "epoch": 0.1, "learning_rate": 1.9707659594742838e-05, "loss": 0.6136, "step": 606 }, { "epoch": 0.11, "learning_rate": 1.9706311272112402e-05, "loss": 0.5904, "step": 607 }, { "epoch": 0.11, "learning_rate": 1.9704959893625065e-05, "loss": 0.5741, "step": 608 }, { "epoch": 0.11, "learning_rate": 1.9703605459706277e-05, "loss": 0.5984, "step": 609 }, { "epoch": 0.11, "learning_rate": 1.9702247970782466e-05, "loss": 0.6133, "step": 610 }, { "epoch": 0.11, "learning_rate": 1.970088742728101e-05, "loss": 0.5793, "step": 611 }, { "epoch": 0.11, "learning_rate": 1.969952382963025e-05, "loss": 0.5931, "step": 612 }, { "epoch": 0.11, "learning_rate": 1.9698157178259486e-05, "loss": 0.6022, "step": 613 }, { "epoch": 0.11, "learning_rate": 1.9696787473598993e-05, "loss": 0.6211, "step": 614 }, { "epoch": 0.11, "learning_rate": 1.9695414716079993e-05, "loss": 0.597, "step": 615 }, { "epoch": 0.11, "learning_rate": 1.9694038906134673e-05, "loss": 0.6061, "step": 616 }, { "epoch": 0.11, "learning_rate": 1.9692660044196183e-05, "loss": 0.6007, "step": 617 }, { "epoch": 0.11, "learning_rate": 1.9691278130698633e-05, "loss": 0.5892, "step": 618 }, { "epoch": 0.11, "learning_rate": 1.9689893166077093e-05, "loss": 0.6199, "step": 619 }, { "epoch": 0.11, "learning_rate": 1.9688505150767596e-05, "loss": 0.6154, "step": 620 }, { "epoch": 0.11, "learning_rate": 1.968711408520713e-05, "loss": 0.5822, "step": 621 }, { "epoch": 0.11, "learning_rate": 1.9685719969833648e-05, "loss": 0.6064, "step": 622 }, { "epoch": 0.11, "learning_rate": 1.9684322805086064e-05, "loss": 0.6208, "step": 623 }, { "epoch": 0.11, "learning_rate": 1.9682922591404246e-05, "loss": 0.6069, "step": 624 }, { "epoch": 0.11, "learning_rate": 1.968151932922903e-05, "loss": 0.604, "step": 625 }, { "epoch": 0.11, "learning_rate": 1.9680113019002212e-05, "loss": 0.6041, "step": 626 }, { "epoch": 0.11, "learning_rate": 1.9678703661166532e-05, "loss": 0.5954, "step": 627 }, { "epoch": 0.11, "learning_rate": 1.9677291256165712e-05, "loss": 0.601, "step": 628 }, { "epoch": 0.11, "learning_rate": 1.9675875804444416e-05, "loss": 0.5964, "step": 629 }, { "epoch": 0.11, "learning_rate": 1.9674457306448272e-05, "loss": 0.5843, "step": 630 }, { "epoch": 0.11, "learning_rate": 1.967303576262387e-05, "loss": 0.5953, "step": 631 }, { "epoch": 0.11, "learning_rate": 1.9671611173418764e-05, "loss": 0.6149, "step": 632 }, { "epoch": 0.11, "learning_rate": 1.967018353928145e-05, "loss": 0.5814, "step": 633 }, { "epoch": 0.11, "learning_rate": 1.96687528606614e-05, "loss": 0.5949, "step": 634 }, { "epoch": 0.11, "learning_rate": 1.9667319138009036e-05, "loss": 0.6025, "step": 635 }, { "epoch": 0.11, "learning_rate": 1.9665882371775735e-05, "loss": 0.5966, "step": 636 }, { "epoch": 0.11, "learning_rate": 1.9664442562413844e-05, "loss": 0.6114, "step": 637 }, { "epoch": 0.11, "learning_rate": 1.9662999710376655e-05, "loss": 0.6086, "step": 638 }, { "epoch": 0.11, "learning_rate": 1.966155381611843e-05, "loss": 0.5752, "step": 639 }, { "epoch": 0.11, "learning_rate": 1.9660104880094374e-05, "loss": 0.5939, "step": 640 }, { "epoch": 0.11, "learning_rate": 1.965865290276067e-05, "loss": 0.5878, "step": 641 }, { "epoch": 0.11, "learning_rate": 1.9657197884574434e-05, "loss": 0.5962, "step": 642 }, { "epoch": 0.11, "learning_rate": 1.965573982599376e-05, "loss": 0.5814, "step": 643 }, { "epoch": 0.11, "learning_rate": 1.9654278727477695e-05, "loss": 0.6065, "step": 644 }, { "epoch": 0.11, "learning_rate": 1.9652814589486228e-05, "loss": 0.6076, "step": 645 }, { "epoch": 0.11, "learning_rate": 1.9651347412480323e-05, "loss": 0.5884, "step": 646 }, { "epoch": 0.11, "learning_rate": 1.96498771969219e-05, "loss": 0.5891, "step": 647 }, { "epoch": 0.11, "learning_rate": 1.9648403943273818e-05, "loss": 0.5963, "step": 648 }, { "epoch": 0.11, "learning_rate": 1.9646927651999915e-05, "loss": 0.5981, "step": 649 }, { "epoch": 0.11, "learning_rate": 1.9645448323564968e-05, "loss": 0.6053, "step": 650 }, { "epoch": 0.11, "learning_rate": 1.964396595843472e-05, "loss": 0.5978, "step": 651 }, { "epoch": 0.11, "learning_rate": 1.9642480557075862e-05, "loss": 0.6106, "step": 652 }, { "epoch": 0.11, "learning_rate": 1.964099211995605e-05, "loss": 0.6073, "step": 653 }, { "epoch": 0.11, "learning_rate": 1.9639500647543892e-05, "loss": 0.6097, "step": 654 }, { "epoch": 0.11, "learning_rate": 1.963800614030895e-05, "loss": 0.6106, "step": 655 }, { "epoch": 0.11, "learning_rate": 1.9636508598721745e-05, "loss": 0.5994, "step": 656 }, { "epoch": 0.11, "learning_rate": 1.963500802325375e-05, "loss": 0.5838, "step": 657 }, { "epoch": 0.11, "learning_rate": 1.963350441437739e-05, "loss": 0.6112, "step": 658 }, { "epoch": 0.11, "learning_rate": 1.9631997772566052e-05, "loss": 0.6065, "step": 659 }, { "epoch": 0.11, "learning_rate": 1.9630488098294072e-05, "loss": 0.6039, "step": 660 }, { "epoch": 0.11, "learning_rate": 1.9628975392036755e-05, "loss": 0.5707, "step": 661 }, { "epoch": 0.11, "learning_rate": 1.9627459654270335e-05, "loss": 0.6088, "step": 662 }, { "epoch": 0.11, "learning_rate": 1.9625940885472018e-05, "loss": 0.6213, "step": 663 }, { "epoch": 0.12, "learning_rate": 1.962441908611997e-05, "loss": 0.6106, "step": 664 }, { "epoch": 0.12, "learning_rate": 1.9622894256693292e-05, "loss": 0.6085, "step": 665 }, { "epoch": 0.12, "learning_rate": 1.9621366397672054e-05, "loss": 0.611, "step": 666 }, { "epoch": 0.12, "learning_rate": 1.9619835509537274e-05, "loss": 0.5842, "step": 667 }, { "epoch": 0.12, "learning_rate": 1.961830159277092e-05, "loss": 0.6073, "step": 668 }, { "epoch": 0.12, "learning_rate": 1.9616764647855924e-05, "loss": 0.6061, "step": 669 }, { "epoch": 0.12, "learning_rate": 1.9615224675276167e-05, "loss": 0.6077, "step": 670 }, { "epoch": 0.12, "learning_rate": 1.9613681675516473e-05, "loss": 0.606, "step": 671 }, { "epoch": 0.12, "learning_rate": 1.9612135649062635e-05, "loss": 0.5944, "step": 672 }, { "epoch": 0.12, "learning_rate": 1.961058659640139e-05, "loss": 0.5928, "step": 673 }, { "epoch": 0.12, "learning_rate": 1.9609034518020425e-05, "loss": 0.5971, "step": 674 }, { "epoch": 0.12, "learning_rate": 1.9607479414408387e-05, "loss": 0.5979, "step": 675 }, { "epoch": 0.12, "learning_rate": 1.9605921286054877e-05, "loss": 0.6144, "step": 676 }, { "epoch": 0.12, "learning_rate": 1.9604360133450436e-05, "loss": 0.5944, "step": 677 }, { "epoch": 0.12, "learning_rate": 1.9602795957086567e-05, "loss": 0.6033, "step": 678 }, { "epoch": 0.12, "learning_rate": 1.9601228757455725e-05, "loss": 0.6037, "step": 679 }, { "epoch": 0.12, "learning_rate": 1.9599658535051312e-05, "loss": 0.5999, "step": 680 }, { "epoch": 0.12, "learning_rate": 1.959808529036769e-05, "loss": 0.6144, "step": 681 }, { "epoch": 0.12, "learning_rate": 1.9596509023900155e-05, "loss": 0.5928, "step": 682 }, { "epoch": 0.12, "learning_rate": 1.9594929736144978e-05, "loss": 0.6044, "step": 683 }, { "epoch": 0.12, "learning_rate": 1.959334742759936e-05, "loss": 0.5948, "step": 684 }, { "epoch": 0.12, "learning_rate": 1.9591762098761467e-05, "loss": 0.604, "step": 685 }, { "epoch": 0.12, "learning_rate": 1.9590173750130413e-05, "loss": 0.5983, "step": 686 }, { "epoch": 0.12, "learning_rate": 1.9588582382206258e-05, "loss": 0.6169, "step": 687 }, { "epoch": 0.12, "learning_rate": 1.9586987995490015e-05, "loss": 0.6039, "step": 688 }, { "epoch": 0.12, "learning_rate": 1.958539059048365e-05, "loss": 0.6156, "step": 689 }, { "epoch": 0.12, "learning_rate": 1.9583790167690074e-05, "loss": 0.6061, "step": 690 }, { "epoch": 0.12, "learning_rate": 1.9582186727613152e-05, "loss": 0.5966, "step": 691 }, { "epoch": 0.12, "learning_rate": 1.9580580270757702e-05, "loss": 0.5805, "step": 692 }, { "epoch": 0.12, "learning_rate": 1.9578970797629488e-05, "loss": 0.6046, "step": 693 }, { "epoch": 0.12, "learning_rate": 1.9577358308735217e-05, "loss": 0.6113, "step": 694 }, { "epoch": 0.12, "learning_rate": 1.9575742804582557e-05, "loss": 0.6156, "step": 695 }, { "epoch": 0.12, "learning_rate": 1.957412428568012e-05, "loss": 0.5987, "step": 696 }, { "epoch": 0.12, "learning_rate": 1.9572502752537462e-05, "loss": 0.5913, "step": 697 }, { "epoch": 0.12, "learning_rate": 1.9570878205665104e-05, "loss": 0.5947, "step": 698 }, { "epoch": 0.12, "learning_rate": 1.9569250645574498e-05, "loss": 0.6079, "step": 699 }, { "epoch": 0.12, "learning_rate": 1.956762007277805e-05, "loss": 0.595, "step": 700 }, { "epoch": 0.12, "learning_rate": 1.9565986487789124e-05, "loss": 0.6072, "step": 701 }, { "epoch": 0.12, "learning_rate": 1.9564349891122017e-05, "loss": 0.6025, "step": 702 }, { "epoch": 0.12, "learning_rate": 1.9562710283291988e-05, "loss": 0.6233, "step": 703 }, { "epoch": 0.12, "learning_rate": 1.9561067664815233e-05, "loss": 0.5939, "step": 704 }, { "epoch": 0.12, "learning_rate": 1.9559422036208905e-05, "loss": 0.6028, "step": 705 }, { "epoch": 0.12, "learning_rate": 1.95577733979911e-05, "loss": 0.6115, "step": 706 }, { "epoch": 0.12, "learning_rate": 1.9556121750680857e-05, "loss": 0.6081, "step": 707 }, { "epoch": 0.12, "learning_rate": 1.9554467094798173e-05, "loss": 0.5941, "step": 708 }, { "epoch": 0.12, "learning_rate": 1.9552809430863985e-05, "loss": 0.6062, "step": 709 }, { "epoch": 0.12, "learning_rate": 1.9551148759400177e-05, "loss": 0.6072, "step": 710 }, { "epoch": 0.12, "learning_rate": 1.954948508092958e-05, "loss": 0.6063, "step": 711 }, { "epoch": 0.12, "learning_rate": 1.954781839597598e-05, "loss": 0.5871, "step": 712 }, { "epoch": 0.12, "learning_rate": 1.9546148705064097e-05, "loss": 0.6166, "step": 713 }, { "epoch": 0.12, "learning_rate": 1.95444760087196e-05, "loss": 0.6027, "step": 714 }, { "epoch": 0.12, "learning_rate": 1.9542800307469116e-05, "loss": 0.6054, "step": 715 }, { "epoch": 0.12, "learning_rate": 1.95411216018402e-05, "loss": 0.5932, "step": 716 }, { "epoch": 0.12, "learning_rate": 1.9539439892361375e-05, "loss": 0.6028, "step": 717 }, { "epoch": 0.12, "learning_rate": 1.9537755179562086e-05, "loss": 0.5864, "step": 718 }, { "epoch": 0.12, "learning_rate": 1.9536067463972733e-05, "loss": 0.6003, "step": 719 }, { "epoch": 0.12, "learning_rate": 1.953437674612467e-05, "loss": 0.6181, "step": 720 }, { "epoch": 0.12, "learning_rate": 1.9532683026550187e-05, "loss": 0.5947, "step": 721 }, { "epoch": 0.13, "learning_rate": 1.9530986305782517e-05, "loss": 0.5942, "step": 722 }, { "epoch": 0.13, "learning_rate": 1.9529286584355848e-05, "loss": 0.5894, "step": 723 }, { "epoch": 0.13, "learning_rate": 1.9527583862805303e-05, "loss": 0.5979, "step": 724 }, { "epoch": 0.13, "learning_rate": 1.9525878141666954e-05, "loss": 0.6152, "step": 725 }, { "epoch": 0.13, "learning_rate": 1.9524169421477818e-05, "loss": 0.5895, "step": 726 }, { "epoch": 0.13, "learning_rate": 1.952245770277585e-05, "loss": 0.6003, "step": 727 }, { "epoch": 0.13, "learning_rate": 1.9520742986099958e-05, "loss": 0.5944, "step": 728 }, { "epoch": 0.13, "learning_rate": 1.9519025271989986e-05, "loss": 0.6014, "step": 729 }, { "epoch": 0.13, "learning_rate": 1.951730456098673e-05, "loss": 0.6043, "step": 730 }, { "epoch": 0.13, "learning_rate": 1.9515580853631922e-05, "loss": 0.5906, "step": 731 }, { "epoch": 0.13, "learning_rate": 1.9513854150468238e-05, "loss": 0.5985, "step": 732 }, { "epoch": 0.13, "learning_rate": 1.9512124452039307e-05, "loss": 0.6042, "step": 733 }, { "epoch": 0.13, "learning_rate": 1.9510391758889683e-05, "loss": 0.5941, "step": 734 }, { "epoch": 0.13, "learning_rate": 1.9508656071564883e-05, "loss": 0.5808, "step": 735 }, { "epoch": 0.13, "learning_rate": 1.950691739061135e-05, "loss": 0.5914, "step": 736 }, { "epoch": 0.13, "learning_rate": 1.9505175716576476e-05, "loss": 0.5931, "step": 737 }, { "epoch": 0.13, "learning_rate": 1.95034310500086e-05, "loss": 0.6063, "step": 738 }, { "epoch": 0.13, "learning_rate": 1.9501683391457e-05, "loss": 0.5896, "step": 739 }, { "epoch": 0.13, "learning_rate": 1.9499932741471887e-05, "loss": 0.6029, "step": 740 }, { "epoch": 0.13, "learning_rate": 1.949817910060443e-05, "loss": 0.5809, "step": 741 }, { "epoch": 0.13, "learning_rate": 1.9496422469406725e-05, "loss": 0.5898, "step": 742 }, { "epoch": 0.13, "learning_rate": 1.9494662848431816e-05, "loss": 0.6037, "step": 743 }, { "epoch": 0.13, "learning_rate": 1.9492900238233696e-05, "loss": 0.5974, "step": 744 }, { "epoch": 0.13, "learning_rate": 1.949113463936728e-05, "loss": 0.6003, "step": 745 }, { "epoch": 0.13, "learning_rate": 1.9489366052388443e-05, "loss": 0.602, "step": 746 }, { "epoch": 0.13, "learning_rate": 1.9487594477853986e-05, "loss": 0.5768, "step": 747 }, { "epoch": 0.13, "learning_rate": 1.9485819916321663e-05, "loss": 0.5965, "step": 748 }, { "epoch": 0.13, "learning_rate": 1.9484042368350163e-05, "loss": 0.6125, "step": 749 }, { "epoch": 0.13, "learning_rate": 1.9482261834499107e-05, "loss": 0.5988, "step": 750 }, { "epoch": 0.13, "learning_rate": 1.9480478315329073e-05, "loss": 0.5981, "step": 751 }, { "epoch": 0.13, "learning_rate": 1.9478691811401563e-05, "loss": 0.6013, "step": 752 }, { "epoch": 0.13, "learning_rate": 1.9476902323279034e-05, "loss": 0.5986, "step": 753 }, { "epoch": 0.13, "learning_rate": 1.9475109851524862e-05, "loss": 0.6134, "step": 754 }, { "epoch": 0.13, "learning_rate": 1.9473314396703385e-05, "loss": 0.6084, "step": 755 }, { "epoch": 0.13, "learning_rate": 1.9471515959379866e-05, "loss": 0.5949, "step": 756 }, { "epoch": 0.13, "learning_rate": 1.946971454012051e-05, "loss": 0.6018, "step": 757 }, { "epoch": 0.13, "learning_rate": 1.946791013949246e-05, "loss": 0.5911, "step": 758 }, { "epoch": 0.13, "learning_rate": 1.94661027580638e-05, "loss": 0.5884, "step": 759 }, { "epoch": 0.13, "learning_rate": 1.9464292396403553e-05, "loss": 0.6006, "step": 760 }, { "epoch": 0.13, "learning_rate": 1.9462479055081676e-05, "loss": 0.6117, "step": 761 }, { "epoch": 0.13, "learning_rate": 1.946066273466907e-05, "loss": 0.5862, "step": 762 }, { "epoch": 0.13, "learning_rate": 1.945884343573757e-05, "loss": 0.6047, "step": 763 }, { "epoch": 0.13, "learning_rate": 1.9457021158859946e-05, "loss": 0.6001, "step": 764 }, { "epoch": 0.13, "learning_rate": 1.9455195904609913e-05, "loss": 0.6115, "step": 765 }, { "epoch": 0.13, "learning_rate": 1.9453367673562116e-05, "loss": 0.5909, "step": 766 }, { "epoch": 0.13, "learning_rate": 1.9451536466292142e-05, "loss": 0.5987, "step": 767 }, { "epoch": 0.13, "learning_rate": 1.9449702283376516e-05, "loss": 0.5966, "step": 768 }, { "epoch": 0.13, "learning_rate": 1.9447865125392696e-05, "loss": 0.5931, "step": 769 }, { "epoch": 0.13, "learning_rate": 1.9446024992919077e-05, "loss": 0.5889, "step": 770 }, { "epoch": 0.13, "learning_rate": 1.944418188653499e-05, "loss": 0.5979, "step": 771 }, { "epoch": 0.13, "learning_rate": 1.9442335806820706e-05, "loss": 0.6004, "step": 772 }, { "epoch": 0.13, "learning_rate": 1.944048675435743e-05, "loss": 0.596, "step": 773 }, { "epoch": 0.13, "learning_rate": 1.9438634729727304e-05, "loss": 0.5944, "step": 774 }, { "epoch": 0.13, "learning_rate": 1.9436779733513398e-05, "loss": 0.5853, "step": 775 }, { "epoch": 0.13, "learning_rate": 1.9434921766299733e-05, "loss": 0.6009, "step": 776 }, { "epoch": 0.13, "learning_rate": 1.943306082867125e-05, "loss": 0.593, "step": 777 }, { "epoch": 0.13, "learning_rate": 1.9431196921213837e-05, "loss": 0.5969, "step": 778 }, { "epoch": 0.13, "learning_rate": 1.9429330044514305e-05, "loss": 0.593, "step": 779 }, { "epoch": 0.14, "learning_rate": 1.942746019916041e-05, "loss": 0.5944, "step": 780 }, { "epoch": 0.14, "learning_rate": 1.9425587385740844e-05, "loss": 0.5961, "step": 781 }, { "epoch": 0.14, "learning_rate": 1.942371160484522e-05, "loss": 0.6013, "step": 782 }, { "epoch": 0.14, "learning_rate": 1.9421832857064093e-05, "loss": 0.5873, "step": 783 }, { "epoch": 0.14, "learning_rate": 1.9419951142988963e-05, "loss": 0.583, "step": 784 }, { "epoch": 0.14, "learning_rate": 1.9418066463212246e-05, "loss": 0.5989, "step": 785 }, { "epoch": 0.14, "learning_rate": 1.94161788183273e-05, "loss": 0.615, "step": 786 }, { "epoch": 0.14, "learning_rate": 1.941428820892842e-05, "loss": 0.6065, "step": 787 }, { "epoch": 0.14, "learning_rate": 1.9412394635610824e-05, "loss": 0.5945, "step": 788 }, { "epoch": 0.14, "learning_rate": 1.9410498098970675e-05, "loss": 0.612, "step": 789 }, { "epoch": 0.14, "learning_rate": 1.9408598599605062e-05, "loss": 0.5941, "step": 790 }, { "epoch": 0.14, "learning_rate": 1.9406696138112006e-05, "loss": 0.6054, "step": 791 }, { "epoch": 0.14, "learning_rate": 1.9404790715090463e-05, "loss": 0.5965, "step": 792 }, { "epoch": 0.14, "learning_rate": 1.9402882331140322e-05, "loss": 0.5907, "step": 793 }, { "epoch": 0.14, "learning_rate": 1.9400970986862404e-05, "loss": 0.5961, "step": 794 }, { "epoch": 0.14, "learning_rate": 1.9399056682858458e-05, "loss": 0.5837, "step": 795 }, { "epoch": 0.14, "learning_rate": 1.9397139419731176e-05, "loss": 0.5916, "step": 796 }, { "epoch": 0.14, "learning_rate": 1.9395219198084163e-05, "loss": 0.6074, "step": 797 }, { "epoch": 0.14, "learning_rate": 1.9393296018521973e-05, "loss": 0.5995, "step": 798 }, { "epoch": 0.14, "learning_rate": 1.9391369881650083e-05, "loss": 0.5906, "step": 799 }, { "epoch": 0.14, "learning_rate": 1.93894407880749e-05, "loss": 0.5869, "step": 800 }, { "epoch": 0.14, "learning_rate": 1.938750873840377e-05, "loss": 0.5948, "step": 801 }, { "epoch": 0.14, "learning_rate": 1.9385573733244956e-05, "loss": 0.5944, "step": 802 }, { "epoch": 0.14, "learning_rate": 1.9383635773207664e-05, "loss": 0.5992, "step": 803 }, { "epoch": 0.14, "learning_rate": 1.938169485890203e-05, "loss": 0.6082, "step": 804 }, { "epoch": 0.14, "learning_rate": 1.937975099093911e-05, "loss": 0.5821, "step": 805 }, { "epoch": 0.14, "learning_rate": 1.9377804169930893e-05, "loss": 0.5883, "step": 806 }, { "epoch": 0.14, "learning_rate": 1.9375854396490305e-05, "loss": 0.5923, "step": 807 }, { "epoch": 0.14, "learning_rate": 1.9373901671231202e-05, "loss": 0.5881, "step": 808 }, { "epoch": 0.14, "learning_rate": 1.9371945994768354e-05, "loss": 0.5883, "step": 809 }, { "epoch": 0.14, "learning_rate": 1.9369987367717477e-05, "loss": 0.6326, "step": 810 }, { "epoch": 0.14, "learning_rate": 1.9368025790695207e-05, "loss": 0.582, "step": 811 }, { "epoch": 0.14, "learning_rate": 1.9366061264319112e-05, "loss": 0.5774, "step": 812 }, { "epoch": 0.14, "learning_rate": 1.9364093789207686e-05, "loss": 0.5693, "step": 813 }, { "epoch": 0.14, "learning_rate": 1.9362123365980356e-05, "loss": 0.6021, "step": 814 }, { "epoch": 0.14, "learning_rate": 1.9360149995257474e-05, "loss": 0.5958, "step": 815 }, { "epoch": 0.14, "learning_rate": 1.9358173677660317e-05, "loss": 0.5991, "step": 816 }, { "epoch": 0.14, "learning_rate": 1.9356194413811092e-05, "loss": 0.5968, "step": 817 }, { "epoch": 0.14, "learning_rate": 1.935421220433294e-05, "loss": 0.5954, "step": 818 }, { "epoch": 0.14, "learning_rate": 1.9352227049849923e-05, "loss": 0.5957, "step": 819 }, { "epoch": 0.14, "learning_rate": 1.9350238950987024e-05, "loss": 0.5847, "step": 820 }, { "epoch": 0.14, "learning_rate": 1.934824790837017e-05, "loss": 0.5918, "step": 821 }, { "epoch": 0.14, "learning_rate": 1.93462539226262e-05, "loss": 0.6008, "step": 822 }, { "epoch": 0.14, "learning_rate": 1.934425699438288e-05, "loss": 0.62, "step": 823 }, { "epoch": 0.14, "learning_rate": 1.9342257124268914e-05, "loss": 0.5972, "step": 824 }, { "epoch": 0.14, "learning_rate": 1.9340254312913923e-05, "loss": 0.6149, "step": 825 }, { "epoch": 0.14, "learning_rate": 1.9338248560948452e-05, "loss": 0.5851, "step": 826 }, { "epoch": 0.14, "learning_rate": 1.9336239869003982e-05, "loss": 0.5992, "step": 827 }, { "epoch": 0.14, "learning_rate": 1.933422823771291e-05, "loss": 0.5932, "step": 828 }, { "epoch": 0.14, "learning_rate": 1.933221366770856e-05, "loss": 0.5915, "step": 829 }, { "epoch": 0.14, "learning_rate": 1.9330196159625188e-05, "loss": 0.5859, "step": 830 }, { "epoch": 0.14, "learning_rate": 1.9328175714097964e-05, "loss": 0.5961, "step": 831 }, { "epoch": 0.14, "learning_rate": 1.9326152331762995e-05, "loss": 0.6179, "step": 832 }, { "epoch": 0.14, "learning_rate": 1.9324126013257304e-05, "loss": 0.5817, "step": 833 }, { "epoch": 0.14, "learning_rate": 1.9322096759218838e-05, "loss": 0.598, "step": 834 }, { "epoch": 0.14, "learning_rate": 1.9320064570286473e-05, "loss": 0.5752, "step": 835 }, { "epoch": 0.14, "learning_rate": 1.931802944710001e-05, "loss": 0.6198, "step": 836 }, { "epoch": 0.14, "learning_rate": 1.9315991390300166e-05, "loss": 0.5905, "step": 837 }, { "epoch": 0.15, "learning_rate": 1.931395040052859e-05, "loss": 0.6004, "step": 838 }, { "epoch": 0.15, "learning_rate": 1.9311906478427848e-05, "loss": 0.6136, "step": 839 }, { "epoch": 0.15, "learning_rate": 1.9309859624641432e-05, "loss": 0.6045, "step": 840 }, { "epoch": 0.15, "learning_rate": 1.930780983981376e-05, "loss": 0.5853, "step": 841 }, { "epoch": 0.15, "learning_rate": 1.9305757124590164e-05, "loss": 0.5906, "step": 842 }, { "epoch": 0.15, "learning_rate": 1.9303701479616915e-05, "loss": 0.602, "step": 843 }, { "epoch": 0.15, "learning_rate": 1.9301642905541186e-05, "loss": 0.5967, "step": 844 }, { "epoch": 0.15, "learning_rate": 1.9299581403011082e-05, "loss": 0.5936, "step": 845 }, { "epoch": 0.15, "learning_rate": 1.9297516972675638e-05, "loss": 0.5991, "step": 846 }, { "epoch": 0.15, "learning_rate": 1.9295449615184792e-05, "loss": 0.594, "step": 847 }, { "epoch": 0.15, "learning_rate": 1.9293379331189425e-05, "loss": 0.6054, "step": 848 }, { "epoch": 0.15, "learning_rate": 1.9291306121341318e-05, "loss": 0.5876, "step": 849 }, { "epoch": 0.15, "learning_rate": 1.928922998629319e-05, "loss": 0.6127, "step": 850 }, { "epoch": 0.15, "learning_rate": 1.9287150926698673e-05, "loss": 0.6114, "step": 851 }, { "epoch": 0.15, "learning_rate": 1.9285068943212327e-05, "loss": 0.5954, "step": 852 }, { "epoch": 0.15, "learning_rate": 1.9282984036489618e-05, "loss": 0.5861, "step": 853 }, { "epoch": 0.15, "learning_rate": 1.928089620718694e-05, "loss": 0.5905, "step": 854 }, { "epoch": 0.15, "learning_rate": 1.9278805455961623e-05, "loss": 0.5867, "step": 855 }, { "epoch": 0.15, "learning_rate": 1.9276711783471888e-05, "loss": 0.5971, "step": 856 }, { "epoch": 0.15, "learning_rate": 1.9274615190376894e-05, "loss": 0.5926, "step": 857 }, { "epoch": 0.15, "learning_rate": 1.927251567733672e-05, "loss": 0.5944, "step": 858 }, { "epoch": 0.15, "learning_rate": 1.927041324501235e-05, "loss": 0.5878, "step": 859 }, { "epoch": 0.15, "learning_rate": 1.9268307894065704e-05, "loss": 0.579, "step": 860 }, { "epoch": 0.15, "learning_rate": 1.9266199625159615e-05, "loss": 0.5877, "step": 861 }, { "epoch": 0.15, "learning_rate": 1.926408843895783e-05, "loss": 0.5884, "step": 862 }, { "epoch": 0.15, "learning_rate": 1.926197433612502e-05, "loss": 0.6041, "step": 863 }, { "epoch": 0.15, "learning_rate": 1.925985731732677e-05, "loss": 0.6081, "step": 864 }, { "epoch": 0.15, "learning_rate": 1.9257737383229586e-05, "loss": 0.5952, "step": 865 }, { "epoch": 0.15, "learning_rate": 1.925561453450089e-05, "loss": 0.6192, "step": 866 }, { "epoch": 0.15, "learning_rate": 1.9253488771809024e-05, "loss": 0.5897, "step": 867 }, { "epoch": 0.15, "learning_rate": 1.9251360095823246e-05, "loss": 0.5987, "step": 868 }, { "epoch": 0.15, "learning_rate": 1.9249228507213732e-05, "loss": 0.5888, "step": 869 }, { "epoch": 0.15, "learning_rate": 1.924709400665157e-05, "loss": 0.5992, "step": 870 }, { "epoch": 0.15, "learning_rate": 1.9244956594808778e-05, "loss": 0.5834, "step": 871 }, { "epoch": 0.15, "learning_rate": 1.9242816272358272e-05, "loss": 0.5788, "step": 872 }, { "epoch": 0.15, "learning_rate": 1.9240673039973897e-05, "loss": 0.5912, "step": 873 }, { "epoch": 0.15, "learning_rate": 1.923852689833041e-05, "loss": 0.5885, "step": 874 }, { "epoch": 0.15, "learning_rate": 1.923637784810349e-05, "loss": 0.5899, "step": 875 }, { "epoch": 0.15, "learning_rate": 1.9234225889969723e-05, "loss": 0.594, "step": 876 }, { "epoch": 0.15, "learning_rate": 1.923207102460661e-05, "loss": 0.6094, "step": 877 }, { "epoch": 0.15, "learning_rate": 1.922991325269258e-05, "loss": 0.5966, "step": 878 }, { "epoch": 0.15, "learning_rate": 1.9227752574906965e-05, "loss": 0.597, "step": 879 }, { "epoch": 0.15, "learning_rate": 1.922558899193001e-05, "loss": 0.5926, "step": 880 }, { "epoch": 0.15, "learning_rate": 1.922342250444289e-05, "loss": 0.6119, "step": 881 }, { "epoch": 0.15, "learning_rate": 1.9221253113127677e-05, "loss": 0.5949, "step": 882 }, { "epoch": 0.15, "learning_rate": 1.9219080818667365e-05, "loss": 0.5953, "step": 883 }, { "epoch": 0.15, "learning_rate": 1.9216905621745866e-05, "loss": 0.5846, "step": 884 }, { "epoch": 0.15, "learning_rate": 1.9214727523048e-05, "loss": 0.5842, "step": 885 }, { "epoch": 0.15, "learning_rate": 1.9212546523259498e-05, "loss": 0.5983, "step": 886 }, { "epoch": 0.15, "learning_rate": 1.9210362623067015e-05, "loss": 0.5915, "step": 887 }, { "epoch": 0.15, "learning_rate": 1.920817582315811e-05, "loss": 0.6004, "step": 888 }, { "epoch": 0.15, "learning_rate": 1.920598612422125e-05, "loss": 0.5862, "step": 889 }, { "epoch": 0.15, "learning_rate": 1.9203793526945835e-05, "loss": 0.5994, "step": 890 }, { "epoch": 0.15, "learning_rate": 1.9201598032022156e-05, "loss": 0.5946, "step": 891 }, { "epoch": 0.15, "learning_rate": 1.9199399640141428e-05, "loss": 0.5859, "step": 892 }, { "epoch": 0.15, "learning_rate": 1.9197198351995774e-05, "loss": 0.5928, "step": 893 }, { "epoch": 0.15, "learning_rate": 1.9194994168278232e-05, "loss": 0.5824, "step": 894 }, { "epoch": 0.16, "learning_rate": 1.919278708968275e-05, "loss": 0.6057, "step": 895 }, { "epoch": 0.16, "learning_rate": 1.919057711690418e-05, "loss": 0.595, "step": 896 }, { "epoch": 0.16, "learning_rate": 1.9188364250638295e-05, "loss": 0.6064, "step": 897 }, { "epoch": 0.16, "learning_rate": 1.9186148491581782e-05, "loss": 0.5991, "step": 898 }, { "epoch": 0.16, "learning_rate": 1.9183929840432227e-05, "loss": 0.5964, "step": 899 }, { "epoch": 0.16, "learning_rate": 1.9181708297888133e-05, "loss": 0.6064, "step": 900 }, { "epoch": 0.16, "learning_rate": 1.9179483864648913e-05, "loss": 0.5806, "step": 901 }, { "epoch": 0.16, "learning_rate": 1.9177256541414894e-05, "loss": 0.591, "step": 902 }, { "epoch": 0.16, "learning_rate": 1.91750263288873e-05, "loss": 0.5949, "step": 903 }, { "epoch": 0.16, "learning_rate": 1.917279322776828e-05, "loss": 0.589, "step": 904 }, { "epoch": 0.16, "learning_rate": 1.917055723876088e-05, "loss": 0.5813, "step": 905 }, { "epoch": 0.16, "learning_rate": 1.916831836256907e-05, "loss": 0.6072, "step": 906 }, { "epoch": 0.16, "learning_rate": 1.9166076599897706e-05, "loss": 0.6004, "step": 907 }, { "epoch": 0.16, "learning_rate": 1.916383195145258e-05, "loss": 0.5779, "step": 908 }, { "epoch": 0.16, "learning_rate": 1.916158441794037e-05, "loss": 0.5975, "step": 909 }, { "epoch": 0.16, "learning_rate": 1.9159334000068675e-05, "loss": 0.5907, "step": 910 }, { "epoch": 0.16, "learning_rate": 1.9157080698546e-05, "loss": 0.5965, "step": 911 }, { "epoch": 0.16, "learning_rate": 1.9154824514081752e-05, "loss": 0.5965, "step": 912 }, { "epoch": 0.16, "learning_rate": 1.9152565447386256e-05, "loss": 0.6015, "step": 913 }, { "epoch": 0.16, "learning_rate": 1.915030349917073e-05, "loss": 0.5978, "step": 914 }, { "epoch": 0.16, "learning_rate": 1.9148038670147315e-05, "loss": 0.5996, "step": 915 }, { "epoch": 0.16, "learning_rate": 1.914577096102905e-05, "loss": 0.5669, "step": 916 }, { "epoch": 0.16, "learning_rate": 1.9143500372529878e-05, "loss": 0.5986, "step": 917 }, { "epoch": 0.16, "learning_rate": 1.9141226905364657e-05, "loss": 0.5883, "step": 918 }, { "epoch": 0.16, "learning_rate": 1.9138950560249147e-05, "loss": 0.5883, "step": 919 }, { "epoch": 0.16, "learning_rate": 1.9136671337900013e-05, "loss": 0.5792, "step": 920 }, { "epoch": 0.16, "learning_rate": 1.9134389239034826e-05, "loss": 0.6009, "step": 921 }, { "epoch": 0.16, "learning_rate": 1.9132104264372065e-05, "loss": 0.6017, "step": 922 }, { "epoch": 0.16, "learning_rate": 1.9129816414631116e-05, "loss": 0.5919, "step": 923 }, { "epoch": 0.16, "learning_rate": 1.9127525690532258e-05, "loss": 0.6065, "step": 924 }, { "epoch": 0.16, "learning_rate": 1.9125232092796697e-05, "loss": 0.6088, "step": 925 }, { "epoch": 0.16, "learning_rate": 1.9122935622146518e-05, "loss": 0.5938, "step": 926 }, { "epoch": 0.16, "learning_rate": 1.9120636279304733e-05, "loss": 0.6032, "step": 927 }, { "epoch": 0.16, "learning_rate": 1.911833406499524e-05, "loss": 0.5952, "step": 928 }, { "epoch": 0.16, "learning_rate": 1.911602897994286e-05, "loss": 0.5894, "step": 929 }, { "epoch": 0.16, "learning_rate": 1.91137210248733e-05, "loss": 0.5982, "step": 930 }, { "epoch": 0.16, "learning_rate": 1.9111410200513182e-05, "loss": 0.5887, "step": 931 }, { "epoch": 0.16, "learning_rate": 1.9109096507590022e-05, "loss": 0.5924, "step": 932 }, { "epoch": 0.16, "learning_rate": 1.910677994683225e-05, "loss": 0.5998, "step": 933 }, { "epoch": 0.16, "learning_rate": 1.910446051896919e-05, "loss": 0.5872, "step": 934 }, { "epoch": 0.16, "learning_rate": 1.9102138224731073e-05, "loss": 0.5933, "step": 935 }, { "epoch": 0.16, "learning_rate": 1.9099813064849034e-05, "loss": 0.5966, "step": 936 }, { "epoch": 0.16, "learning_rate": 1.90974850400551e-05, "loss": 0.5976, "step": 937 }, { "epoch": 0.16, "learning_rate": 1.9095154151082218e-05, "loss": 0.5833, "step": 938 }, { "epoch": 0.16, "learning_rate": 1.909282039866422e-05, "loss": 0.6186, "step": 939 }, { "epoch": 0.16, "learning_rate": 1.9090483783535844e-05, "loss": 0.5819, "step": 940 }, { "epoch": 0.16, "learning_rate": 1.9088144306432737e-05, "loss": 0.5828, "step": 941 }, { "epoch": 0.16, "learning_rate": 1.9085801968091434e-05, "loss": 0.5859, "step": 942 }, { "epoch": 0.16, "learning_rate": 1.9083456769249387e-05, "loss": 0.5943, "step": 943 }, { "epoch": 0.16, "learning_rate": 1.9081108710644933e-05, "loss": 0.6001, "step": 944 }, { "epoch": 0.16, "learning_rate": 1.9078757793017317e-05, "loss": 0.5787, "step": 945 }, { "epoch": 0.16, "learning_rate": 1.907640401710668e-05, "loss": 0.5992, "step": 946 }, { "epoch": 0.16, "learning_rate": 1.9074047383654076e-05, "loss": 0.5944, "step": 947 }, { "epoch": 0.16, "learning_rate": 1.907168789340144e-05, "loss": 0.5834, "step": 948 }, { "epoch": 0.16, "learning_rate": 1.9069325547091614e-05, "loss": 0.6156, "step": 949 }, { "epoch": 0.16, "learning_rate": 1.9066960345468343e-05, "loss": 0.5864, "step": 950 }, { "epoch": 0.16, "learning_rate": 1.9064592289276268e-05, "loss": 0.5991, "step": 951 }, { "epoch": 0.16, "learning_rate": 1.906222137926093e-05, "loss": 0.5804, "step": 952 }, { "epoch": 0.17, "learning_rate": 1.9059847616168766e-05, "loss": 0.6111, "step": 953 }, { "epoch": 0.17, "learning_rate": 1.9057471000747113e-05, "loss": 0.6167, "step": 954 }, { "epoch": 0.17, "learning_rate": 1.9055091533744204e-05, "loss": 0.5921, "step": 955 }, { "epoch": 0.17, "learning_rate": 1.905270921590917e-05, "loss": 0.5899, "step": 956 }, { "epoch": 0.17, "learning_rate": 1.9050324047992045e-05, "loss": 0.5911, "step": 957 }, { "epoch": 0.17, "learning_rate": 1.904793603074375e-05, "loss": 0.5881, "step": 958 }, { "epoch": 0.17, "learning_rate": 1.9045545164916118e-05, "loss": 0.589, "step": 959 }, { "epoch": 0.17, "learning_rate": 1.9043151451261865e-05, "loss": 0.5886, "step": 960 }, { "epoch": 0.17, "learning_rate": 1.9040754890534604e-05, "loss": 0.5911, "step": 961 }, { "epoch": 0.17, "learning_rate": 1.9038355483488857e-05, "loss": 0.6036, "step": 962 }, { "epoch": 0.17, "learning_rate": 1.9035953230880026e-05, "loss": 0.5936, "step": 963 }, { "epoch": 0.17, "learning_rate": 1.9033548133464425e-05, "loss": 0.5976, "step": 964 }, { "epoch": 0.17, "learning_rate": 1.903114019199925e-05, "loss": 0.6108, "step": 965 }, { "epoch": 0.17, "learning_rate": 1.9028729407242598e-05, "loss": 0.5963, "step": 966 }, { "epoch": 0.17, "learning_rate": 1.902631577995347e-05, "loss": 0.5972, "step": 967 }, { "epoch": 0.17, "learning_rate": 1.9023899310891737e-05, "loss": 0.5988, "step": 968 }, { "epoch": 0.17, "learning_rate": 1.9021480000818193e-05, "loss": 0.6, "step": 969 }, { "epoch": 0.17, "learning_rate": 1.901905785049451e-05, "loss": 0.5871, "step": 970 }, { "epoch": 0.17, "learning_rate": 1.9016632860683257e-05, "loss": 0.5901, "step": 971 }, { "epoch": 0.17, "learning_rate": 1.9014205032147904e-05, "loss": 0.5891, "step": 972 }, { "epoch": 0.17, "learning_rate": 1.9011774365652802e-05, "loss": 0.5931, "step": 973 }, { "epoch": 0.17, "learning_rate": 1.9009340861963208e-05, "loss": 0.5779, "step": 974 }, { "epoch": 0.17, "learning_rate": 1.9006904521845263e-05, "loss": 0.5929, "step": 975 }, { "epoch": 0.17, "learning_rate": 1.9004465346066005e-05, "loss": 0.5966, "step": 976 }, { "epoch": 0.17, "learning_rate": 1.9002023335393366e-05, "loss": 0.5986, "step": 977 }, { "epoch": 0.17, "learning_rate": 1.8999578490596167e-05, "loss": 0.6057, "step": 978 }, { "epoch": 0.17, "learning_rate": 1.8997130812444127e-05, "loss": 0.5987, "step": 979 }, { "epoch": 0.17, "learning_rate": 1.899468030170785e-05, "loss": 0.5874, "step": 980 }, { "epoch": 0.17, "learning_rate": 1.8992226959158834e-05, "loss": 0.5933, "step": 981 }, { "epoch": 0.17, "learning_rate": 1.8989770785569472e-05, "loss": 0.5964, "step": 982 }, { "epoch": 0.17, "learning_rate": 1.898731178171305e-05, "loss": 0.6102, "step": 983 }, { "epoch": 0.17, "learning_rate": 1.898484994836373e-05, "loss": 0.5941, "step": 984 }, { "epoch": 0.17, "learning_rate": 1.8982385286296586e-05, "loss": 0.5926, "step": 985 }, { "epoch": 0.17, "learning_rate": 1.897991779628757e-05, "loss": 0.593, "step": 986 }, { "epoch": 0.17, "learning_rate": 1.897744747911352e-05, "loss": 0.5832, "step": 987 }, { "epoch": 0.17, "learning_rate": 1.897497433555218e-05, "loss": 0.5915, "step": 988 }, { "epoch": 0.17, "learning_rate": 1.8972498366382173e-05, "loss": 0.613, "step": 989 }, { "epoch": 0.17, "learning_rate": 1.8970019572383007e-05, "loss": 0.5864, "step": 990 }, { "epoch": 0.17, "learning_rate": 1.896753795433509e-05, "loss": 0.5991, "step": 991 }, { "epoch": 0.17, "learning_rate": 1.8965053513019715e-05, "loss": 0.5943, "step": 992 }, { "epoch": 0.17, "learning_rate": 1.8962566249219062e-05, "loss": 0.5885, "step": 993 }, { "epoch": 0.17, "learning_rate": 1.89600761637162e-05, "loss": 0.5935, "step": 994 }, { "epoch": 0.17, "learning_rate": 1.8957583257295094e-05, "loss": 0.5896, "step": 995 }, { "epoch": 0.17, "learning_rate": 1.8955087530740584e-05, "loss": 0.5989, "step": 996 }, { "epoch": 0.17, "learning_rate": 1.8952588984838404e-05, "loss": 0.5943, "step": 997 }, { "epoch": 0.17, "learning_rate": 1.895008762037518e-05, "loss": 0.6053, "step": 998 }, { "epoch": 0.17, "learning_rate": 1.894758343813842e-05, "loss": 0.5951, "step": 999 }, { "epoch": 0.17, "learning_rate": 1.8945076438916515e-05, "loss": 0.5954, "step": 1000 }, { "epoch": 0.17, "learning_rate": 1.8942566623498758e-05, "loss": 0.5991, "step": 1001 }, { "epoch": 0.17, "learning_rate": 1.894005399267532e-05, "loss": 0.6209, "step": 1002 }, { "epoch": 0.17, "learning_rate": 1.8937538547237247e-05, "loss": 0.6027, "step": 1003 }, { "epoch": 0.17, "learning_rate": 1.8935020287976486e-05, "loss": 0.581, "step": 1004 }, { "epoch": 0.17, "learning_rate": 1.893249921568587e-05, "loss": 0.6185, "step": 1005 }, { "epoch": 0.17, "learning_rate": 1.8929975331159106e-05, "loss": 0.6023, "step": 1006 }, { "epoch": 0.17, "learning_rate": 1.8927448635190796e-05, "loss": 0.6098, "step": 1007 }, { "epoch": 0.17, "learning_rate": 1.8924919128576428e-05, "loss": 0.578, "step": 1008 }, { "epoch": 0.17, "learning_rate": 1.892238681211237e-05, "loss": 0.5859, "step": 1009 }, { "epoch": 0.17, "learning_rate": 1.8919851686595875e-05, "loss": 0.583, "step": 1010 }, { "epoch": 0.18, "learning_rate": 1.891731375282508e-05, "loss": 0.5873, "step": 1011 }, { "epoch": 0.18, "learning_rate": 1.8914773011599012e-05, "loss": 0.5885, "step": 1012 }, { "epoch": 0.18, "learning_rate": 1.891222946371757e-05, "loss": 0.5927, "step": 1013 }, { "epoch": 0.18, "learning_rate": 1.8909683109981553e-05, "loss": 0.5866, "step": 1014 }, { "epoch": 0.18, "learning_rate": 1.890713395119263e-05, "loss": 0.603, "step": 1015 }, { "epoch": 0.18, "learning_rate": 1.890458198815336e-05, "loss": 0.5961, "step": 1016 }, { "epoch": 0.18, "learning_rate": 1.8902027221667177e-05, "loss": 0.5773, "step": 1017 }, { "epoch": 0.18, "learning_rate": 1.889946965253841e-05, "loss": 0.5904, "step": 1018 }, { "epoch": 0.18, "learning_rate": 1.8896909281572258e-05, "loss": 0.5961, "step": 1019 }, { "epoch": 0.18, "learning_rate": 1.889434610957481e-05, "loss": 0.5969, "step": 1020 }, { "epoch": 0.18, "learning_rate": 1.8891780137353036e-05, "loss": 0.6039, "step": 1021 }, { "epoch": 0.18, "learning_rate": 1.888921136571478e-05, "loss": 0.5726, "step": 1022 }, { "epoch": 0.18, "learning_rate": 1.8886639795468783e-05, "loss": 0.5963, "step": 1023 }, { "epoch": 0.18, "learning_rate": 1.8884065427424652e-05, "loss": 0.6029, "step": 1024 }, { "epoch": 0.18, "learning_rate": 1.8881488262392876e-05, "loss": 0.5892, "step": 1025 }, { "epoch": 0.18, "learning_rate": 1.8878908301184836e-05, "loss": 0.5982, "step": 1026 }, { "epoch": 0.18, "learning_rate": 1.8876325544612782e-05, "loss": 0.5967, "step": 1027 }, { "epoch": 0.18, "learning_rate": 1.8873739993489852e-05, "loss": 0.5968, "step": 1028 }, { "epoch": 0.18, "learning_rate": 1.887115164863006e-05, "loss": 0.5845, "step": 1029 }, { "epoch": 0.18, "learning_rate": 1.8868560510848296e-05, "loss": 0.5925, "step": 1030 }, { "epoch": 0.18, "learning_rate": 1.8865966580960334e-05, "loss": 0.607, "step": 1031 }, { "epoch": 0.18, "learning_rate": 1.8863369859782824e-05, "loss": 0.592, "step": 1032 }, { "epoch": 0.18, "learning_rate": 1.8860770348133305e-05, "loss": 0.5865, "step": 1033 }, { "epoch": 0.18, "learning_rate": 1.8858168046830176e-05, "loss": 0.5871, "step": 1034 }, { "epoch": 0.18, "learning_rate": 1.8855562956692732e-05, "loss": 0.6081, "step": 1035 }, { "epoch": 0.18, "learning_rate": 1.8852955078541134e-05, "loss": 0.5961, "step": 1036 }, { "epoch": 0.18, "learning_rate": 1.8850344413196426e-05, "loss": 0.5934, "step": 1037 }, { "epoch": 0.18, "learning_rate": 1.8847730961480534e-05, "loss": 0.6054, "step": 1038 }, { "epoch": 0.18, "learning_rate": 1.8845114724216248e-05, "loss": 0.581, "step": 1039 }, { "epoch": 0.18, "learning_rate": 1.884249570222725e-05, "loss": 0.5991, "step": 1040 }, { "epoch": 0.18, "learning_rate": 1.883987389633809e-05, "loss": 0.6051, "step": 1041 }, { "epoch": 0.18, "learning_rate": 1.8837249307374193e-05, "loss": 0.5883, "step": 1042 }, { "epoch": 0.18, "learning_rate": 1.883462193616187e-05, "loss": 0.5946, "step": 1043 }, { "epoch": 0.18, "learning_rate": 1.8831991783528293e-05, "loss": 0.5906, "step": 1044 }, { "epoch": 0.18, "learning_rate": 1.8829358850301524e-05, "loss": 0.6071, "step": 1045 }, { "epoch": 0.18, "learning_rate": 1.8826723137310492e-05, "loss": 0.5874, "step": 1046 }, { "epoch": 0.18, "learning_rate": 1.8824084645385005e-05, "loss": 0.5929, "step": 1047 }, { "epoch": 0.18, "learning_rate": 1.8821443375355746e-05, "loss": 0.6024, "step": 1048 }, { "epoch": 0.18, "learning_rate": 1.8818799328054265e-05, "loss": 0.5843, "step": 1049 }, { "epoch": 0.18, "learning_rate": 1.8816152504312998e-05, "loss": 0.6022, "step": 1050 }, { "epoch": 0.18, "learning_rate": 1.881350290496525e-05, "loss": 0.5821, "step": 1051 }, { "epoch": 0.18, "learning_rate": 1.88108505308452e-05, "loss": 0.5946, "step": 1052 }, { "epoch": 0.18, "learning_rate": 1.8808195382787894e-05, "loss": 0.581, "step": 1053 }, { "epoch": 0.18, "learning_rate": 1.8805537461629266e-05, "loss": 0.5898, "step": 1054 }, { "epoch": 0.18, "learning_rate": 1.8802876768206106e-05, "loss": 0.5691, "step": 1055 }, { "epoch": 0.18, "learning_rate": 1.880021330335609e-05, "loss": 0.598, "step": 1056 }, { "epoch": 0.18, "learning_rate": 1.8797547067917764e-05, "loss": 0.604, "step": 1057 }, { "epoch": 0.18, "learning_rate": 1.879487806273054e-05, "loss": 0.5989, "step": 1058 }, { "epoch": 0.18, "learning_rate": 1.8792206288634706e-05, "loss": 0.586, "step": 1059 }, { "epoch": 0.18, "learning_rate": 1.8789531746471422e-05, "loss": 0.6059, "step": 1060 }, { "epoch": 0.18, "learning_rate": 1.8786854437082725e-05, "loss": 0.5875, "step": 1061 }, { "epoch": 0.18, "learning_rate": 1.878417436131151e-05, "loss": 0.5911, "step": 1062 }, { "epoch": 0.18, "learning_rate": 1.8781491520001555e-05, "loss": 0.6055, "step": 1063 }, { "epoch": 0.18, "learning_rate": 1.8778805913997503e-05, "loss": 0.5967, "step": 1064 }, { "epoch": 0.18, "learning_rate": 1.8776117544144866e-05, "loss": 0.6076, "step": 1065 }, { "epoch": 0.18, "learning_rate": 1.877342641129003e-05, "loss": 0.5712, "step": 1066 }, { "epoch": 0.18, "learning_rate": 1.8770732516280256e-05, "loss": 0.5979, "step": 1067 }, { "epoch": 0.18, "learning_rate": 1.876803585996366e-05, "loss": 0.5938, "step": 1068 }, { "epoch": 0.19, "learning_rate": 1.876533644318924e-05, "loss": 0.5872, "step": 1069 }, { "epoch": 0.19, "learning_rate": 1.8762634266806852e-05, "loss": 0.6024, "step": 1070 }, { "epoch": 0.19, "learning_rate": 1.875992933166724e-05, "loss": 0.5924, "step": 1071 }, { "epoch": 0.19, "learning_rate": 1.8757221638621993e-05, "loss": 0.6062, "step": 1072 }, { "epoch": 0.19, "learning_rate": 1.8754511188523583e-05, "loss": 0.5948, "step": 1073 }, { "epoch": 0.19, "learning_rate": 1.8751797982225342e-05, "loss": 0.6088, "step": 1074 }, { "epoch": 0.19, "learning_rate": 1.8749082020581485e-05, "loss": 0.6021, "step": 1075 }, { "epoch": 0.19, "learning_rate": 1.8746363304447073e-05, "loss": 0.5812, "step": 1076 }, { "epoch": 0.19, "learning_rate": 1.8743641834678047e-05, "loss": 0.5914, "step": 1077 }, { "epoch": 0.19, "learning_rate": 1.8740917612131218e-05, "loss": 0.5965, "step": 1078 }, { "epoch": 0.19, "learning_rate": 1.873819063766425e-05, "loss": 0.5703, "step": 1079 }, { "epoch": 0.19, "learning_rate": 1.873546091213569e-05, "loss": 0.5895, "step": 1080 }, { "epoch": 0.19, "learning_rate": 1.8732728436404938e-05, "loss": 0.6019, "step": 1081 }, { "epoch": 0.19, "learning_rate": 1.8729993211332263e-05, "loss": 0.5966, "step": 1082 }, { "epoch": 0.19, "learning_rate": 1.8727255237778804e-05, "loss": 0.5864, "step": 1083 }, { "epoch": 0.19, "learning_rate": 1.8724514516606565e-05, "loss": 0.5806, "step": 1084 }, { "epoch": 0.19, "learning_rate": 1.872177104867841e-05, "loss": 0.5891, "step": 1085 }, { "epoch": 0.19, "learning_rate": 1.8719024834858065e-05, "loss": 0.6024, "step": 1086 }, { "epoch": 0.19, "learning_rate": 1.8716275876010135e-05, "loss": 0.5786, "step": 1087 }, { "epoch": 0.19, "learning_rate": 1.8713524173000075e-05, "loss": 0.5904, "step": 1088 }, { "epoch": 0.19, "learning_rate": 1.871076972669421e-05, "loss": 0.5882, "step": 1089 }, { "epoch": 0.19, "learning_rate": 1.870801253795973e-05, "loss": 0.5859, "step": 1090 }, { "epoch": 0.19, "learning_rate": 1.8705252607664683e-05, "loss": 0.5987, "step": 1091 }, { "epoch": 0.19, "learning_rate": 1.8702489936677987e-05, "loss": 0.6025, "step": 1092 }, { "epoch": 0.19, "learning_rate": 1.8699724525869413e-05, "loss": 0.5835, "step": 1093 }, { "epoch": 0.19, "learning_rate": 1.8696956376109602e-05, "loss": 0.5972, "step": 1094 }, { "epoch": 0.19, "learning_rate": 1.869418548827006e-05, "loss": 0.5862, "step": 1095 }, { "epoch": 0.19, "learning_rate": 1.8691411863223147e-05, "loss": 0.6063, "step": 1096 }, { "epoch": 0.19, "learning_rate": 1.8688635501842093e-05, "loss": 0.5841, "step": 1097 }, { "epoch": 0.19, "learning_rate": 1.8685856405000984e-05, "loss": 0.5855, "step": 1098 }, { "epoch": 0.19, "learning_rate": 1.8683074573574763e-05, "loss": 0.5863, "step": 1099 }, { "epoch": 0.19, "learning_rate": 1.868029000843925e-05, "loss": 0.5939, "step": 1100 }, { "epoch": 0.19, "learning_rate": 1.8677502710471105e-05, "loss": 0.5819, "step": 1101 }, { "epoch": 0.19, "learning_rate": 1.8674712680547865e-05, "loss": 0.6181, "step": 1102 }, { "epoch": 0.19, "learning_rate": 1.8671919919547914e-05, "loss": 0.6014, "step": 1103 }, { "epoch": 0.19, "learning_rate": 1.866912442835051e-05, "loss": 0.601, "step": 1104 }, { "epoch": 0.19, "learning_rate": 1.8666326207835758e-05, "loss": 0.5794, "step": 1105 }, { "epoch": 0.19, "learning_rate": 1.8663525258884626e-05, "loss": 0.5907, "step": 1106 }, { "epoch": 0.19, "learning_rate": 1.866072158237895e-05, "loss": 0.6141, "step": 1107 }, { "epoch": 0.19, "learning_rate": 1.8657915179201408e-05, "loss": 0.6042, "step": 1108 }, { "epoch": 0.19, "learning_rate": 1.865510605023555e-05, "loss": 0.5857, "step": 1109 }, { "epoch": 0.19, "learning_rate": 1.865229419636578e-05, "loss": 0.6012, "step": 1110 }, { "epoch": 0.19, "learning_rate": 1.8649479618477357e-05, "loss": 0.5853, "step": 1111 }, { "epoch": 0.19, "learning_rate": 1.86466623174564e-05, "loss": 0.5932, "step": 1112 }, { "epoch": 0.19, "learning_rate": 1.8643842294189888e-05, "loss": 0.5884, "step": 1113 }, { "epoch": 0.19, "learning_rate": 1.8641019549565654e-05, "loss": 0.5846, "step": 1114 }, { "epoch": 0.19, "learning_rate": 1.8638194084472384e-05, "loss": 0.5963, "step": 1115 }, { "epoch": 0.19, "learning_rate": 1.863536589979963e-05, "loss": 0.5857, "step": 1116 }, { "epoch": 0.19, "learning_rate": 1.8632534996437793e-05, "loss": 0.5835, "step": 1117 }, { "epoch": 0.19, "learning_rate": 1.862970137527813e-05, "loss": 0.6107, "step": 1118 }, { "epoch": 0.19, "learning_rate": 1.8626865037212756e-05, "loss": 0.6012, "step": 1119 }, { "epoch": 0.19, "learning_rate": 1.8624025983134643e-05, "loss": 0.5911, "step": 1120 }, { "epoch": 0.19, "learning_rate": 1.8621184213937615e-05, "loss": 0.587, "step": 1121 }, { "epoch": 0.19, "learning_rate": 1.8618339730516353e-05, "loss": 0.579, "step": 1122 }, { "epoch": 0.19, "learning_rate": 1.8615492533766387e-05, "loss": 0.5704, "step": 1123 }, { "epoch": 0.19, "learning_rate": 1.861264262458411e-05, "loss": 0.589, "step": 1124 }, { "epoch": 0.19, "learning_rate": 1.860979000386676e-05, "loss": 0.5886, "step": 1125 }, { "epoch": 0.2, "learning_rate": 1.860693467251244e-05, "loss": 0.5866, "step": 1126 }, { "epoch": 0.2, "learning_rate": 1.8604076631420095e-05, "loss": 0.5875, "step": 1127 }, { "epoch": 0.2, "learning_rate": 1.8601215881489528e-05, "loss": 0.5912, "step": 1128 }, { "epoch": 0.2, "learning_rate": 1.8598352423621394e-05, "loss": 0.5796, "step": 1129 }, { "epoch": 0.2, "learning_rate": 1.8595486258717198e-05, "loss": 0.5958, "step": 1130 }, { "epoch": 0.2, "learning_rate": 1.8592617387679304e-05, "loss": 0.596, "step": 1131 }, { "epoch": 0.2, "learning_rate": 1.858974581141093e-05, "loss": 0.5848, "step": 1132 }, { "epoch": 0.2, "learning_rate": 1.858687153081613e-05, "loss": 0.6046, "step": 1133 }, { "epoch": 0.2, "learning_rate": 1.8583994546799822e-05, "loss": 0.5836, "step": 1134 }, { "epoch": 0.2, "learning_rate": 1.858111486026778e-05, "loss": 0.5925, "step": 1135 }, { "epoch": 0.2, "learning_rate": 1.857823247212661e-05, "loss": 0.6041, "step": 1136 }, { "epoch": 0.2, "learning_rate": 1.8575347383283788e-05, "loss": 0.6006, "step": 1137 }, { "epoch": 0.2, "learning_rate": 1.8572459594647626e-05, "loss": 0.6089, "step": 1138 }, { "epoch": 0.2, "learning_rate": 1.8569569107127297e-05, "loss": 0.5736, "step": 1139 }, { "epoch": 0.2, "learning_rate": 1.8566675921632817e-05, "loss": 0.5941, "step": 1140 }, { "epoch": 0.2, "learning_rate": 1.8563780039075055e-05, "loss": 0.5767, "step": 1141 }, { "epoch": 0.2, "learning_rate": 1.8560881460365726e-05, "loss": 0.5897, "step": 1142 }, { "epoch": 0.2, "learning_rate": 1.855798018641739e-05, "loss": 0.6032, "step": 1143 }, { "epoch": 0.2, "learning_rate": 1.855507621814347e-05, "loss": 0.5854, "step": 1144 }, { "epoch": 0.2, "learning_rate": 1.8552169556458224e-05, "loss": 0.5915, "step": 1145 }, { "epoch": 0.2, "learning_rate": 1.854926020227676e-05, "loss": 0.5874, "step": 1146 }, { "epoch": 0.2, "learning_rate": 1.8546348156515036e-05, "loss": 0.6002, "step": 1147 }, { "epoch": 0.2, "learning_rate": 1.8543433420089862e-05, "loss": 0.5708, "step": 1148 }, { "epoch": 0.2, "learning_rate": 1.8540515993918886e-05, "loss": 0.5735, "step": 1149 }, { "epoch": 0.2, "learning_rate": 1.8537595878920607e-05, "loss": 0.5857, "step": 1150 }, { "epoch": 0.2, "learning_rate": 1.853467307601437e-05, "loss": 0.5817, "step": 1151 }, { "epoch": 0.2, "learning_rate": 1.8531747586120368e-05, "loss": 0.5919, "step": 1152 }, { "epoch": 0.2, "learning_rate": 1.8528819410159638e-05, "loss": 0.5913, "step": 1153 }, { "epoch": 0.2, "learning_rate": 1.8525888549054067e-05, "loss": 0.5974, "step": 1154 }, { "epoch": 0.2, "learning_rate": 1.8522955003726375e-05, "loss": 0.5993, "step": 1155 }, { "epoch": 0.2, "learning_rate": 1.8520018775100146e-05, "loss": 0.5909, "step": 1156 }, { "epoch": 0.2, "learning_rate": 1.851707986409979e-05, "loss": 0.608, "step": 1157 }, { "epoch": 0.2, "learning_rate": 1.8514138271650578e-05, "loss": 0.6027, "step": 1158 }, { "epoch": 0.2, "learning_rate": 1.851119399867861e-05, "loss": 0.5808, "step": 1159 }, { "epoch": 0.2, "learning_rate": 1.8508247046110843e-05, "loss": 0.5834, "step": 1160 }, { "epoch": 0.2, "learning_rate": 1.8505297414875066e-05, "loss": 0.5869, "step": 1161 }, { "epoch": 0.2, "learning_rate": 1.8502345105899922e-05, "loss": 0.5839, "step": 1162 }, { "epoch": 0.2, "learning_rate": 1.849939012011489e-05, "loss": 0.6121, "step": 1163 }, { "epoch": 0.2, "learning_rate": 1.8496432458450297e-05, "loss": 0.6093, "step": 1164 }, { "epoch": 0.2, "learning_rate": 1.8493472121837302e-05, "loss": 0.598, "step": 1165 }, { "epoch": 0.2, "learning_rate": 1.8490509111207925e-05, "loss": 0.6105, "step": 1166 }, { "epoch": 0.2, "learning_rate": 1.8487543427495004e-05, "loss": 0.5941, "step": 1167 }, { "epoch": 0.2, "learning_rate": 1.8484575071632242e-05, "loss": 0.5916, "step": 1168 }, { "epoch": 0.2, "learning_rate": 1.8481604044554164e-05, "loss": 0.6019, "step": 1169 }, { "epoch": 0.2, "learning_rate": 1.8478630347196147e-05, "loss": 0.588, "step": 1170 }, { "epoch": 0.2, "learning_rate": 1.8475653980494408e-05, "loss": 0.591, "step": 1171 }, { "epoch": 0.2, "learning_rate": 1.8472674945386e-05, "loss": 0.59, "step": 1172 }, { "epoch": 0.2, "learning_rate": 1.846969324280882e-05, "loss": 0.5995, "step": 1173 }, { "epoch": 0.2, "learning_rate": 1.84667088737016e-05, "loss": 0.592, "step": 1174 }, { "epoch": 0.2, "learning_rate": 1.8463721839003917e-05, "loss": 0.6115, "step": 1175 }, { "epoch": 0.2, "learning_rate": 1.846073213965619e-05, "loss": 0.5817, "step": 1176 }, { "epoch": 0.2, "learning_rate": 1.845773977659966e-05, "loss": 0.6005, "step": 1177 }, { "epoch": 0.2, "learning_rate": 1.845474475077643e-05, "loss": 0.5936, "step": 1178 }, { "epoch": 0.2, "learning_rate": 1.845174706312942e-05, "loss": 0.5963, "step": 1179 }, { "epoch": 0.2, "learning_rate": 1.844874671460241e-05, "loss": 0.5819, "step": 1180 }, { "epoch": 0.2, "learning_rate": 1.8445743706139994e-05, "loss": 0.5974, "step": 1181 }, { "epoch": 0.2, "learning_rate": 1.8442738038687623e-05, "loss": 0.5835, "step": 1182 }, { "epoch": 0.2, "learning_rate": 1.8439729713191572e-05, "loss": 0.6036, "step": 1183 }, { "epoch": 0.21, "learning_rate": 1.843671873059896e-05, "loss": 0.5982, "step": 1184 }, { "epoch": 0.21, "learning_rate": 1.8433705091857738e-05, "loss": 0.5797, "step": 1185 }, { "epoch": 0.21, "learning_rate": 1.8430688797916702e-05, "loss": 0.5784, "step": 1186 }, { "epoch": 0.21, "learning_rate": 1.842766984972547e-05, "loss": 0.5867, "step": 1187 }, { "epoch": 0.21, "learning_rate": 1.842464824823451e-05, "loss": 0.5816, "step": 1188 }, { "epoch": 0.21, "learning_rate": 1.8421623994395114e-05, "loss": 0.5838, "step": 1189 }, { "epoch": 0.21, "learning_rate": 1.8418597089159415e-05, "loss": 0.5726, "step": 1190 }, { "epoch": 0.21, "learning_rate": 1.841556753348038e-05, "loss": 0.5877, "step": 1191 }, { "epoch": 0.21, "learning_rate": 1.8412535328311813e-05, "loss": 0.594, "step": 1192 }, { "epoch": 0.21, "learning_rate": 1.8409500474608343e-05, "loss": 0.5845, "step": 1193 }, { "epoch": 0.21, "learning_rate": 1.8406462973325445e-05, "loss": 0.5836, "step": 1194 }, { "epoch": 0.21, "learning_rate": 1.840342282541942e-05, "loss": 0.589, "step": 1195 }, { "epoch": 0.21, "learning_rate": 1.8400380031847395e-05, "loss": 0.5901, "step": 1196 }, { "epoch": 0.21, "learning_rate": 1.8397334593567347e-05, "loss": 0.6005, "step": 1197 }, { "epoch": 0.21, "learning_rate": 1.8394286511538077e-05, "loss": 0.6083, "step": 1198 }, { "epoch": 0.21, "learning_rate": 1.839123578671922e-05, "loss": 0.5932, "step": 1199 }, { "epoch": 0.21, "learning_rate": 1.8388182420071237e-05, "loss": 0.5884, "step": 1200 }, { "epoch": 0.21, "learning_rate": 1.838512641255543e-05, "loss": 0.5828, "step": 1201 }, { "epoch": 0.21, "learning_rate": 1.8382067765133926e-05, "loss": 0.6054, "step": 1202 }, { "epoch": 0.21, "learning_rate": 1.8379006478769677e-05, "loss": 0.579, "step": 1203 }, { "epoch": 0.21, "learning_rate": 1.8375942554426486e-05, "loss": 0.593, "step": 1204 }, { "epoch": 0.21, "learning_rate": 1.837287599306897e-05, "loss": 0.5986, "step": 1205 }, { "epoch": 0.21, "learning_rate": 1.8369806795662582e-05, "loss": 0.576, "step": 1206 }, { "epoch": 0.21, "learning_rate": 1.8366734963173597e-05, "loss": 0.5945, "step": 1207 }, { "epoch": 0.21, "learning_rate": 1.836366049656913e-05, "loss": 0.5974, "step": 1208 }, { "epoch": 0.21, "learning_rate": 1.8360583396817123e-05, "loss": 0.5721, "step": 1209 }, { "epoch": 0.21, "learning_rate": 1.8357503664886345e-05, "loss": 0.5979, "step": 1210 }, { "epoch": 0.21, "learning_rate": 1.8354421301746393e-05, "loss": 0.5964, "step": 1211 }, { "epoch": 0.21, "learning_rate": 1.835133630836769e-05, "loss": 0.5862, "step": 1212 }, { "epoch": 0.21, "learning_rate": 1.8348248685721495e-05, "loss": 0.5958, "step": 1213 }, { "epoch": 0.21, "learning_rate": 1.834515843477989e-05, "loss": 0.5901, "step": 1214 }, { "epoch": 0.21, "learning_rate": 1.8342065556515787e-05, "loss": 0.5843, "step": 1215 }, { "epoch": 0.21, "learning_rate": 1.8338970051902913e-05, "loss": 0.5889, "step": 1216 }, { "epoch": 0.21, "learning_rate": 1.8335871921915843e-05, "loss": 0.5973, "step": 1217 }, { "epoch": 0.21, "learning_rate": 1.833277116752996e-05, "loss": 0.5856, "step": 1218 }, { "epoch": 0.21, "learning_rate": 1.8329667789721487e-05, "loss": 0.5965, "step": 1219 }, { "epoch": 0.21, "learning_rate": 1.8326561789467458e-05, "loss": 0.5893, "step": 1220 }, { "epoch": 0.21, "learning_rate": 1.8323453167745747e-05, "loss": 0.5861, "step": 1221 }, { "epoch": 0.21, "learning_rate": 1.832034192553505e-05, "loss": 0.5984, "step": 1222 }, { "epoch": 0.21, "learning_rate": 1.831722806381488e-05, "loss": 0.5895, "step": 1223 }, { "epoch": 0.21, "learning_rate": 1.831411158356558e-05, "loss": 0.5885, "step": 1224 }, { "epoch": 0.21, "learning_rate": 1.831099248576832e-05, "loss": 0.6007, "step": 1225 }, { "epoch": 0.21, "learning_rate": 1.830787077140509e-05, "loss": 0.5923, "step": 1226 }, { "epoch": 0.21, "learning_rate": 1.830474644145871e-05, "loss": 0.6029, "step": 1227 }, { "epoch": 0.21, "learning_rate": 1.8301619496912813e-05, "loss": 0.5734, "step": 1228 }, { "epoch": 0.21, "learning_rate": 1.8298489938751864e-05, "loss": 0.5912, "step": 1229 }, { "epoch": 0.21, "learning_rate": 1.8295357767961144e-05, "loss": 0.5908, "step": 1230 }, { "epoch": 0.21, "learning_rate": 1.8292222985526772e-05, "loss": 0.6007, "step": 1231 }, { "epoch": 0.21, "learning_rate": 1.8289085592435663e-05, "loss": 0.582, "step": 1232 }, { "epoch": 0.21, "learning_rate": 1.8285945589675576e-05, "loss": 0.5895, "step": 1233 }, { "epoch": 0.21, "learning_rate": 1.8282802978235084e-05, "loss": 0.5938, "step": 1234 }, { "epoch": 0.21, "learning_rate": 1.827965775910358e-05, "loss": 0.5912, "step": 1235 }, { "epoch": 0.21, "learning_rate": 1.827650993327128e-05, "loss": 0.5927, "step": 1236 }, { "epoch": 0.21, "learning_rate": 1.827335950172922e-05, "loss": 0.5965, "step": 1237 }, { "epoch": 0.21, "learning_rate": 1.827020646546926e-05, "loss": 0.5858, "step": 1238 }, { "epoch": 0.21, "learning_rate": 1.826705082548407e-05, "loss": 0.6157, "step": 1239 }, { "epoch": 0.21, "learning_rate": 1.826389258276715e-05, "loss": 0.5981, "step": 1240 }, { "epoch": 0.21, "learning_rate": 1.8260731738312817e-05, "loss": 0.5899, "step": 1241 }, { "epoch": 0.22, "learning_rate": 1.8257568293116204e-05, "loss": 0.574, "step": 1242 }, { "epoch": 0.22, "learning_rate": 1.8254402248173265e-05, "loss": 0.5752, "step": 1243 }, { "epoch": 0.22, "learning_rate": 1.825123360448077e-05, "loss": 0.5868, "step": 1244 }, { "epoch": 0.22, "learning_rate": 1.824806236303631e-05, "loss": 0.5879, "step": 1245 }, { "epoch": 0.22, "learning_rate": 1.82448885248383e-05, "loss": 0.5834, "step": 1246 }, { "epoch": 0.22, "learning_rate": 1.8241712090885956e-05, "loss": 0.5919, "step": 1247 }, { "epoch": 0.22, "learning_rate": 1.8238533062179325e-05, "loss": 0.5795, "step": 1248 }, { "epoch": 0.22, "learning_rate": 1.8235351439719266e-05, "loss": 0.5741, "step": 1249 }, { "epoch": 0.22, "learning_rate": 1.823216722450746e-05, "loss": 0.5864, "step": 1250 }, { "epoch": 0.22, "learning_rate": 1.8228980417546392e-05, "loss": 0.5912, "step": 1251 }, { "epoch": 0.22, "learning_rate": 1.8225791019839375e-05, "loss": 0.5851, "step": 1252 }, { "epoch": 0.22, "learning_rate": 1.8222599032390534e-05, "loss": 0.5933, "step": 1253 }, { "epoch": 0.22, "learning_rate": 1.8219404456204808e-05, "loss": 0.583, "step": 1254 }, { "epoch": 0.22, "learning_rate": 1.8216207292287945e-05, "loss": 0.5969, "step": 1255 }, { "epoch": 0.22, "learning_rate": 1.8213007541646527e-05, "loss": 0.5918, "step": 1256 }, { "epoch": 0.22, "learning_rate": 1.8209805205287932e-05, "loss": 0.5837, "step": 1257 }, { "epoch": 0.22, "learning_rate": 1.8206600284220353e-05, "loss": 0.5828, "step": 1258 }, { "epoch": 0.22, "learning_rate": 1.8203392779452804e-05, "loss": 0.5957, "step": 1259 }, { "epoch": 0.22, "learning_rate": 1.8200182691995113e-05, "loss": 0.5702, "step": 1260 }, { "epoch": 0.22, "learning_rate": 1.819697002285792e-05, "loss": 0.5847, "step": 1261 }, { "epoch": 0.22, "learning_rate": 1.819375477305267e-05, "loss": 0.5793, "step": 1262 }, { "epoch": 0.22, "learning_rate": 1.8190536943591627e-05, "loss": 0.6093, "step": 1263 }, { "epoch": 0.22, "learning_rate": 1.8187316535487868e-05, "loss": 0.6011, "step": 1264 }, { "epoch": 0.22, "learning_rate": 1.8184093549755284e-05, "loss": 0.601, "step": 1265 }, { "epoch": 0.22, "learning_rate": 1.8180867987408567e-05, "loss": 0.5908, "step": 1266 }, { "epoch": 0.22, "learning_rate": 1.8177639849463234e-05, "loss": 0.5892, "step": 1267 }, { "epoch": 0.22, "learning_rate": 1.8174409136935603e-05, "loss": 0.6083, "step": 1268 }, { "epoch": 0.22, "learning_rate": 1.81711758508428e-05, "loss": 0.601, "step": 1269 }, { "epoch": 0.22, "learning_rate": 1.816793999220278e-05, "loss": 0.5842, "step": 1270 }, { "epoch": 0.22, "learning_rate": 1.816470156203428e-05, "loss": 0.5704, "step": 1271 }, { "epoch": 0.22, "learning_rate": 1.816146056135687e-05, "loss": 0.5997, "step": 1272 }, { "epoch": 0.22, "learning_rate": 1.8158216991190917e-05, "loss": 0.5888, "step": 1273 }, { "epoch": 0.22, "learning_rate": 1.8154970852557604e-05, "loss": 0.5993, "step": 1274 }, { "epoch": 0.22, "learning_rate": 1.8151722146478913e-05, "loss": 0.5807, "step": 1275 }, { "epoch": 0.22, "learning_rate": 1.814847087397765e-05, "loss": 0.5875, "step": 1276 }, { "epoch": 0.22, "learning_rate": 1.814521703607741e-05, "loss": 0.5865, "step": 1277 }, { "epoch": 0.22, "learning_rate": 1.8141960633802612e-05, "loss": 0.583, "step": 1278 }, { "epoch": 0.22, "learning_rate": 1.813870166817847e-05, "loss": 0.5743, "step": 1279 }, { "epoch": 0.22, "learning_rate": 1.8135440140231013e-05, "loss": 0.5957, "step": 1280 }, { "epoch": 0.22, "learning_rate": 1.8132176050987077e-05, "loss": 0.5984, "step": 1281 }, { "epoch": 0.22, "learning_rate": 1.8128909401474298e-05, "loss": 0.5941, "step": 1282 }, { "epoch": 0.22, "learning_rate": 1.812564019272112e-05, "loss": 0.5791, "step": 1283 }, { "epoch": 0.22, "learning_rate": 1.81223684257568e-05, "loss": 0.5848, "step": 1284 }, { "epoch": 0.22, "learning_rate": 1.811909410161139e-05, "loss": 0.5896, "step": 1285 }, { "epoch": 0.22, "learning_rate": 1.8115817221315753e-05, "loss": 0.5834, "step": 1286 }, { "epoch": 0.22, "learning_rate": 1.8112537785901557e-05, "loss": 0.5897, "step": 1287 }, { "epoch": 0.22, "learning_rate": 1.8109255796401278e-05, "loss": 0.6102, "step": 1288 }, { "epoch": 0.22, "learning_rate": 1.8105971253848178e-05, "loss": 0.5906, "step": 1289 }, { "epoch": 0.22, "learning_rate": 1.8102684159276345e-05, "loss": 0.6039, "step": 1290 }, { "epoch": 0.22, "learning_rate": 1.809939451372066e-05, "loss": 0.5931, "step": 1291 }, { "epoch": 0.22, "learning_rate": 1.8096102318216807e-05, "loss": 0.5985, "step": 1292 }, { "epoch": 0.22, "learning_rate": 1.8092807573801277e-05, "loss": 0.5945, "step": 1293 }, { "epoch": 0.22, "learning_rate": 1.8089510281511357e-05, "loss": 0.6139, "step": 1294 }, { "epoch": 0.22, "learning_rate": 1.8086210442385146e-05, "loss": 0.6027, "step": 1295 }, { "epoch": 0.22, "learning_rate": 1.8082908057461534e-05, "loss": 0.6247, "step": 1296 }, { "epoch": 0.22, "learning_rate": 1.8079603127780216e-05, "loss": 0.6077, "step": 1297 }, { "epoch": 0.22, "learning_rate": 1.807629565438169e-05, "loss": 0.5951, "step": 1298 }, { "epoch": 0.23, "learning_rate": 1.8072985638307265e-05, "loss": 0.6001, "step": 1299 }, { "epoch": 0.23, "learning_rate": 1.8069673080599024e-05, "loss": 0.5903, "step": 1300 }, { "epoch": 0.23, "learning_rate": 1.8066357982299873e-05, "loss": 0.5817, "step": 1301 }, { "epoch": 0.23, "learning_rate": 1.8063040344453513e-05, "loss": 0.5936, "step": 1302 }, { "epoch": 0.23, "learning_rate": 1.805972016810444e-05, "loss": 0.581, "step": 1303 }, { "epoch": 0.23, "learning_rate": 1.8056397454297953e-05, "loss": 0.6131, "step": 1304 }, { "epoch": 0.23, "learning_rate": 1.8053072204080142e-05, "loss": 0.5882, "step": 1305 }, { "epoch": 0.23, "learning_rate": 1.8049744418497913e-05, "loss": 0.5726, "step": 1306 }, { "epoch": 0.23, "learning_rate": 1.8046414098598947e-05, "loss": 0.5935, "step": 1307 }, { "epoch": 0.23, "learning_rate": 1.804308124543175e-05, "loss": 0.5949, "step": 1308 }, { "epoch": 0.23, "learning_rate": 1.8039745860045595e-05, "loss": 0.6001, "step": 1309 }, { "epoch": 0.23, "learning_rate": 1.803640794349058e-05, "loss": 0.5931, "step": 1310 }, { "epoch": 0.23, "learning_rate": 1.803306749681758e-05, "loss": 0.609, "step": 1311 }, { "epoch": 0.23, "learning_rate": 1.8029724521078278e-05, "loss": 0.5965, "step": 1312 }, { "epoch": 0.23, "learning_rate": 1.8026379017325147e-05, "loss": 0.5904, "step": 1313 }, { "epoch": 0.23, "learning_rate": 1.8023030986611463e-05, "loss": 0.5799, "step": 1314 }, { "epoch": 0.23, "learning_rate": 1.8019680429991293e-05, "loss": 0.5848, "step": 1315 }, { "epoch": 0.23, "learning_rate": 1.8016327348519495e-05, "loss": 0.6127, "step": 1316 }, { "epoch": 0.23, "learning_rate": 1.8012971743251722e-05, "loss": 0.5885, "step": 1317 }, { "epoch": 0.23, "learning_rate": 1.8009613615244438e-05, "loss": 0.5973, "step": 1318 }, { "epoch": 0.23, "learning_rate": 1.800625296555488e-05, "loss": 0.5937, "step": 1319 }, { "epoch": 0.23, "learning_rate": 1.8002889795241087e-05, "loss": 0.6015, "step": 1320 }, { "epoch": 0.23, "learning_rate": 1.79995241053619e-05, "loss": 0.5598, "step": 1321 }, { "epoch": 0.23, "learning_rate": 1.799615589697694e-05, "loss": 0.5915, "step": 1322 }, { "epoch": 0.23, "learning_rate": 1.7992785171146633e-05, "loss": 0.5791, "step": 1323 }, { "epoch": 0.23, "learning_rate": 1.798941192893218e-05, "loss": 0.583, "step": 1324 }, { "epoch": 0.23, "learning_rate": 1.7986036171395594e-05, "loss": 0.5892, "step": 1325 }, { "epoch": 0.23, "learning_rate": 1.7982657899599672e-05, "loss": 0.6005, "step": 1326 }, { "epoch": 0.23, "learning_rate": 1.7979277114607996e-05, "loss": 0.5904, "step": 1327 }, { "epoch": 0.23, "learning_rate": 1.7975893817484948e-05, "loss": 0.5991, "step": 1328 }, { "epoch": 0.23, "learning_rate": 1.79725080092957e-05, "loss": 0.5903, "step": 1329 }, { "epoch": 0.23, "learning_rate": 1.796911969110621e-05, "loss": 0.6104, "step": 1330 }, { "epoch": 0.23, "learning_rate": 1.7965728863983228e-05, "loss": 0.5892, "step": 1331 }, { "epoch": 0.23, "learning_rate": 1.7962335528994296e-05, "loss": 0.5903, "step": 1332 }, { "epoch": 0.23, "learning_rate": 1.795893968720775e-05, "loss": 0.5998, "step": 1333 }, { "epoch": 0.23, "learning_rate": 1.7955541339692697e-05, "loss": 0.5871, "step": 1334 }, { "epoch": 0.23, "learning_rate": 1.7952140487519055e-05, "loss": 0.581, "step": 1335 }, { "epoch": 0.23, "learning_rate": 1.7948737131757518e-05, "loss": 0.5871, "step": 1336 }, { "epoch": 0.23, "learning_rate": 1.7945331273479577e-05, "loss": 0.5933, "step": 1337 }, { "epoch": 0.23, "learning_rate": 1.7941922913757492e-05, "loss": 0.5806, "step": 1338 }, { "epoch": 0.23, "learning_rate": 1.7938512053664335e-05, "loss": 0.605, "step": 1339 }, { "epoch": 0.23, "learning_rate": 1.793509869427395e-05, "loss": 0.5984, "step": 1340 }, { "epoch": 0.23, "learning_rate": 1.793168283666097e-05, "loss": 0.5971, "step": 1341 }, { "epoch": 0.23, "learning_rate": 1.7928264481900815e-05, "loss": 0.5884, "step": 1342 }, { "epoch": 0.23, "learning_rate": 1.79248436310697e-05, "loss": 0.5931, "step": 1343 }, { "epoch": 0.23, "learning_rate": 1.7921420285244612e-05, "loss": 0.607, "step": 1344 }, { "epoch": 0.23, "learning_rate": 1.791799444550333e-05, "loss": 0.5794, "step": 1345 }, { "epoch": 0.23, "learning_rate": 1.7914566112924413e-05, "loss": 0.5814, "step": 1346 }, { "epoch": 0.23, "learning_rate": 1.791113528858722e-05, "loss": 0.594, "step": 1347 }, { "epoch": 0.23, "learning_rate": 1.7907701973571878e-05, "loss": 0.5743, "step": 1348 }, { "epoch": 0.23, "learning_rate": 1.7904266168959306e-05, "loss": 0.5905, "step": 1349 }, { "epoch": 0.23, "learning_rate": 1.7900827875831207e-05, "loss": 0.6062, "step": 1350 }, { "epoch": 0.23, "learning_rate": 1.7897387095270058e-05, "loss": 0.5832, "step": 1351 }, { "epoch": 0.23, "learning_rate": 1.7893943828359136e-05, "loss": 0.5914, "step": 1352 }, { "epoch": 0.23, "learning_rate": 1.7890498076182484e-05, "loss": 0.5993, "step": 1353 }, { "epoch": 0.23, "learning_rate": 1.7887049839824938e-05, "loss": 0.5923, "step": 1354 }, { "epoch": 0.23, "learning_rate": 1.7883599120372116e-05, "loss": 0.5877, "step": 1355 }, { "epoch": 0.23, "learning_rate": 1.7880145918910406e-05, "loss": 0.5814, "step": 1356 }, { "epoch": 0.24, "learning_rate": 1.7876690236526995e-05, "loss": 0.5912, "step": 1357 }, { "epoch": 0.24, "learning_rate": 1.787323207430984e-05, "loss": 0.6054, "step": 1358 }, { "epoch": 0.24, "learning_rate": 1.786977143334768e-05, "loss": 0.5996, "step": 1359 }, { "epoch": 0.24, "learning_rate": 1.786630831473003e-05, "loss": 0.6031, "step": 1360 }, { "epoch": 0.24, "learning_rate": 1.7862842719547204e-05, "loss": 0.5843, "step": 1361 }, { "epoch": 0.24, "learning_rate": 1.785937464889027e-05, "loss": 0.5846, "step": 1362 }, { "epoch": 0.24, "learning_rate": 1.785590410385109e-05, "loss": 0.5867, "step": 1363 }, { "epoch": 0.24, "learning_rate": 1.7852431085522305e-05, "loss": 0.5924, "step": 1364 }, { "epoch": 0.24, "learning_rate": 1.784895559499733e-05, "loss": 0.5871, "step": 1365 }, { "epoch": 0.24, "learning_rate": 1.7845477633370364e-05, "loss": 0.5959, "step": 1366 }, { "epoch": 0.24, "learning_rate": 1.7841997201736375e-05, "loss": 0.5971, "step": 1367 }, { "epoch": 0.24, "learning_rate": 1.7838514301191117e-05, "loss": 0.5858, "step": 1368 }, { "epoch": 0.24, "learning_rate": 1.783502893283112e-05, "loss": 0.5746, "step": 1369 }, { "epoch": 0.24, "learning_rate": 1.7831541097753683e-05, "loss": 0.5937, "step": 1370 }, { "epoch": 0.24, "learning_rate": 1.78280507970569e-05, "loss": 0.6087, "step": 1371 }, { "epoch": 0.24, "learning_rate": 1.7824558031839613e-05, "loss": 0.5908, "step": 1372 }, { "epoch": 0.24, "learning_rate": 1.782106280320147e-05, "loss": 0.5969, "step": 1373 }, { "epoch": 0.24, "learning_rate": 1.781756511224287e-05, "loss": 0.5888, "step": 1374 }, { "epoch": 0.24, "learning_rate": 1.781406496006501e-05, "loss": 0.5915, "step": 1375 }, { "epoch": 0.24, "learning_rate": 1.7810562347769842e-05, "loss": 0.603, "step": 1376 }, { "epoch": 0.24, "learning_rate": 1.78070572764601e-05, "loss": 0.5859, "step": 1377 }, { "epoch": 0.24, "learning_rate": 1.780354974723929e-05, "loss": 0.5835, "step": 1378 }, { "epoch": 0.24, "learning_rate": 1.78000397612117e-05, "loss": 0.5805, "step": 1379 }, { "epoch": 0.24, "learning_rate": 1.7796527319482385e-05, "loss": 0.596, "step": 1380 }, { "epoch": 0.24, "learning_rate": 1.7793012423157172e-05, "loss": 0.5776, "step": 1381 }, { "epoch": 0.24, "learning_rate": 1.778949507334266e-05, "loss": 0.5768, "step": 1382 }, { "epoch": 0.24, "learning_rate": 1.778597527114623e-05, "loss": 0.5734, "step": 1383 }, { "epoch": 0.24, "learning_rate": 1.7782453017676025e-05, "loss": 0.5719, "step": 1384 }, { "epoch": 0.24, "learning_rate": 1.777892831404096e-05, "loss": 0.5967, "step": 1385 }, { "epoch": 0.24, "learning_rate": 1.777540116135073e-05, "loss": 0.5994, "step": 1386 }, { "epoch": 0.24, "learning_rate": 1.777187156071579e-05, "loss": 0.5949, "step": 1387 }, { "epoch": 0.24, "learning_rate": 1.7768339513247374e-05, "loss": 0.578, "step": 1388 }, { "epoch": 0.24, "learning_rate": 1.7764805020057476e-05, "loss": 0.5905, "step": 1389 }, { "epoch": 0.24, "learning_rate": 1.776126808225888e-05, "loss": 0.5921, "step": 1390 }, { "epoch": 0.24, "learning_rate": 1.7757728700965122e-05, "loss": 0.5893, "step": 1391 }, { "epoch": 0.24, "learning_rate": 1.7754186877290508e-05, "loss": 0.5837, "step": 1392 }, { "epoch": 0.24, "learning_rate": 1.7750642612350124e-05, "loss": 0.5851, "step": 1393 }, { "epoch": 0.24, "learning_rate": 1.7747095907259807e-05, "loss": 0.5809, "step": 1394 }, { "epoch": 0.24, "learning_rate": 1.7743546763136187e-05, "loss": 0.5926, "step": 1395 }, { "epoch": 0.24, "learning_rate": 1.773999518109664e-05, "loss": 0.5883, "step": 1396 }, { "epoch": 0.24, "learning_rate": 1.773644116225932e-05, "loss": 0.5962, "step": 1397 }, { "epoch": 0.24, "learning_rate": 1.7732884707743143e-05, "loss": 0.5951, "step": 1398 }, { "epoch": 0.24, "learning_rate": 1.7729325818667794e-05, "loss": 0.5991, "step": 1399 }, { "epoch": 0.24, "learning_rate": 1.772576449615373e-05, "loss": 0.5895, "step": 1400 }, { "epoch": 0.24, "learning_rate": 1.7722200741322163e-05, "loss": 0.5954, "step": 1401 }, { "epoch": 0.24, "learning_rate": 1.771863455529508e-05, "loss": 0.584, "step": 1402 }, { "epoch": 0.24, "learning_rate": 1.7715065939195235e-05, "loss": 0.5965, "step": 1403 }, { "epoch": 0.24, "learning_rate": 1.7711494894146138e-05, "loss": 0.5896, "step": 1404 }, { "epoch": 0.24, "learning_rate": 1.7707921421272064e-05, "loss": 0.5932, "step": 1405 }, { "epoch": 0.24, "learning_rate": 1.7704345521698057e-05, "loss": 0.5819, "step": 1406 }, { "epoch": 0.24, "learning_rate": 1.7700767196549934e-05, "loss": 0.5768, "step": 1407 }, { "epoch": 0.24, "learning_rate": 1.7697186446954257e-05, "loss": 0.597, "step": 1408 }, { "epoch": 0.24, "learning_rate": 1.7693603274038365e-05, "loss": 0.6008, "step": 1409 }, { "epoch": 0.24, "learning_rate": 1.769001767893035e-05, "loss": 0.5928, "step": 1410 }, { "epoch": 0.24, "learning_rate": 1.7686429662759076e-05, "loss": 0.5732, "step": 1411 }, { "epoch": 0.24, "learning_rate": 1.7682839226654168e-05, "loss": 0.5876, "step": 1412 }, { "epoch": 0.24, "learning_rate": 1.7679246371746e-05, "loss": 0.5791, "step": 1413 }, { "epoch": 0.24, "learning_rate": 1.7675651099165732e-05, "loss": 0.5833, "step": 1414 }, { "epoch": 0.25, "learning_rate": 1.767205341004526e-05, "loss": 0.5991, "step": 1415 }, { "epoch": 0.25, "learning_rate": 1.766845330551725e-05, "loss": 0.5785, "step": 1416 }, { "epoch": 0.25, "learning_rate": 1.766485078671514e-05, "loss": 0.5828, "step": 1417 }, { "epoch": 0.25, "learning_rate": 1.7661245854773103e-05, "loss": 0.5837, "step": 1418 }, { "epoch": 0.25, "learning_rate": 1.76576385108261e-05, "loss": 0.5953, "step": 1419 }, { "epoch": 0.25, "learning_rate": 1.765402875600984e-05, "loss": 0.6141, "step": 1420 }, { "epoch": 0.25, "learning_rate": 1.7650416591460776e-05, "loss": 0.6022, "step": 1421 }, { "epoch": 0.25, "learning_rate": 1.7646802018316143e-05, "loss": 0.5884, "step": 1422 }, { "epoch": 0.25, "learning_rate": 1.764318503771392e-05, "loss": 0.6088, "step": 1423 }, { "epoch": 0.25, "learning_rate": 1.7639565650792846e-05, "loss": 0.5684, "step": 1424 }, { "epoch": 0.25, "learning_rate": 1.763594385869243e-05, "loss": 0.5939, "step": 1425 }, { "epoch": 0.25, "learning_rate": 1.7632319662552914e-05, "loss": 0.579, "step": 1426 }, { "epoch": 0.25, "learning_rate": 1.762869306351532e-05, "loss": 0.5846, "step": 1427 }, { "epoch": 0.25, "learning_rate": 1.7625064062721414e-05, "loss": 0.5896, "step": 1428 }, { "epoch": 0.25, "learning_rate": 1.762143266131372e-05, "loss": 0.5958, "step": 1429 }, { "epoch": 0.25, "learning_rate": 1.7617798860435526e-05, "loss": 0.5747, "step": 1430 }, { "epoch": 0.25, "learning_rate": 1.761416266123086e-05, "loss": 0.5948, "step": 1431 }, { "epoch": 0.25, "learning_rate": 1.761052406484452e-05, "loss": 0.5858, "step": 1432 }, { "epoch": 0.25, "learning_rate": 1.7606883072422048e-05, "loss": 0.5965, "step": 1433 }, { "epoch": 0.25, "learning_rate": 1.7603239685109746e-05, "loss": 0.5792, "step": 1434 }, { "epoch": 0.25, "learning_rate": 1.759959390405467e-05, "loss": 0.5893, "step": 1435 }, { "epoch": 0.25, "learning_rate": 1.7595945730404627e-05, "loss": 0.5761, "step": 1436 }, { "epoch": 0.25, "learning_rate": 1.759229516530818e-05, "loss": 0.5969, "step": 1437 }, { "epoch": 0.25, "learning_rate": 1.7588642209914644e-05, "loss": 0.5908, "step": 1438 }, { "epoch": 0.25, "learning_rate": 1.7584986865374084e-05, "loss": 0.5862, "step": 1439 }, { "epoch": 0.25, "learning_rate": 1.758132913283732e-05, "loss": 0.5951, "step": 1440 }, { "epoch": 0.25, "learning_rate": 1.757766901345592e-05, "loss": 0.598, "step": 1441 }, { "epoch": 0.25, "learning_rate": 1.7574006508382214e-05, "loss": 0.5828, "step": 1442 }, { "epoch": 0.25, "learning_rate": 1.757034161876927e-05, "loss": 0.5805, "step": 1443 }, { "epoch": 0.25, "learning_rate": 1.7566674345770912e-05, "loss": 0.6095, "step": 1444 }, { "epoch": 0.25, "learning_rate": 1.756300469054172e-05, "loss": 0.578, "step": 1445 }, { "epoch": 0.25, "learning_rate": 1.755933265423701e-05, "loss": 0.5911, "step": 1446 }, { "epoch": 0.25, "learning_rate": 1.7555658238012867e-05, "loss": 0.58, "step": 1447 }, { "epoch": 0.25, "learning_rate": 1.7551981443026104e-05, "loss": 0.5931, "step": 1448 }, { "epoch": 0.25, "learning_rate": 1.75483022704343e-05, "loss": 0.5956, "step": 1449 }, { "epoch": 0.25, "learning_rate": 1.754462072139578e-05, "loss": 0.6017, "step": 1450 }, { "epoch": 0.25, "learning_rate": 1.75409367970696e-05, "loss": 0.5981, "step": 1451 }, { "epoch": 0.25, "learning_rate": 1.753725049861559e-05, "loss": 0.5895, "step": 1452 }, { "epoch": 0.25, "learning_rate": 1.753356182719431e-05, "loss": 0.5875, "step": 1453 }, { "epoch": 0.25, "learning_rate": 1.7529870783967066e-05, "loss": 0.5953, "step": 1454 }, { "epoch": 0.25, "learning_rate": 1.7526177370095924e-05, "loss": 0.5874, "step": 1455 }, { "epoch": 0.25, "learning_rate": 1.752248158674369e-05, "loss": 0.5885, "step": 1456 }, { "epoch": 0.25, "learning_rate": 1.751878343507391e-05, "loss": 0.5982, "step": 1457 }, { "epoch": 0.25, "learning_rate": 1.7515082916250876e-05, "loss": 0.58, "step": 1458 }, { "epoch": 0.25, "learning_rate": 1.7511380031439635e-05, "loss": 0.5967, "step": 1459 }, { "epoch": 0.25, "learning_rate": 1.7507674781805976e-05, "loss": 0.58, "step": 1460 }, { "epoch": 0.25, "learning_rate": 1.7503967168516426e-05, "loss": 0.5859, "step": 1461 }, { "epoch": 0.25, "learning_rate": 1.7500257192738263e-05, "loss": 0.5777, "step": 1462 }, { "epoch": 0.25, "learning_rate": 1.74965448556395e-05, "loss": 0.5978, "step": 1463 }, { "epoch": 0.25, "learning_rate": 1.749283015838891e-05, "loss": 0.5834, "step": 1464 }, { "epoch": 0.25, "learning_rate": 1.748911310215599e-05, "loss": 0.5946, "step": 1465 }, { "epoch": 0.25, "learning_rate": 1.7485393688110987e-05, "loss": 0.5925, "step": 1466 }, { "epoch": 0.25, "learning_rate": 1.7481671917424895e-05, "loss": 0.5883, "step": 1467 }, { "epoch": 0.25, "learning_rate": 1.7477947791269447e-05, "loss": 0.5896, "step": 1468 }, { "epoch": 0.25, "learning_rate": 1.7474221310817114e-05, "loss": 0.6068, "step": 1469 }, { "epoch": 0.25, "learning_rate": 1.7470492477241113e-05, "loss": 0.5918, "step": 1470 }, { "epoch": 0.25, "learning_rate": 1.7466761291715403e-05, "loss": 0.5765, "step": 1471 }, { "epoch": 0.25, "learning_rate": 1.746302775541467e-05, "loss": 0.5907, "step": 1472 }, { "epoch": 0.26, "learning_rate": 1.7459291869514363e-05, "loss": 0.5726, "step": 1473 }, { "epoch": 0.26, "learning_rate": 1.7455553635190652e-05, "loss": 0.5996, "step": 1474 }, { "epoch": 0.26, "learning_rate": 1.7451813053620452e-05, "loss": 0.5878, "step": 1475 }, { "epoch": 0.26, "learning_rate": 1.744807012598142e-05, "loss": 0.5736, "step": 1476 }, { "epoch": 0.26, "learning_rate": 1.7444324853451947e-05, "loss": 0.5934, "step": 1477 }, { "epoch": 0.26, "learning_rate": 1.7440577237211168e-05, "loss": 0.5694, "step": 1478 }, { "epoch": 0.26, "learning_rate": 1.7436827278438947e-05, "loss": 0.599, "step": 1479 }, { "epoch": 0.26, "learning_rate": 1.743307497831589e-05, "loss": 0.5784, "step": 1480 }, { "epoch": 0.26, "learning_rate": 1.7429320338023355e-05, "loss": 0.609, "step": 1481 }, { "epoch": 0.26, "learning_rate": 1.7425563358743403e-05, "loss": 0.5989, "step": 1482 }, { "epoch": 0.26, "learning_rate": 1.7421804041658867e-05, "loss": 0.5718, "step": 1483 }, { "epoch": 0.26, "learning_rate": 1.741804238795329e-05, "loss": 0.5923, "step": 1484 }, { "epoch": 0.26, "learning_rate": 1.7414278398810966e-05, "loss": 0.5921, "step": 1485 }, { "epoch": 0.26, "learning_rate": 1.7410512075416915e-05, "loss": 0.5957, "step": 1486 }, { "epoch": 0.26, "learning_rate": 1.74067434189569e-05, "loss": 0.5919, "step": 1487 }, { "epoch": 0.26, "learning_rate": 1.7402972430617414e-05, "loss": 0.5953, "step": 1488 }, { "epoch": 0.26, "learning_rate": 1.7399199111585683e-05, "loss": 0.5828, "step": 1489 }, { "epoch": 0.26, "learning_rate": 1.739542346304967e-05, "loss": 0.5947, "step": 1490 }, { "epoch": 0.26, "learning_rate": 1.7391645486198068e-05, "loss": 0.5756, "step": 1491 }, { "epoch": 0.26, "learning_rate": 1.7387865182220305e-05, "loss": 0.5868, "step": 1492 }, { "epoch": 0.26, "learning_rate": 1.738408255230654e-05, "loss": 0.5955, "step": 1493 }, { "epoch": 0.26, "learning_rate": 1.738029759764767e-05, "loss": 0.5894, "step": 1494 }, { "epoch": 0.26, "learning_rate": 1.7376510319435314e-05, "loss": 0.5756, "step": 1495 }, { "epoch": 0.26, "learning_rate": 1.7372720718861838e-05, "loss": 0.5911, "step": 1496 }, { "epoch": 0.26, "learning_rate": 1.736892879712032e-05, "loss": 0.6076, "step": 1497 }, { "epoch": 0.26, "learning_rate": 1.736513455540458e-05, "loss": 0.5948, "step": 1498 }, { "epoch": 0.26, "learning_rate": 1.7361337994909166e-05, "loss": 0.5669, "step": 1499 }, { "epoch": 0.26, "learning_rate": 1.735753911682936e-05, "loss": 0.5829, "step": 1500 }, { "epoch": 0.26, "learning_rate": 1.735373792236117e-05, "loss": 0.5809, "step": 1501 }, { "epoch": 0.26, "learning_rate": 1.7349934412701328e-05, "loss": 0.5921, "step": 1502 }, { "epoch": 0.26, "learning_rate": 1.7346128589047308e-05, "loss": 0.5824, "step": 1503 }, { "epoch": 0.26, "learning_rate": 1.7342320452597296e-05, "loss": 0.5818, "step": 1504 }, { "epoch": 0.26, "learning_rate": 1.7338510004550225e-05, "loss": 0.5845, "step": 1505 }, { "epoch": 0.26, "learning_rate": 1.7334697246105738e-05, "loss": 0.5969, "step": 1506 }, { "epoch": 0.26, "learning_rate": 1.7330882178464215e-05, "loss": 0.5735, "step": 1507 }, { "epoch": 0.26, "learning_rate": 1.7327064802826762e-05, "loss": 0.6134, "step": 1508 }, { "epoch": 0.26, "learning_rate": 1.7323245120395217e-05, "loss": 0.5739, "step": 1509 }, { "epoch": 0.26, "learning_rate": 1.7319423132372125e-05, "loss": 0.5893, "step": 1510 }, { "epoch": 0.26, "learning_rate": 1.731559883996078e-05, "loss": 0.5848, "step": 1511 }, { "epoch": 0.26, "learning_rate": 1.731177224436519e-05, "loss": 0.5999, "step": 1512 }, { "epoch": 0.26, "learning_rate": 1.7307943346790086e-05, "loss": 0.5815, "step": 1513 }, { "epoch": 0.26, "learning_rate": 1.7304112148440927e-05, "loss": 0.5955, "step": 1514 }, { "epoch": 0.26, "learning_rate": 1.73002786505239e-05, "loss": 0.5715, "step": 1515 }, { "epoch": 0.26, "learning_rate": 1.7296442854245915e-05, "loss": 0.5869, "step": 1516 }, { "epoch": 0.26, "learning_rate": 1.7292604760814597e-05, "loss": 0.595, "step": 1517 }, { "epoch": 0.26, "learning_rate": 1.7288764371438304e-05, "loss": 0.5833, "step": 1518 }, { "epoch": 0.26, "learning_rate": 1.728492168732611e-05, "loss": 0.5843, "step": 1519 }, { "epoch": 0.26, "learning_rate": 1.728107670968782e-05, "loss": 0.5814, "step": 1520 }, { "epoch": 0.26, "learning_rate": 1.727722943973395e-05, "loss": 0.5697, "step": 1521 }, { "epoch": 0.26, "learning_rate": 1.7273379878675752e-05, "loss": 0.5683, "step": 1522 }, { "epoch": 0.26, "learning_rate": 1.7269528027725182e-05, "loss": 0.5748, "step": 1523 }, { "epoch": 0.26, "learning_rate": 1.726567388809493e-05, "loss": 0.5874, "step": 1524 }, { "epoch": 0.26, "learning_rate": 1.7261817460998398e-05, "loss": 0.6045, "step": 1525 }, { "epoch": 0.26, "learning_rate": 1.725795874764972e-05, "loss": 0.5891, "step": 1526 }, { "epoch": 0.26, "learning_rate": 1.7254097749263735e-05, "loss": 0.5916, "step": 1527 }, { "epoch": 0.26, "learning_rate": 1.725023446705601e-05, "loss": 0.6018, "step": 1528 }, { "epoch": 0.26, "learning_rate": 1.7246368902242837e-05, "loss": 0.5781, "step": 1529 }, { "epoch": 0.27, "learning_rate": 1.724250105604121e-05, "loss": 0.5883, "step": 1530 }, { "epoch": 0.27, "learning_rate": 1.7238630929668855e-05, "loss": 0.5667, "step": 1531 }, { "epoch": 0.27, "learning_rate": 1.723475852434421e-05, "loss": 0.5925, "step": 1532 }, { "epoch": 0.27, "learning_rate": 1.7230883841286433e-05, "loss": 0.5993, "step": 1533 }, { "epoch": 0.27, "learning_rate": 1.72270068817154e-05, "loss": 0.5759, "step": 1534 }, { "epoch": 0.27, "learning_rate": 1.7223127646851698e-05, "loss": 0.5861, "step": 1535 }, { "epoch": 0.27, "learning_rate": 1.721924613791663e-05, "loss": 0.584, "step": 1536 }, { "epoch": 0.27, "learning_rate": 1.7215362356132232e-05, "loss": 0.5978, "step": 1537 }, { "epoch": 0.27, "learning_rate": 1.721147630272123e-05, "loss": 0.5978, "step": 1538 }, { "epoch": 0.27, "learning_rate": 1.7207587978907085e-05, "loss": 0.5749, "step": 1539 }, { "epoch": 0.27, "learning_rate": 1.7203697385913965e-05, "loss": 0.5842, "step": 1540 }, { "epoch": 0.27, "learning_rate": 1.719980452496675e-05, "loss": 0.5952, "step": 1541 }, { "epoch": 0.27, "learning_rate": 1.7195909397291036e-05, "loss": 0.6013, "step": 1542 }, { "epoch": 0.27, "learning_rate": 1.7192012004113138e-05, "loss": 0.5732, "step": 1543 }, { "epoch": 0.27, "learning_rate": 1.7188112346660077e-05, "loss": 0.6002, "step": 1544 }, { "epoch": 0.27, "learning_rate": 1.7184210426159594e-05, "loss": 0.5867, "step": 1545 }, { "epoch": 0.27, "learning_rate": 1.7180306243840133e-05, "loss": 0.5922, "step": 1546 }, { "epoch": 0.27, "learning_rate": 1.7176399800930852e-05, "loss": 0.5865, "step": 1547 }, { "epoch": 0.27, "learning_rate": 1.7172491098661636e-05, "loss": 0.6037, "step": 1548 }, { "epoch": 0.27, "learning_rate": 1.7168580138263064e-05, "loss": 0.5733, "step": 1549 }, { "epoch": 0.27, "learning_rate": 1.7164666920966424e-05, "loss": 0.598, "step": 1550 }, { "epoch": 0.27, "learning_rate": 1.716075144800373e-05, "loss": 0.5701, "step": 1551 }, { "epoch": 0.27, "learning_rate": 1.7156833720607698e-05, "loss": 0.5943, "step": 1552 }, { "epoch": 0.27, "learning_rate": 1.7152913740011746e-05, "loss": 0.5963, "step": 1553 }, { "epoch": 0.27, "learning_rate": 1.714899150745002e-05, "loss": 0.5692, "step": 1554 }, { "epoch": 0.27, "learning_rate": 1.7145067024157355e-05, "loss": 0.5867, "step": 1555 }, { "epoch": 0.27, "learning_rate": 1.7141140291369312e-05, "loss": 0.5817, "step": 1556 }, { "epoch": 0.27, "learning_rate": 1.713721131032214e-05, "loss": 0.5674, "step": 1557 }, { "epoch": 0.27, "learning_rate": 1.713328008225282e-05, "loss": 0.5951, "step": 1558 }, { "epoch": 0.27, "learning_rate": 1.7129346608399027e-05, "loss": 0.5773, "step": 1559 }, { "epoch": 0.27, "learning_rate": 1.7125410889999135e-05, "loss": 0.5849, "step": 1560 }, { "epoch": 0.27, "learning_rate": 1.712147292829224e-05, "loss": 0.5804, "step": 1561 }, { "epoch": 0.27, "learning_rate": 1.711753272451814e-05, "loss": 0.6007, "step": 1562 }, { "epoch": 0.27, "learning_rate": 1.7113590279917332e-05, "loss": 0.6036, "step": 1563 }, { "epoch": 0.27, "learning_rate": 1.7109645595731028e-05, "loss": 0.5972, "step": 1564 }, { "epoch": 0.27, "learning_rate": 1.710569867320114e-05, "loss": 0.5858, "step": 1565 }, { "epoch": 0.27, "learning_rate": 1.7101749513570282e-05, "loss": 0.5781, "step": 1566 }, { "epoch": 0.27, "learning_rate": 1.709779811808178e-05, "loss": 0.5937, "step": 1567 }, { "epoch": 0.27, "learning_rate": 1.7093844487979658e-05, "loss": 0.5968, "step": 1568 }, { "epoch": 0.27, "learning_rate": 1.7089888624508638e-05, "loss": 0.5728, "step": 1569 }, { "epoch": 0.27, "learning_rate": 1.7085930528914166e-05, "loss": 0.6125, "step": 1570 }, { "epoch": 0.27, "learning_rate": 1.7081970202442363e-05, "loss": 0.5845, "step": 1571 }, { "epoch": 0.27, "learning_rate": 1.707800764634008e-05, "loss": 0.6075, "step": 1572 }, { "epoch": 0.27, "learning_rate": 1.7074042861854842e-05, "loss": 0.5912, "step": 1573 }, { "epoch": 0.27, "learning_rate": 1.70700758502349e-05, "loss": 0.5916, "step": 1574 }, { "epoch": 0.27, "learning_rate": 1.7066106612729196e-05, "loss": 0.6012, "step": 1575 }, { "epoch": 0.27, "learning_rate": 1.7062135150587367e-05, "loss": 0.5882, "step": 1576 }, { "epoch": 0.27, "learning_rate": 1.705816146505976e-05, "loss": 0.5911, "step": 1577 }, { "epoch": 0.27, "learning_rate": 1.7054185557397413e-05, "loss": 0.586, "step": 1578 }, { "epoch": 0.27, "learning_rate": 1.705020742885208e-05, "loss": 0.5877, "step": 1579 }, { "epoch": 0.27, "learning_rate": 1.704622708067619e-05, "loss": 0.5829, "step": 1580 }, { "epoch": 0.27, "learning_rate": 1.704224451412289e-05, "loss": 0.5791, "step": 1581 }, { "epoch": 0.27, "learning_rate": 1.703825973044602e-05, "loss": 0.5915, "step": 1582 }, { "epoch": 0.27, "learning_rate": 1.7034272730900116e-05, "loss": 0.5863, "step": 1583 }, { "epoch": 0.27, "learning_rate": 1.7030283516740414e-05, "loss": 0.5879, "step": 1584 }, { "epoch": 0.27, "learning_rate": 1.7026292089222844e-05, "loss": 0.5816, "step": 1585 }, { "epoch": 0.27, "learning_rate": 1.7022298449604037e-05, "loss": 0.5786, "step": 1586 }, { "epoch": 0.27, "learning_rate": 1.7018302599141313e-05, "loss": 0.5849, "step": 1587 }, { "epoch": 0.28, "learning_rate": 1.7014304539092705e-05, "loss": 0.5851, "step": 1588 }, { "epoch": 0.28, "learning_rate": 1.7010304270716917e-05, "loss": 0.6043, "step": 1589 }, { "epoch": 0.28, "learning_rate": 1.7006301795273365e-05, "loss": 0.5698, "step": 1590 }, { "epoch": 0.28, "learning_rate": 1.700229711402216e-05, "loss": 0.5888, "step": 1591 }, { "epoch": 0.28, "learning_rate": 1.6998290228224106e-05, "loss": 0.5926, "step": 1592 }, { "epoch": 0.28, "learning_rate": 1.699428113914069e-05, "loss": 0.596, "step": 1593 }, { "epoch": 0.28, "learning_rate": 1.6990269848034104e-05, "loss": 0.5695, "step": 1594 }, { "epoch": 0.28, "learning_rate": 1.6986256356167237e-05, "loss": 0.5887, "step": 1595 }, { "epoch": 0.28, "learning_rate": 1.698224066480366e-05, "loss": 0.5798, "step": 1596 }, { "epoch": 0.28, "learning_rate": 1.6978222775207636e-05, "loss": 0.5949, "step": 1597 }, { "epoch": 0.28, "learning_rate": 1.6974202688644132e-05, "loss": 0.577, "step": 1598 }, { "epoch": 0.28, "learning_rate": 1.69701804063788e-05, "loss": 0.5924, "step": 1599 }, { "epoch": 0.28, "learning_rate": 1.6966155929677973e-05, "loss": 0.5978, "step": 1600 }, { "epoch": 0.28, "learning_rate": 1.69621292598087e-05, "loss": 0.5988, "step": 1601 }, { "epoch": 0.28, "learning_rate": 1.695810039803869e-05, "loss": 0.5786, "step": 1602 }, { "epoch": 0.28, "learning_rate": 1.695406934563637e-05, "loss": 0.589, "step": 1603 }, { "epoch": 0.28, "learning_rate": 1.695003610387084e-05, "loss": 0.5896, "step": 1604 }, { "epoch": 0.28, "learning_rate": 1.6946000674011892e-05, "loss": 0.6019, "step": 1605 }, { "epoch": 0.28, "learning_rate": 1.6941963057330006e-05, "loss": 0.5781, "step": 1606 }, { "epoch": 0.28, "learning_rate": 1.6937923255096354e-05, "loss": 0.5852, "step": 1607 }, { "epoch": 0.28, "learning_rate": 1.69338812685828e-05, "loss": 0.5733, "step": 1608 }, { "epoch": 0.28, "learning_rate": 1.6929837099061885e-05, "loss": 0.5701, "step": 1609 }, { "epoch": 0.28, "learning_rate": 1.6925790747806845e-05, "loss": 0.5741, "step": 1610 }, { "epoch": 0.28, "learning_rate": 1.6921742216091602e-05, "loss": 0.5743, "step": 1611 }, { "epoch": 0.28, "learning_rate": 1.6917691505190756e-05, "loss": 0.5722, "step": 1612 }, { "epoch": 0.28, "learning_rate": 1.691363861637961e-05, "loss": 0.6066, "step": 1613 }, { "epoch": 0.28, "learning_rate": 1.6909583550934137e-05, "loss": 0.5668, "step": 1614 }, { "epoch": 0.28, "learning_rate": 1.6905526310131e-05, "loss": 0.5894, "step": 1615 }, { "epoch": 0.28, "learning_rate": 1.6901466895247554e-05, "loss": 0.595, "step": 1616 }, { "epoch": 0.28, "learning_rate": 1.6897405307561826e-05, "loss": 0.5914, "step": 1617 }, { "epoch": 0.28, "learning_rate": 1.689334154835254e-05, "loss": 0.5871, "step": 1618 }, { "epoch": 0.28, "learning_rate": 1.688927561889909e-05, "loss": 0.5872, "step": 1619 }, { "epoch": 0.28, "learning_rate": 1.6885207520481568e-05, "loss": 0.61, "step": 1620 }, { "epoch": 0.28, "learning_rate": 1.6881137254380736e-05, "loss": 0.5827, "step": 1621 }, { "epoch": 0.28, "learning_rate": 1.6877064821878044e-05, "loss": 0.57, "step": 1622 }, { "epoch": 0.28, "learning_rate": 1.687299022425563e-05, "loss": 0.5842, "step": 1623 }, { "epoch": 0.28, "learning_rate": 1.6868913462796298e-05, "loss": 0.6083, "step": 1624 }, { "epoch": 0.28, "learning_rate": 1.686483453878355e-05, "loss": 0.5816, "step": 1625 }, { "epoch": 0.28, "learning_rate": 1.686075345350156e-05, "loss": 0.5955, "step": 1626 }, { "epoch": 0.28, "learning_rate": 1.685667020823518e-05, "loss": 0.5816, "step": 1627 }, { "epoch": 0.28, "learning_rate": 1.685258480426995e-05, "loss": 0.5844, "step": 1628 }, { "epoch": 0.28, "learning_rate": 1.6848497242892086e-05, "loss": 0.5898, "step": 1629 }, { "epoch": 0.28, "learning_rate": 1.6844407525388486e-05, "loss": 0.5807, "step": 1630 }, { "epoch": 0.28, "learning_rate": 1.6840315653046715e-05, "loss": 0.5975, "step": 1631 }, { "epoch": 0.28, "learning_rate": 1.6836221627155033e-05, "loss": 0.5775, "step": 1632 }, { "epoch": 0.28, "learning_rate": 1.6832125449002363e-05, "loss": 0.5953, "step": 1633 }, { "epoch": 0.28, "learning_rate": 1.6828027119878316e-05, "loss": 0.568, "step": 1634 }, { "epoch": 0.28, "learning_rate": 1.6823926641073184e-05, "loss": 0.5847, "step": 1635 }, { "epoch": 0.28, "learning_rate": 1.6819824013877917e-05, "loss": 0.5907, "step": 1636 }, { "epoch": 0.28, "learning_rate": 1.681571923958416e-05, "loss": 0.5722, "step": 1637 }, { "epoch": 0.28, "learning_rate": 1.681161231948423e-05, "loss": 0.5711, "step": 1638 }, { "epoch": 0.28, "learning_rate": 1.6807503254871105e-05, "loss": 0.594, "step": 1639 }, { "epoch": 0.28, "learning_rate": 1.6803392047038457e-05, "loss": 0.5931, "step": 1640 }, { "epoch": 0.28, "learning_rate": 1.679927869728063e-05, "loss": 0.5861, "step": 1641 }, { "epoch": 0.28, "learning_rate": 1.6795163206892632e-05, "loss": 0.5983, "step": 1642 }, { "epoch": 0.28, "learning_rate": 1.6791045577170157e-05, "loss": 0.6027, "step": 1643 }, { "epoch": 0.28, "learning_rate": 1.6786925809409555e-05, "loss": 0.5699, "step": 1644 }, { "epoch": 0.28, "learning_rate": 1.678280390490787e-05, "loss": 0.5847, "step": 1645 }, { "epoch": 0.29, "learning_rate": 1.6778679864962807e-05, "loss": 0.5785, "step": 1646 }, { "epoch": 0.29, "learning_rate": 1.6774553690872745e-05, "loss": 0.5844, "step": 1647 }, { "epoch": 0.29, "learning_rate": 1.6770425383936734e-05, "loss": 0.5936, "step": 1648 }, { "epoch": 0.29, "learning_rate": 1.6766294945454502e-05, "loss": 0.59, "step": 1649 }, { "epoch": 0.29, "learning_rate": 1.6762162376726438e-05, "loss": 0.5716, "step": 1650 }, { "epoch": 0.29, "learning_rate": 1.6758027679053607e-05, "loss": 0.5754, "step": 1651 }, { "epoch": 0.29, "learning_rate": 1.675389085373775e-05, "loss": 0.5979, "step": 1652 }, { "epoch": 0.29, "learning_rate": 1.674975190208126e-05, "loss": 0.5786, "step": 1653 }, { "epoch": 0.29, "learning_rate": 1.674561082538722e-05, "loss": 0.6052, "step": 1654 }, { "epoch": 0.29, "learning_rate": 1.6741467624959372e-05, "loss": 0.5916, "step": 1655 }, { "epoch": 0.29, "learning_rate": 1.673732230210213e-05, "loss": 0.5778, "step": 1656 }, { "epoch": 0.29, "learning_rate": 1.6733174858120568e-05, "loss": 0.6019, "step": 1657 }, { "epoch": 0.29, "learning_rate": 1.672902529432044e-05, "loss": 0.5711, "step": 1658 }, { "epoch": 0.29, "learning_rate": 1.6724873612008155e-05, "loss": 0.5932, "step": 1659 }, { "epoch": 0.29, "learning_rate": 1.67207198124908e-05, "loss": 0.5828, "step": 1660 }, { "epoch": 0.29, "learning_rate": 1.6716563897076122e-05, "loss": 0.5747, "step": 1661 }, { "epoch": 0.29, "learning_rate": 1.6712405867072537e-05, "loss": 0.6034, "step": 1662 }, { "epoch": 0.29, "learning_rate": 1.670824572378913e-05, "loss": 0.5791, "step": 1663 }, { "epoch": 0.29, "learning_rate": 1.6704083468535636e-05, "loss": 0.5854, "step": 1664 }, { "epoch": 0.29, "learning_rate": 1.6699919102622474e-05, "loss": 0.5695, "step": 1665 }, { "epoch": 0.29, "learning_rate": 1.6695752627360718e-05, "loss": 0.5889, "step": 1666 }, { "epoch": 0.29, "learning_rate": 1.6691584044062102e-05, "loss": 0.5774, "step": 1667 }, { "epoch": 0.29, "learning_rate": 1.668741335403904e-05, "loss": 0.5787, "step": 1668 }, { "epoch": 0.29, "learning_rate": 1.668324055860459e-05, "loss": 0.5869, "step": 1669 }, { "epoch": 0.29, "learning_rate": 1.6679065659072486e-05, "loss": 0.5981, "step": 1670 }, { "epoch": 0.29, "learning_rate": 1.667488865675712e-05, "loss": 0.5888, "step": 1671 }, { "epoch": 0.29, "learning_rate": 1.667070955297354e-05, "loss": 0.5789, "step": 1672 }, { "epoch": 0.29, "learning_rate": 1.6666528349037467e-05, "loss": 0.6004, "step": 1673 }, { "epoch": 0.29, "learning_rate": 1.6662345046265273e-05, "loss": 0.5843, "step": 1674 }, { "epoch": 0.29, "learning_rate": 1.6658159645974e-05, "loss": 0.5757, "step": 1675 }, { "epoch": 0.29, "learning_rate": 1.6653972149481342e-05, "loss": 0.5704, "step": 1676 }, { "epoch": 0.29, "learning_rate": 1.664978255810566e-05, "loss": 0.5957, "step": 1677 }, { "epoch": 0.29, "learning_rate": 1.6645590873165968e-05, "loss": 0.566, "step": 1678 }, { "epoch": 0.29, "learning_rate": 1.6641397095981946e-05, "loss": 0.6014, "step": 1679 }, { "epoch": 0.29, "learning_rate": 1.6637201227873926e-05, "loss": 0.5742, "step": 1680 }, { "epoch": 0.29, "learning_rate": 1.6633003270162903e-05, "loss": 0.5975, "step": 1681 }, { "epoch": 0.29, "learning_rate": 1.6628803224170526e-05, "loss": 0.5761, "step": 1682 }, { "epoch": 0.29, "learning_rate": 1.662460109121911e-05, "loss": 0.5898, "step": 1683 }, { "epoch": 0.29, "learning_rate": 1.6620396872631608e-05, "loss": 0.5942, "step": 1684 }, { "epoch": 0.29, "learning_rate": 1.6616190569731655e-05, "loss": 0.6007, "step": 1685 }, { "epoch": 0.29, "learning_rate": 1.6611982183843524e-05, "loss": 0.5707, "step": 1686 }, { "epoch": 0.29, "learning_rate": 1.6607771716292153e-05, "loss": 0.5927, "step": 1687 }, { "epoch": 0.29, "learning_rate": 1.6603559168403126e-05, "loss": 0.5828, "step": 1688 }, { "epoch": 0.29, "learning_rate": 1.659934454150269e-05, "loss": 0.5876, "step": 1689 }, { "epoch": 0.29, "learning_rate": 1.6595127836917744e-05, "loss": 0.579, "step": 1690 }, { "epoch": 0.29, "learning_rate": 1.6590909055975846e-05, "loss": 0.5951, "step": 1691 }, { "epoch": 0.29, "learning_rate": 1.6586688200005193e-05, "loss": 0.5795, "step": 1692 }, { "epoch": 0.29, "learning_rate": 1.6582465270334655e-05, "loss": 0.588, "step": 1693 }, { "epoch": 0.29, "learning_rate": 1.6578240268293745e-05, "loss": 0.5729, "step": 1694 }, { "epoch": 0.29, "learning_rate": 1.657401319521262e-05, "loss": 0.5886, "step": 1695 }, { "epoch": 0.29, "learning_rate": 1.6569784052422105e-05, "loss": 0.5858, "step": 1696 }, { "epoch": 0.29, "learning_rate": 1.6565552841253663e-05, "loss": 0.6034, "step": 1697 }, { "epoch": 0.29, "learning_rate": 1.6561319563039428e-05, "loss": 0.5853, "step": 1698 }, { "epoch": 0.29, "learning_rate": 1.655708421911215e-05, "loss": 0.5777, "step": 1699 }, { "epoch": 0.29, "learning_rate": 1.655284681080527e-05, "loss": 0.5866, "step": 1700 }, { "epoch": 0.29, "learning_rate": 1.6548607339452853e-05, "loss": 0.6017, "step": 1701 }, { "epoch": 0.29, "learning_rate": 1.6544365806389614e-05, "loss": 0.5756, "step": 1702 }, { "epoch": 0.29, "learning_rate": 1.654012221295093e-05, "loss": 0.5753, "step": 1703 }, { "epoch": 0.3, "learning_rate": 1.653587656047282e-05, "loss": 0.5924, "step": 1704 }, { "epoch": 0.3, "learning_rate": 1.6531628850291947e-05, "loss": 0.5832, "step": 1705 }, { "epoch": 0.3, "learning_rate": 1.652737908374563e-05, "loss": 0.5851, "step": 1706 }, { "epoch": 0.3, "learning_rate": 1.6523127262171827e-05, "loss": 0.5646, "step": 1707 }, { "epoch": 0.3, "learning_rate": 1.6518873386909148e-05, "loss": 0.5968, "step": 1708 }, { "epoch": 0.3, "learning_rate": 1.6514617459296852e-05, "loss": 0.5919, "step": 1709 }, { "epoch": 0.3, "learning_rate": 1.651035948067484e-05, "loss": 0.59, "step": 1710 }, { "epoch": 0.3, "learning_rate": 1.6506099452383656e-05, "loss": 0.5807, "step": 1711 }, { "epoch": 0.3, "learning_rate": 1.65018373757645e-05, "loss": 0.5872, "step": 1712 }, { "epoch": 0.3, "learning_rate": 1.6497573252159203e-05, "loss": 0.5871, "step": 1713 }, { "epoch": 0.3, "learning_rate": 1.649330708291025e-05, "loss": 0.5942, "step": 1714 }, { "epoch": 0.3, "learning_rate": 1.648903886936077e-05, "loss": 0.5662, "step": 1715 }, { "epoch": 0.3, "learning_rate": 1.648476861285453e-05, "loss": 0.5922, "step": 1716 }, { "epoch": 0.3, "learning_rate": 1.6480496314735944e-05, "loss": 0.5698, "step": 1717 }, { "epoch": 0.3, "learning_rate": 1.6476221976350064e-05, "loss": 0.5996, "step": 1718 }, { "epoch": 0.3, "learning_rate": 1.6471945599042592e-05, "loss": 0.5764, "step": 1719 }, { "epoch": 0.3, "learning_rate": 1.6467667184159872e-05, "loss": 0.5709, "step": 1720 }, { "epoch": 0.3, "learning_rate": 1.646338673304888e-05, "loss": 0.5806, "step": 1721 }, { "epoch": 0.3, "learning_rate": 1.645910424705724e-05, "loss": 0.5817, "step": 1722 }, { "epoch": 0.3, "learning_rate": 1.6454819727533216e-05, "loss": 0.5808, "step": 1723 }, { "epoch": 0.3, "learning_rate": 1.6450533175825715e-05, "loss": 0.5929, "step": 1724 }, { "epoch": 0.3, "learning_rate": 1.6446244593284277e-05, "loss": 0.5819, "step": 1725 }, { "epoch": 0.3, "learning_rate": 1.644195398125908e-05, "loss": 0.574, "step": 1726 }, { "epoch": 0.3, "learning_rate": 1.6437661341100954e-05, "loss": 0.5765, "step": 1727 }, { "epoch": 0.3, "learning_rate": 1.6433366674161354e-05, "loss": 0.5867, "step": 1728 }, { "epoch": 0.3, "learning_rate": 1.6429069981792386e-05, "loss": 0.5791, "step": 1729 }, { "epoch": 0.3, "learning_rate": 1.6424771265346775e-05, "loss": 0.5993, "step": 1730 }, { "epoch": 0.3, "learning_rate": 1.64204705261779e-05, "loss": 0.5835, "step": 1731 }, { "epoch": 0.3, "learning_rate": 1.6416167765639773e-05, "loss": 0.5894, "step": 1732 }, { "epoch": 0.3, "learning_rate": 1.6411862985087036e-05, "loss": 0.5779, "step": 1733 }, { "epoch": 0.3, "learning_rate": 1.6407556185874975e-05, "loss": 0.5758, "step": 1734 }, { "epoch": 0.3, "learning_rate": 1.6403247369359502e-05, "loss": 0.6006, "step": 1735 }, { "epoch": 0.3, "learning_rate": 1.6398936536897182e-05, "loss": 0.5876, "step": 1736 }, { "epoch": 0.3, "learning_rate": 1.639462368984519e-05, "loss": 0.5889, "step": 1737 }, { "epoch": 0.3, "learning_rate": 1.6390308829561358e-05, "loss": 0.5848, "step": 1738 }, { "epoch": 0.3, "learning_rate": 1.6385991957404136e-05, "loss": 0.5664, "step": 1739 }, { "epoch": 0.3, "learning_rate": 1.6381673074732615e-05, "loss": 0.5868, "step": 1740 }, { "epoch": 0.3, "learning_rate": 1.6377352182906515e-05, "loss": 0.5885, "step": 1741 }, { "epoch": 0.3, "learning_rate": 1.637302928328619e-05, "loss": 0.5878, "step": 1742 }, { "epoch": 0.3, "learning_rate": 1.6368704377232637e-05, "loss": 0.5811, "step": 1743 }, { "epoch": 0.3, "learning_rate": 1.6364377466107465e-05, "loss": 0.5846, "step": 1744 }, { "epoch": 0.3, "learning_rate": 1.6360048551272926e-05, "loss": 0.5747, "step": 1745 }, { "epoch": 0.3, "learning_rate": 1.63557176340919e-05, "loss": 0.5993, "step": 1746 }, { "epoch": 0.3, "learning_rate": 1.6351384715927897e-05, "loss": 0.5729, "step": 1747 }, { "epoch": 0.3, "learning_rate": 1.6347049798145064e-05, "loss": 0.5994, "step": 1748 }, { "epoch": 0.3, "learning_rate": 1.6342712882108166e-05, "loss": 0.5765, "step": 1749 }, { "epoch": 0.3, "learning_rate": 1.6338373969182605e-05, "loss": 0.5813, "step": 1750 }, { "epoch": 0.3, "learning_rate": 1.633403306073441e-05, "loss": 0.5635, "step": 1751 }, { "epoch": 0.3, "learning_rate": 1.6329690158130233e-05, "loss": 0.5895, "step": 1752 }, { "epoch": 0.3, "learning_rate": 1.6325345262737363e-05, "loss": 0.5821, "step": 1753 }, { "epoch": 0.3, "learning_rate": 1.6320998375923708e-05, "loss": 0.584, "step": 1754 }, { "epoch": 0.3, "learning_rate": 1.6316649499057816e-05, "loss": 0.548, "step": 1755 }, { "epoch": 0.3, "learning_rate": 1.6312298633508844e-05, "loss": 0.5839, "step": 1756 }, { "epoch": 0.3, "learning_rate": 1.6307945780646584e-05, "loss": 0.5771, "step": 1757 }, { "epoch": 0.3, "learning_rate": 1.6303590941841457e-05, "loss": 0.5829, "step": 1758 }, { "epoch": 0.3, "learning_rate": 1.62992341184645e-05, "loss": 0.5852, "step": 1759 }, { "epoch": 0.3, "learning_rate": 1.6294875311887384e-05, "loss": 0.5861, "step": 1760 }, { "epoch": 0.31, "learning_rate": 1.6290514523482405e-05, "loss": 0.5832, "step": 1761 }, { "epoch": 0.31, "learning_rate": 1.628615175462247e-05, "loss": 0.6013, "step": 1762 }, { "epoch": 0.31, "learning_rate": 1.6281787006681116e-05, "loss": 0.5657, "step": 1763 }, { "epoch": 0.31, "learning_rate": 1.627742028103252e-05, "loss": 0.5795, "step": 1764 }, { "epoch": 0.31, "learning_rate": 1.6273051579051453e-05, "loss": 0.5773, "step": 1765 }, { "epoch": 0.31, "learning_rate": 1.626868090211333e-05, "loss": 0.5873, "step": 1766 }, { "epoch": 0.31, "learning_rate": 1.626430825159417e-05, "loss": 0.5834, "step": 1767 }, { "epoch": 0.31, "learning_rate": 1.625993362887063e-05, "loss": 0.5887, "step": 1768 }, { "epoch": 0.31, "learning_rate": 1.625555703531998e-05, "loss": 0.5751, "step": 1769 }, { "epoch": 0.31, "learning_rate": 1.6251178472320114e-05, "loss": 0.5614, "step": 1770 }, { "epoch": 0.31, "learning_rate": 1.624679794124954e-05, "loss": 0.5877, "step": 1771 }, { "epoch": 0.31, "learning_rate": 1.6242415443487393e-05, "loss": 0.5848, "step": 1772 }, { "epoch": 0.31, "learning_rate": 1.6238030980413418e-05, "loss": 0.5743, "step": 1773 }, { "epoch": 0.31, "learning_rate": 1.6233644553407987e-05, "loss": 0.5881, "step": 1774 }, { "epoch": 0.31, "learning_rate": 1.6229256163852086e-05, "loss": 0.5832, "step": 1775 }, { "epoch": 0.31, "learning_rate": 1.622486581312732e-05, "loss": 0.5792, "step": 1776 }, { "epoch": 0.31, "learning_rate": 1.6220473502615918e-05, "loss": 0.5871, "step": 1777 }, { "epoch": 0.31, "learning_rate": 1.621607923370071e-05, "loss": 0.5939, "step": 1778 }, { "epoch": 0.31, "learning_rate": 1.6211683007765157e-05, "loss": 0.5638, "step": 1779 }, { "epoch": 0.31, "learning_rate": 1.6207284826193334e-05, "loss": 0.5887, "step": 1780 }, { "epoch": 0.31, "learning_rate": 1.6202884690369924e-05, "loss": 0.5805, "step": 1781 }, { "epoch": 0.31, "learning_rate": 1.6198482601680233e-05, "loss": 0.5813, "step": 1782 }, { "epoch": 0.31, "learning_rate": 1.619407856151018e-05, "loss": 0.583, "step": 1783 }, { "epoch": 0.31, "learning_rate": 1.61896725712463e-05, "loss": 0.5855, "step": 1784 }, { "epoch": 0.31, "learning_rate": 1.618526463227573e-05, "loss": 0.586, "step": 1785 }, { "epoch": 0.31, "learning_rate": 1.618085474598624e-05, "loss": 0.5876, "step": 1786 }, { "epoch": 0.31, "learning_rate": 1.6176442913766192e-05, "loss": 0.5841, "step": 1787 }, { "epoch": 0.31, "learning_rate": 1.6172029137004584e-05, "loss": 0.6121, "step": 1788 }, { "epoch": 0.31, "learning_rate": 1.6167613417091007e-05, "loss": 0.5706, "step": 1789 }, { "epoch": 0.31, "learning_rate": 1.6163195755415676e-05, "loss": 0.5976, "step": 1790 }, { "epoch": 0.31, "learning_rate": 1.6158776153369406e-05, "loss": 0.5923, "step": 1791 }, { "epoch": 0.31, "learning_rate": 1.615435461234363e-05, "loss": 0.5826, "step": 1792 }, { "epoch": 0.31, "learning_rate": 1.6149931133730396e-05, "loss": 0.5893, "step": 1793 }, { "epoch": 0.31, "learning_rate": 1.6145505718922348e-05, "loss": 0.587, "step": 1794 }, { "epoch": 0.31, "learning_rate": 1.6141078369312752e-05, "loss": 0.5842, "step": 1795 }, { "epoch": 0.31, "learning_rate": 1.613664908629548e-05, "loss": 0.586, "step": 1796 }, { "epoch": 0.31, "learning_rate": 1.6132217871265012e-05, "loss": 0.575, "step": 1797 }, { "epoch": 0.31, "learning_rate": 1.6127784725616434e-05, "loss": 0.5857, "step": 1798 }, { "epoch": 0.31, "learning_rate": 1.6123349650745445e-05, "loss": 0.6041, "step": 1799 }, { "epoch": 0.31, "learning_rate": 1.6118912648048345e-05, "loss": 0.5925, "step": 1800 }, { "epoch": 0.31, "learning_rate": 1.6114473718922042e-05, "loss": 0.5838, "step": 1801 }, { "epoch": 0.31, "learning_rate": 1.611003286476406e-05, "loss": 0.5844, "step": 1802 }, { "epoch": 0.31, "learning_rate": 1.610559008697252e-05, "loss": 0.571, "step": 1803 }, { "epoch": 0.31, "learning_rate": 1.6101145386946145e-05, "loss": 0.577, "step": 1804 }, { "epoch": 0.31, "learning_rate": 1.6096698766084274e-05, "loss": 0.5802, "step": 1805 }, { "epoch": 0.31, "learning_rate": 1.6092250225786842e-05, "loss": 0.5834, "step": 1806 }, { "epoch": 0.31, "learning_rate": 1.6087799767454394e-05, "loss": 0.5719, "step": 1807 }, { "epoch": 0.31, "learning_rate": 1.6083347392488072e-05, "loss": 0.5812, "step": 1808 }, { "epoch": 0.31, "learning_rate": 1.6078893102289634e-05, "loss": 0.5879, "step": 1809 }, { "epoch": 0.31, "learning_rate": 1.6074436898261424e-05, "loss": 0.5858, "step": 1810 }, { "epoch": 0.31, "learning_rate": 1.6069978781806403e-05, "loss": 0.5685, "step": 1811 }, { "epoch": 0.31, "learning_rate": 1.6065518754328125e-05, "loss": 0.5813, "step": 1812 }, { "epoch": 0.31, "learning_rate": 1.6061056817230754e-05, "loss": 0.5805, "step": 1813 }, { "epoch": 0.31, "learning_rate": 1.6056592971919047e-05, "loss": 0.5715, "step": 1814 }, { "epoch": 0.31, "learning_rate": 1.6052127219798367e-05, "loss": 0.5906, "step": 1815 }, { "epoch": 0.31, "learning_rate": 1.604765956227467e-05, "loss": 0.5826, "step": 1816 }, { "epoch": 0.31, "learning_rate": 1.6043190000754525e-05, "loss": 0.582, "step": 1817 }, { "epoch": 0.31, "learning_rate": 1.6038718536645088e-05, "loss": 0.5804, "step": 1818 }, { "epoch": 0.32, "learning_rate": 1.6034245171354118e-05, "loss": 0.5781, "step": 1819 }, { "epoch": 0.32, "learning_rate": 1.6029769906289982e-05, "loss": 0.5764, "step": 1820 }, { "epoch": 0.32, "learning_rate": 1.6025292742861622e-05, "loss": 0.5868, "step": 1821 }, { "epoch": 0.32, "learning_rate": 1.6020813682478603e-05, "loss": 0.5773, "step": 1822 }, { "epoch": 0.32, "learning_rate": 1.601633272655107e-05, "loss": 0.5854, "step": 1823 }, { "epoch": 0.32, "learning_rate": 1.6011849876489777e-05, "loss": 0.5731, "step": 1824 }, { "epoch": 0.32, "learning_rate": 1.6007365133706065e-05, "loss": 0.5882, "step": 1825 }, { "epoch": 0.32, "learning_rate": 1.6002878499611876e-05, "loss": 0.5556, "step": 1826 }, { "epoch": 0.32, "learning_rate": 1.5998389975619747e-05, "loss": 0.5883, "step": 1827 }, { "epoch": 0.32, "learning_rate": 1.5993899563142804e-05, "loss": 0.5961, "step": 1828 }, { "epoch": 0.32, "learning_rate": 1.598940726359477e-05, "loss": 0.592, "step": 1829 }, { "epoch": 0.32, "learning_rate": 1.5984913078389974e-05, "loss": 0.5802, "step": 1830 }, { "epoch": 0.32, "learning_rate": 1.5980417008943323e-05, "loss": 0.5681, "step": 1831 }, { "epoch": 0.32, "learning_rate": 1.5975919056670325e-05, "loss": 0.5745, "step": 1832 }, { "epoch": 0.32, "learning_rate": 1.5971419222987078e-05, "loss": 0.5835, "step": 1833 }, { "epoch": 0.32, "learning_rate": 1.5966917509310277e-05, "loss": 0.5661, "step": 1834 }, { "epoch": 0.32, "learning_rate": 1.59624139170572e-05, "loss": 0.5799, "step": 1835 }, { "epoch": 0.32, "learning_rate": 1.5957908447645722e-05, "loss": 0.5793, "step": 1836 }, { "epoch": 0.32, "learning_rate": 1.5953401102494315e-05, "loss": 0.5752, "step": 1837 }, { "epoch": 0.32, "learning_rate": 1.594889188302203e-05, "loss": 0.593, "step": 1838 }, { "epoch": 0.32, "learning_rate": 1.594438079064851e-05, "loss": 0.5771, "step": 1839 }, { "epoch": 0.32, "learning_rate": 1.5939867826794e-05, "loss": 0.5935, "step": 1840 }, { "epoch": 0.32, "learning_rate": 1.593535299287932e-05, "loss": 0.5859, "step": 1841 }, { "epoch": 0.32, "learning_rate": 1.5930836290325884e-05, "loss": 0.6017, "step": 1842 }, { "epoch": 0.32, "learning_rate": 1.59263177205557e-05, "loss": 0.6002, "step": 1843 }, { "epoch": 0.32, "learning_rate": 1.5921797284991342e-05, "loss": 0.5762, "step": 1844 }, { "epoch": 0.32, "learning_rate": 1.5917274985056007e-05, "loss": 0.6055, "step": 1845 }, { "epoch": 0.32, "learning_rate": 1.5912750822173446e-05, "loss": 0.5924, "step": 1846 }, { "epoch": 0.32, "learning_rate": 1.590822479776802e-05, "loss": 0.5957, "step": 1847 }, { "epoch": 0.32, "learning_rate": 1.590369691326466e-05, "loss": 0.5909, "step": 1848 }, { "epoch": 0.32, "learning_rate": 1.5899167170088886e-05, "loss": 0.5753, "step": 1849 }, { "epoch": 0.32, "learning_rate": 1.5894635569666816e-05, "loss": 0.5817, "step": 1850 }, { "epoch": 0.32, "learning_rate": 1.589010211342513e-05, "loss": 0.5811, "step": 1851 }, { "epoch": 0.32, "learning_rate": 1.5885566802791115e-05, "loss": 0.5734, "step": 1852 }, { "epoch": 0.32, "learning_rate": 1.5881029639192626e-05, "loss": 0.5908, "step": 1853 }, { "epoch": 0.32, "learning_rate": 1.5876490624058113e-05, "loss": 0.5711, "step": 1854 }, { "epoch": 0.32, "learning_rate": 1.5871949758816594e-05, "loss": 0.5897, "step": 1855 }, { "epoch": 0.32, "learning_rate": 1.586740704489769e-05, "loss": 0.5639, "step": 1856 }, { "epoch": 0.32, "learning_rate": 1.5862862483731574e-05, "loss": 0.5973, "step": 1857 }, { "epoch": 0.32, "learning_rate": 1.585831607674904e-05, "loss": 0.5806, "step": 1858 }, { "epoch": 0.32, "learning_rate": 1.5853767825381434e-05, "loss": 0.5887, "step": 1859 }, { "epoch": 0.32, "learning_rate": 1.5849217731060686e-05, "loss": 0.5707, "step": 1860 }, { "epoch": 0.32, "learning_rate": 1.5844665795219314e-05, "loss": 0.5943, "step": 1861 }, { "epoch": 0.32, "learning_rate": 1.584011201929042e-05, "loss": 0.5773, "step": 1862 }, { "epoch": 0.32, "learning_rate": 1.5835556404707668e-05, "loss": 0.5969, "step": 1863 }, { "epoch": 0.32, "learning_rate": 1.5830998952905316e-05, "loss": 0.5591, "step": 1864 }, { "epoch": 0.32, "learning_rate": 1.582643966531819e-05, "loss": 0.5898, "step": 1865 }, { "epoch": 0.32, "learning_rate": 1.5821878543381707e-05, "loss": 0.5895, "step": 1866 }, { "epoch": 0.32, "learning_rate": 1.581731558853185e-05, "loss": 0.5745, "step": 1867 }, { "epoch": 0.32, "learning_rate": 1.5812750802205187e-05, "loss": 0.5759, "step": 1868 }, { "epoch": 0.32, "learning_rate": 1.5808184185838854e-05, "loss": 0.5974, "step": 1869 }, { "epoch": 0.32, "learning_rate": 1.580361574087057e-05, "loss": 0.5768, "step": 1870 }, { "epoch": 0.32, "learning_rate": 1.5799045468738626e-05, "loss": 0.5787, "step": 1871 }, { "epoch": 0.32, "learning_rate": 1.5794473370881886e-05, "loss": 0.5768, "step": 1872 }, { "epoch": 0.32, "learning_rate": 1.57898994487398e-05, "loss": 0.5867, "step": 1873 }, { "epoch": 0.32, "learning_rate": 1.5785323703752383e-05, "loss": 0.5814, "step": 1874 }, { "epoch": 0.32, "learning_rate": 1.578074613736022e-05, "loss": 0.5805, "step": 1875 }, { "epoch": 0.32, "learning_rate": 1.5776166751004483e-05, "loss": 0.5887, "step": 1876 }, { "epoch": 0.33, "learning_rate": 1.5771585546126907e-05, "loss": 0.5836, "step": 1877 }, { "epoch": 0.33, "learning_rate": 1.5767002524169797e-05, "loss": 0.5835, "step": 1878 }, { "epoch": 0.33, "learning_rate": 1.576241768657604e-05, "loss": 0.5759, "step": 1879 }, { "epoch": 0.33, "learning_rate": 1.5757831034789085e-05, "loss": 0.5901, "step": 1880 }, { "epoch": 0.33, "learning_rate": 1.575324257025296e-05, "loss": 0.5865, "step": 1881 }, { "epoch": 0.33, "learning_rate": 1.574865229441226e-05, "loss": 0.5824, "step": 1882 }, { "epoch": 0.33, "learning_rate": 1.5744060208712148e-05, "loss": 0.5843, "step": 1883 }, { "epoch": 0.33, "learning_rate": 1.573946631459836e-05, "loss": 0.5671, "step": 1884 }, { "epoch": 0.33, "learning_rate": 1.5734870613517204e-05, "loss": 0.6187, "step": 1885 }, { "epoch": 0.33, "learning_rate": 1.573027310691555e-05, "loss": 0.5855, "step": 1886 }, { "epoch": 0.33, "learning_rate": 1.572567379624084e-05, "loss": 0.605, "step": 1887 }, { "epoch": 0.33, "learning_rate": 1.5721072682941084e-05, "loss": 0.5754, "step": 1888 }, { "epoch": 0.33, "learning_rate": 1.5716469768464864e-05, "loss": 0.5971, "step": 1889 }, { "epoch": 0.33, "learning_rate": 1.571186505426132e-05, "loss": 0.5763, "step": 1890 }, { "epoch": 0.33, "learning_rate": 1.5707258541780162e-05, "loss": 0.5941, "step": 1891 }, { "epoch": 0.33, "learning_rate": 1.570265023247167e-05, "loss": 0.5994, "step": 1892 }, { "epoch": 0.33, "learning_rate": 1.5698040127786688e-05, "loss": 0.5673, "step": 1893 }, { "epoch": 0.33, "learning_rate": 1.569342822917662e-05, "loss": 0.6122, "step": 1894 }, { "epoch": 0.33, "learning_rate": 1.568881453809344e-05, "loss": 0.5786, "step": 1895 }, { "epoch": 0.33, "learning_rate": 1.5684199055989693e-05, "loss": 0.5661, "step": 1896 }, { "epoch": 0.33, "learning_rate": 1.567958178431847e-05, "loss": 0.576, "step": 1897 }, { "epoch": 0.33, "learning_rate": 1.567496272453344e-05, "loss": 0.5701, "step": 1898 }, { "epoch": 0.33, "learning_rate": 1.5670341878088834e-05, "loss": 0.5825, "step": 1899 }, { "epoch": 0.33, "learning_rate": 1.5665719246439435e-05, "loss": 0.5936, "step": 1900 }, { "epoch": 0.33, "learning_rate": 1.56610948310406e-05, "loss": 0.5757, "step": 1901 }, { "epoch": 0.33, "learning_rate": 1.5656468633348243e-05, "loss": 0.5887, "step": 1902 }, { "epoch": 0.33, "learning_rate": 1.5651840654818834e-05, "loss": 0.5801, "step": 1903 }, { "epoch": 0.33, "learning_rate": 1.5647210896909414e-05, "loss": 0.5902, "step": 1904 }, { "epoch": 0.33, "learning_rate": 1.5642579361077576e-05, "loss": 0.5987, "step": 1905 }, { "epoch": 0.33, "learning_rate": 1.563794604878148e-05, "loss": 0.5818, "step": 1906 }, { "epoch": 0.33, "learning_rate": 1.563331096147983e-05, "loss": 0.5873, "step": 1907 }, { "epoch": 0.33, "learning_rate": 1.562867410063191e-05, "loss": 0.5739, "step": 1908 }, { "epoch": 0.33, "learning_rate": 1.5624035467697547e-05, "loss": 0.5899, "step": 1909 }, { "epoch": 0.33, "learning_rate": 1.561939506413713e-05, "loss": 0.5972, "step": 1910 }, { "epoch": 0.33, "learning_rate": 1.5614752891411613e-05, "loss": 0.5843, "step": 1911 }, { "epoch": 0.33, "learning_rate": 1.5610108950982494e-05, "loss": 0.5813, "step": 1912 }, { "epoch": 0.33, "learning_rate": 1.5605463244311834e-05, "loss": 0.5914, "step": 1913 }, { "epoch": 0.33, "learning_rate": 1.560081577286225e-05, "loss": 0.5735, "step": 1914 }, { "epoch": 0.33, "learning_rate": 1.559616653809692e-05, "loss": 0.5838, "step": 1915 }, { "epoch": 0.33, "learning_rate": 1.5591515541479566e-05, "loss": 0.5919, "step": 1916 }, { "epoch": 0.33, "learning_rate": 1.558686278447447e-05, "loss": 0.5744, "step": 1917 }, { "epoch": 0.33, "learning_rate": 1.5582208268546473e-05, "loss": 0.5834, "step": 1918 }, { "epoch": 0.33, "learning_rate": 1.557755199516096e-05, "loss": 0.583, "step": 1919 }, { "epoch": 0.33, "learning_rate": 1.557289396578388e-05, "loss": 0.588, "step": 1920 }, { "epoch": 0.33, "learning_rate": 1.5568234181881725e-05, "loss": 0.5889, "step": 1921 }, { "epoch": 0.33, "learning_rate": 1.5563572644921546e-05, "loss": 0.5886, "step": 1922 }, { "epoch": 0.33, "learning_rate": 1.5558909356370944e-05, "loss": 0.5738, "step": 1923 }, { "epoch": 0.33, "learning_rate": 1.5554244317698072e-05, "loss": 0.5531, "step": 1924 }, { "epoch": 0.33, "learning_rate": 1.554957753037163e-05, "loss": 0.5823, "step": 1925 }, { "epoch": 0.33, "learning_rate": 1.554490899586087e-05, "loss": 0.5918, "step": 1926 }, { "epoch": 0.33, "learning_rate": 1.5540238715635604e-05, "loss": 0.5762, "step": 1927 }, { "epoch": 0.33, "learning_rate": 1.553556669116618e-05, "loss": 0.5784, "step": 1928 }, { "epoch": 0.33, "learning_rate": 1.5530892923923504e-05, "loss": 0.5746, "step": 1929 }, { "epoch": 0.33, "learning_rate": 1.552621741537902e-05, "loss": 0.5835, "step": 1930 }, { "epoch": 0.33, "learning_rate": 1.5521540167004736e-05, "loss": 0.5892, "step": 1931 }, { "epoch": 0.33, "learning_rate": 1.5516861180273196e-05, "loss": 0.593, "step": 1932 }, { "epoch": 0.33, "learning_rate": 1.551218045665749e-05, "loss": 0.5745, "step": 1933 }, { "epoch": 0.34, "learning_rate": 1.5507497997631267e-05, "loss": 0.5801, "step": 1934 }, { "epoch": 0.34, "learning_rate": 1.5502813804668712e-05, "loss": 0.5916, "step": 1935 }, { "epoch": 0.34, "learning_rate": 1.5498127879244554e-05, "loss": 0.5921, "step": 1936 }, { "epoch": 0.34, "learning_rate": 1.5493440222834073e-05, "loss": 0.5835, "step": 1937 }, { "epoch": 0.34, "learning_rate": 1.54887508369131e-05, "loss": 0.5827, "step": 1938 }, { "epoch": 0.34, "learning_rate": 1.5484059722958e-05, "loss": 0.5944, "step": 1939 }, { "epoch": 0.34, "learning_rate": 1.5479366882445684e-05, "loss": 0.5708, "step": 1940 }, { "epoch": 0.34, "learning_rate": 1.5474672316853605e-05, "loss": 0.5884, "step": 1941 }, { "epoch": 0.34, "learning_rate": 1.546997602765977e-05, "loss": 0.5798, "step": 1942 }, { "epoch": 0.34, "learning_rate": 1.5465278016342717e-05, "loss": 0.5844, "step": 1943 }, { "epoch": 0.34, "learning_rate": 1.5460578284381526e-05, "loss": 0.6016, "step": 1944 }, { "epoch": 0.34, "learning_rate": 1.545587683325583e-05, "loss": 0.5881, "step": 1945 }, { "epoch": 0.34, "learning_rate": 1.5451173664445795e-05, "loss": 0.5914, "step": 1946 }, { "epoch": 0.34, "learning_rate": 1.5446468779432123e-05, "loss": 0.5729, "step": 1947 }, { "epoch": 0.34, "learning_rate": 1.5441762179696066e-05, "loss": 0.5912, "step": 1948 }, { "epoch": 0.34, "learning_rate": 1.5437053866719412e-05, "loss": 0.5778, "step": 1949 }, { "epoch": 0.34, "learning_rate": 1.543234384198449e-05, "loss": 0.5846, "step": 1950 }, { "epoch": 0.34, "learning_rate": 1.5427632106974165e-05, "loss": 0.5846, "step": 1951 }, { "epoch": 0.34, "learning_rate": 1.542291866317184e-05, "loss": 0.5846, "step": 1952 }, { "epoch": 0.34, "learning_rate": 1.5418203512061455e-05, "loss": 0.584, "step": 1953 }, { "epoch": 0.34, "learning_rate": 1.5413486655127498e-05, "loss": 0.5844, "step": 1954 }, { "epoch": 0.34, "learning_rate": 1.540876809385498e-05, "loss": 0.5626, "step": 1955 }, { "epoch": 0.34, "learning_rate": 1.540404782972946e-05, "loss": 0.5849, "step": 1956 }, { "epoch": 0.34, "learning_rate": 1.5399325864237022e-05, "loss": 0.5803, "step": 1957 }, { "epoch": 0.34, "learning_rate": 1.5394602198864297e-05, "loss": 0.5839, "step": 1958 }, { "epoch": 0.34, "learning_rate": 1.538987683509844e-05, "loss": 0.5834, "step": 1959 }, { "epoch": 0.34, "learning_rate": 1.5385149774427153e-05, "loss": 0.5798, "step": 1960 }, { "epoch": 0.34, "learning_rate": 1.5380421018338663e-05, "loss": 0.5828, "step": 1961 }, { "epoch": 0.34, "learning_rate": 1.537569056832173e-05, "loss": 0.5648, "step": 1962 }, { "epoch": 0.34, "learning_rate": 1.537095842586566e-05, "loss": 0.5642, "step": 1963 }, { "epoch": 0.34, "learning_rate": 1.536622459246027e-05, "loss": 0.5901, "step": 1964 }, { "epoch": 0.34, "learning_rate": 1.5361489069595932e-05, "loss": 0.5806, "step": 1965 }, { "epoch": 0.34, "learning_rate": 1.5356751858763535e-05, "loss": 0.5725, "step": 1966 }, { "epoch": 0.34, "learning_rate": 1.535201296145451e-05, "loss": 0.5764, "step": 1967 }, { "epoch": 0.34, "learning_rate": 1.5347272379160805e-05, "loss": 0.5759, "step": 1968 }, { "epoch": 0.34, "learning_rate": 1.5342530113374906e-05, "loss": 0.5787, "step": 1969 }, { "epoch": 0.34, "learning_rate": 1.5337786165589845e-05, "loss": 0.5836, "step": 1970 }, { "epoch": 0.34, "learning_rate": 1.533304053729915e-05, "loss": 0.5729, "step": 1971 }, { "epoch": 0.34, "learning_rate": 1.5328293229996907e-05, "loss": 0.5811, "step": 1972 }, { "epoch": 0.34, "learning_rate": 1.532354424517772e-05, "loss": 0.5945, "step": 1973 }, { "epoch": 0.34, "learning_rate": 1.5318793584336718e-05, "loss": 0.5719, "step": 1974 }, { "epoch": 0.34, "learning_rate": 1.5314041248969558e-05, "loss": 0.5596, "step": 1975 }, { "epoch": 0.34, "learning_rate": 1.530928724057243e-05, "loss": 0.5773, "step": 1976 }, { "epoch": 0.34, "learning_rate": 1.5304531560642052e-05, "loss": 0.5837, "step": 1977 }, { "epoch": 0.34, "learning_rate": 1.529977421067566e-05, "loss": 0.5764, "step": 1978 }, { "epoch": 0.34, "learning_rate": 1.5295015192171016e-05, "loss": 0.5741, "step": 1979 }, { "epoch": 0.34, "learning_rate": 1.5290254506626417e-05, "loss": 0.5798, "step": 1980 }, { "epoch": 0.34, "learning_rate": 1.5285492155540676e-05, "loss": 0.5773, "step": 1981 }, { "epoch": 0.34, "learning_rate": 1.5280728140413134e-05, "loss": 0.5746, "step": 1982 }, { "epoch": 0.34, "learning_rate": 1.5275962462743657e-05, "loss": 0.5926, "step": 1983 }, { "epoch": 0.34, "learning_rate": 1.527119512403263e-05, "loss": 0.5882, "step": 1984 }, { "epoch": 0.34, "learning_rate": 1.5266426125780965e-05, "loss": 0.5774, "step": 1985 }, { "epoch": 0.34, "learning_rate": 1.5261655469490095e-05, "loss": 0.5778, "step": 1986 }, { "epoch": 0.34, "learning_rate": 1.5256883156661972e-05, "loss": 0.568, "step": 1987 }, { "epoch": 0.34, "learning_rate": 1.5252109188799077e-05, "loss": 0.5823, "step": 1988 }, { "epoch": 0.34, "learning_rate": 1.5247333567404407e-05, "loss": 0.5916, "step": 1989 }, { "epoch": 0.34, "learning_rate": 1.5242556293981477e-05, "loss": 0.6029, "step": 1990 }, { "epoch": 0.34, "learning_rate": 1.5237777370034326e-05, "loss": 0.5731, "step": 1991 }, { "epoch": 0.35, "learning_rate": 1.5232996797067513e-05, "loss": 0.5859, "step": 1992 }, { "epoch": 0.35, "learning_rate": 1.5228214576586118e-05, "loss": 0.5772, "step": 1993 }, { "epoch": 0.35, "learning_rate": 1.522343071009573e-05, "loss": 0.5994, "step": 1994 }, { "epoch": 0.35, "learning_rate": 1.521864519910247e-05, "loss": 0.5812, "step": 1995 }, { "epoch": 0.35, "learning_rate": 1.5213858045112968e-05, "loss": 0.5771, "step": 1996 }, { "epoch": 0.35, "learning_rate": 1.5209069249634367e-05, "loss": 0.5774, "step": 1997 }, { "epoch": 0.35, "learning_rate": 1.520427881417434e-05, "loss": 0.5883, "step": 1998 }, { "epoch": 0.35, "learning_rate": 1.5199486740241067e-05, "loss": 0.5812, "step": 1999 }, { "epoch": 0.35, "learning_rate": 1.5194693029343249e-05, "loss": 0.583, "step": 2000 }, { "epoch": 0.35, "learning_rate": 1.518989768299009e-05, "loss": 0.5843, "step": 2001 }, { "epoch": 0.35, "learning_rate": 1.518510070269133e-05, "loss": 0.5819, "step": 2002 }, { "epoch": 0.35, "learning_rate": 1.5180302089957201e-05, "loss": 0.5699, "step": 2003 }, { "epoch": 0.35, "learning_rate": 1.5175501846298468e-05, "loss": 0.5818, "step": 2004 }, { "epoch": 0.35, "learning_rate": 1.5170699973226395e-05, "loss": 0.5796, "step": 2005 }, { "epoch": 0.35, "learning_rate": 1.5165896472252768e-05, "loss": 0.592, "step": 2006 }, { "epoch": 0.35, "learning_rate": 1.5161091344889885e-05, "loss": 0.5904, "step": 2007 }, { "epoch": 0.35, "learning_rate": 1.5156284592650548e-05, "loss": 0.57, "step": 2008 }, { "epoch": 0.35, "learning_rate": 1.5151476217048083e-05, "loss": 0.5778, "step": 2009 }, { "epoch": 0.35, "learning_rate": 1.5146666219596311e-05, "loss": 0.6012, "step": 2010 }, { "epoch": 0.35, "learning_rate": 1.5141854601809583e-05, "loss": 0.5652, "step": 2011 }, { "epoch": 0.35, "learning_rate": 1.513704136520274e-05, "loss": 0.5831, "step": 2012 }, { "epoch": 0.35, "learning_rate": 1.513222651129115e-05, "loss": 0.5636, "step": 2013 }, { "epoch": 0.35, "learning_rate": 1.5127410041590684e-05, "loss": 0.584, "step": 2014 }, { "epoch": 0.35, "learning_rate": 1.5122591957617713e-05, "loss": 0.5868, "step": 2015 }, { "epoch": 0.35, "learning_rate": 1.5117772260889132e-05, "loss": 0.5834, "step": 2016 }, { "epoch": 0.35, "learning_rate": 1.5112950952922329e-05, "loss": 0.5822, "step": 2017 }, { "epoch": 0.35, "learning_rate": 1.5108128035235207e-05, "loss": 0.5873, "step": 2018 }, { "epoch": 0.35, "learning_rate": 1.510330350934618e-05, "loss": 0.5746, "step": 2019 }, { "epoch": 0.35, "learning_rate": 1.5098477376774156e-05, "loss": 0.5694, "step": 2020 }, { "epoch": 0.35, "learning_rate": 1.5093649639038557e-05, "loss": 0.5778, "step": 2021 }, { "epoch": 0.35, "learning_rate": 1.5088820297659314e-05, "loss": 0.555, "step": 2022 }, { "epoch": 0.35, "learning_rate": 1.5083989354156855e-05, "loss": 0.6032, "step": 2023 }, { "epoch": 0.35, "learning_rate": 1.5079156810052111e-05, "loss": 0.5603, "step": 2024 }, { "epoch": 0.35, "learning_rate": 1.507432266686653e-05, "loss": 0.5862, "step": 2025 }, { "epoch": 0.35, "learning_rate": 1.5069486926122045e-05, "loss": 0.5962, "step": 2026 }, { "epoch": 0.35, "learning_rate": 1.5064649589341112e-05, "loss": 0.5789, "step": 2027 }, { "epoch": 0.35, "learning_rate": 1.505981065804667e-05, "loss": 0.5884, "step": 2028 }, { "epoch": 0.35, "learning_rate": 1.5054970133762173e-05, "loss": 0.5871, "step": 2029 }, { "epoch": 0.35, "learning_rate": 1.5050128018011576e-05, "loss": 0.5829, "step": 2030 }, { "epoch": 0.35, "learning_rate": 1.5045284312319326e-05, "loss": 0.5776, "step": 2031 }, { "epoch": 0.35, "learning_rate": 1.5040439018210377e-05, "loss": 0.5818, "step": 2032 }, { "epoch": 0.35, "learning_rate": 1.5035592137210188e-05, "loss": 0.5667, "step": 2033 }, { "epoch": 0.35, "learning_rate": 1.5030743670844703e-05, "loss": 0.5959, "step": 2034 }, { "epoch": 0.35, "learning_rate": 1.5025893620640384e-05, "loss": 0.589, "step": 2035 }, { "epoch": 0.35, "learning_rate": 1.5021041988124173e-05, "loss": 0.573, "step": 2036 }, { "epoch": 0.35, "learning_rate": 1.5016188774823528e-05, "loss": 0.5785, "step": 2037 }, { "epoch": 0.35, "learning_rate": 1.5011333982266389e-05, "loss": 0.5869, "step": 2038 }, { "epoch": 0.35, "learning_rate": 1.5006477611981201e-05, "loss": 0.5914, "step": 2039 }, { "epoch": 0.35, "learning_rate": 1.5001619665496904e-05, "loss": 0.5898, "step": 2040 }, { "epoch": 0.35, "learning_rate": 1.4996760144342934e-05, "loss": 0.5784, "step": 2041 }, { "epoch": 0.35, "learning_rate": 1.4991899050049225e-05, "loss": 0.5771, "step": 2042 }, { "epoch": 0.35, "learning_rate": 1.498703638414621e-05, "loss": 0.5752, "step": 2043 }, { "epoch": 0.35, "learning_rate": 1.4982172148164804e-05, "loss": 0.5654, "step": 2044 }, { "epoch": 0.35, "learning_rate": 1.4977306343636425e-05, "loss": 0.5873, "step": 2045 }, { "epoch": 0.35, "learning_rate": 1.4972438972092986e-05, "loss": 0.5746, "step": 2046 }, { "epoch": 0.35, "learning_rate": 1.496757003506689e-05, "loss": 0.6029, "step": 2047 }, { "epoch": 0.35, "learning_rate": 1.4962699534091037e-05, "loss": 0.5858, "step": 2048 }, { "epoch": 0.35, "learning_rate": 1.4957827470698809e-05, "loss": 0.5823, "step": 2049 }, { "epoch": 0.36, "learning_rate": 1.4952953846424092e-05, "loss": 0.5732, "step": 2050 }, { "epoch": 0.36, "learning_rate": 1.4948078662801263e-05, "loss": 0.576, "step": 2051 }, { "epoch": 0.36, "learning_rate": 1.4943201921365178e-05, "loss": 0.5798, "step": 2052 }, { "epoch": 0.36, "learning_rate": 1.4938323623651193e-05, "loss": 0.5932, "step": 2053 }, { "epoch": 0.36, "learning_rate": 1.4933443771195153e-05, "loss": 0.5803, "step": 2054 }, { "epoch": 0.36, "learning_rate": 1.4928562365533393e-05, "loss": 0.5846, "step": 2055 }, { "epoch": 0.36, "learning_rate": 1.4923679408202733e-05, "loss": 0.5691, "step": 2056 }, { "epoch": 0.36, "learning_rate": 1.4918794900740485e-05, "loss": 0.569, "step": 2057 }, { "epoch": 0.36, "learning_rate": 1.4913908844684448e-05, "loss": 0.5728, "step": 2058 }, { "epoch": 0.36, "learning_rate": 1.490902124157291e-05, "loss": 0.5791, "step": 2059 }, { "epoch": 0.36, "learning_rate": 1.4904132092944638e-05, "loss": 0.574, "step": 2060 }, { "epoch": 0.36, "learning_rate": 1.4899241400338901e-05, "loss": 0.5915, "step": 2061 }, { "epoch": 0.36, "learning_rate": 1.489434916529544e-05, "loss": 0.582, "step": 2062 }, { "epoch": 0.36, "learning_rate": 1.488945538935449e-05, "loss": 0.589, "step": 2063 }, { "epoch": 0.36, "learning_rate": 1.4884560074056767e-05, "loss": 0.5811, "step": 2064 }, { "epoch": 0.36, "learning_rate": 1.487966322094347e-05, "loss": 0.5921, "step": 2065 }, { "epoch": 0.36, "learning_rate": 1.4874764831556285e-05, "loss": 0.5628, "step": 2066 }, { "epoch": 0.36, "learning_rate": 1.4869864907437389e-05, "loss": 0.5754, "step": 2067 }, { "epoch": 0.36, "learning_rate": 1.4864963450129423e-05, "loss": 0.5647, "step": 2068 }, { "epoch": 0.36, "learning_rate": 1.4860060461175534e-05, "loss": 0.5966, "step": 2069 }, { "epoch": 0.36, "learning_rate": 1.4855155942119331e-05, "loss": 0.5885, "step": 2070 }, { "epoch": 0.36, "learning_rate": 1.4850249894504915e-05, "loss": 0.5901, "step": 2071 }, { "epoch": 0.36, "learning_rate": 1.484534231987687e-05, "loss": 0.575, "step": 2072 }, { "epoch": 0.36, "learning_rate": 1.4840433219780255e-05, "loss": 0.5685, "step": 2073 }, { "epoch": 0.36, "learning_rate": 1.4835522595760612e-05, "loss": 0.5651, "step": 2074 }, { "epoch": 0.36, "learning_rate": 1.4830610449363957e-05, "loss": 0.5797, "step": 2075 }, { "epoch": 0.36, "learning_rate": 1.48256967821368e-05, "loss": 0.5691, "step": 2076 }, { "epoch": 0.36, "learning_rate": 1.4820781595626116e-05, "loss": 0.6044, "step": 2077 }, { "epoch": 0.36, "learning_rate": 1.4815864891379362e-05, "loss": 0.5946, "step": 2078 }, { "epoch": 0.36, "learning_rate": 1.4810946670944476e-05, "loss": 0.5924, "step": 2079 }, { "epoch": 0.36, "learning_rate": 1.4806026935869867e-05, "loss": 0.5705, "step": 2080 }, { "epoch": 0.36, "learning_rate": 1.4801105687704429e-05, "loss": 0.5925, "step": 2081 }, { "epoch": 0.36, "learning_rate": 1.4796182927997525e-05, "loss": 0.5913, "step": 2082 }, { "epoch": 0.36, "learning_rate": 1.4791258658298997e-05, "loss": 0.5791, "step": 2083 }, { "epoch": 0.36, "learning_rate": 1.4786332880159166e-05, "loss": 0.5871, "step": 2084 }, { "epoch": 0.36, "learning_rate": 1.4781405595128822e-05, "loss": 0.5789, "step": 2085 }, { "epoch": 0.36, "learning_rate": 1.4776476804759233e-05, "loss": 0.5909, "step": 2086 }, { "epoch": 0.36, "learning_rate": 1.4771546510602135e-05, "loss": 0.5841, "step": 2087 }, { "epoch": 0.36, "learning_rate": 1.4766614714209748e-05, "loss": 0.5693, "step": 2088 }, { "epoch": 0.36, "learning_rate": 1.4761681417134754e-05, "loss": 0.5802, "step": 2089 }, { "epoch": 0.36, "learning_rate": 1.4756746620930317e-05, "loss": 0.5782, "step": 2090 }, { "epoch": 0.36, "learning_rate": 1.4751810327150066e-05, "loss": 0.5911, "step": 2091 }, { "epoch": 0.36, "learning_rate": 1.4746872537348108e-05, "loss": 0.5868, "step": 2092 }, { "epoch": 0.36, "learning_rate": 1.474193325307901e-05, "loss": 0.5724, "step": 2093 }, { "epoch": 0.36, "learning_rate": 1.473699247589782e-05, "loss": 0.569, "step": 2094 }, { "epoch": 0.36, "learning_rate": 1.4732050207360056e-05, "loss": 0.572, "step": 2095 }, { "epoch": 0.36, "learning_rate": 1.4727106449021696e-05, "loss": 0.5703, "step": 2096 }, { "epoch": 0.36, "learning_rate": 1.4722161202439195e-05, "loss": 0.5863, "step": 2097 }, { "epoch": 0.36, "learning_rate": 1.4717214469169478e-05, "loss": 0.58, "step": 2098 }, { "epoch": 0.36, "learning_rate": 1.4712266250769932e-05, "loss": 0.5801, "step": 2099 }, { "epoch": 0.36, "learning_rate": 1.4707316548798414e-05, "loss": 0.5846, "step": 2100 }, { "epoch": 0.36, "learning_rate": 1.4702365364813248e-05, "loss": 0.5892, "step": 2101 }, { "epoch": 0.36, "learning_rate": 1.4697412700373223e-05, "loss": 0.5838, "step": 2102 }, { "epoch": 0.36, "learning_rate": 1.4692458557037603e-05, "loss": 0.5773, "step": 2103 }, { "epoch": 0.36, "learning_rate": 1.4687502936366105e-05, "loss": 0.5782, "step": 2104 }, { "epoch": 0.36, "learning_rate": 1.4682545839918918e-05, "loss": 0.5909, "step": 2105 }, { "epoch": 0.36, "learning_rate": 1.4677587269256693e-05, "loss": 0.6019, "step": 2106 }, { "epoch": 0.36, "learning_rate": 1.467262722594055e-05, "loss": 0.6023, "step": 2107 }, { "epoch": 0.37, "learning_rate": 1.4667665711532066e-05, "loss": 0.5615, "step": 2108 }, { "epoch": 0.37, "learning_rate": 1.4662702727593284e-05, "loss": 0.5876, "step": 2109 }, { "epoch": 0.37, "learning_rate": 1.4657738275686711e-05, "loss": 0.5701, "step": 2110 }, { "epoch": 0.37, "learning_rate": 1.4652772357375316e-05, "loss": 0.5658, "step": 2111 }, { "epoch": 0.37, "learning_rate": 1.4647804974222525e-05, "loss": 0.6002, "step": 2112 }, { "epoch": 0.37, "learning_rate": 1.4642836127792237e-05, "loss": 0.58, "step": 2113 }, { "epoch": 0.37, "learning_rate": 1.4637865819648794e-05, "loss": 0.5858, "step": 2114 }, { "epoch": 0.37, "learning_rate": 1.4632894051357016e-05, "loss": 0.58, "step": 2115 }, { "epoch": 0.37, "learning_rate": 1.462792082448217e-05, "loss": 0.5955, "step": 2116 }, { "epoch": 0.37, "learning_rate": 1.4622946140589984e-05, "loss": 0.577, "step": 2117 }, { "epoch": 0.37, "learning_rate": 1.4617970001246652e-05, "loss": 0.5575, "step": 2118 }, { "epoch": 0.37, "learning_rate": 1.461299240801882e-05, "loss": 0.5784, "step": 2119 }, { "epoch": 0.37, "learning_rate": 1.4608013362473594e-05, "loss": 0.5687, "step": 2120 }, { "epoch": 0.37, "learning_rate": 1.460303286617854e-05, "loss": 0.5722, "step": 2121 }, { "epoch": 0.37, "learning_rate": 1.4598050920701673e-05, "loss": 0.5734, "step": 2122 }, { "epoch": 0.37, "learning_rate": 1.4593067527611467e-05, "loss": 0.5971, "step": 2123 }, { "epoch": 0.37, "learning_rate": 1.4588082688476858e-05, "loss": 0.5747, "step": 2124 }, { "epoch": 0.37, "learning_rate": 1.4583096404867227e-05, "loss": 0.5899, "step": 2125 }, { "epoch": 0.37, "learning_rate": 1.4578108678352421e-05, "loss": 0.5789, "step": 2126 }, { "epoch": 0.37, "learning_rate": 1.4573119510502732e-05, "loss": 0.5886, "step": 2127 }, { "epoch": 0.37, "learning_rate": 1.4568128902888912e-05, "loss": 0.5814, "step": 2128 }, { "epoch": 0.37, "learning_rate": 1.4563136857082164e-05, "loss": 0.5789, "step": 2129 }, { "epoch": 0.37, "learning_rate": 1.4558143374654141e-05, "loss": 0.5906, "step": 2130 }, { "epoch": 0.37, "learning_rate": 1.4553148457176954e-05, "loss": 0.5723, "step": 2131 }, { "epoch": 0.37, "learning_rate": 1.4548152106223157e-05, "loss": 0.5856, "step": 2132 }, { "epoch": 0.37, "learning_rate": 1.4543154323365765e-05, "loss": 0.5808, "step": 2133 }, { "epoch": 0.37, "learning_rate": 1.4538155110178242e-05, "loss": 0.581, "step": 2134 }, { "epoch": 0.37, "learning_rate": 1.4533154468234493e-05, "loss": 0.6061, "step": 2135 }, { "epoch": 0.37, "learning_rate": 1.4528152399108889e-05, "loss": 0.5621, "step": 2136 }, { "epoch": 0.37, "learning_rate": 1.4523148904376234e-05, "loss": 0.5901, "step": 2137 }, { "epoch": 0.37, "learning_rate": 1.451814398561179e-05, "loss": 0.5577, "step": 2138 }, { "epoch": 0.37, "learning_rate": 1.4513137644391266e-05, "loss": 0.5736, "step": 2139 }, { "epoch": 0.37, "learning_rate": 1.4508129882290818e-05, "loss": 0.5695, "step": 2140 }, { "epoch": 0.37, "learning_rate": 1.4503120700887048e-05, "loss": 0.5681, "step": 2141 }, { "epoch": 0.37, "learning_rate": 1.4498110101757009e-05, "loss": 0.5596, "step": 2142 }, { "epoch": 0.37, "learning_rate": 1.4493098086478196e-05, "loss": 0.5924, "step": 2143 }, { "epoch": 0.37, "learning_rate": 1.4488084656628552e-05, "loss": 0.5771, "step": 2144 }, { "epoch": 0.37, "learning_rate": 1.4483069813786466e-05, "loss": 0.5825, "step": 2145 }, { "epoch": 0.37, "learning_rate": 1.4478053559530767e-05, "loss": 0.587, "step": 2146 }, { "epoch": 0.37, "learning_rate": 1.4473035895440735e-05, "loss": 0.5876, "step": 2147 }, { "epoch": 0.37, "learning_rate": 1.446801682309609e-05, "loss": 0.5736, "step": 2148 }, { "epoch": 0.37, "learning_rate": 1.4462996344077e-05, "loss": 0.585, "step": 2149 }, { "epoch": 0.37, "learning_rate": 1.4457974459964068e-05, "loss": 0.579, "step": 2150 }, { "epoch": 0.37, "learning_rate": 1.4452951172338343e-05, "loss": 0.5691, "step": 2151 }, { "epoch": 0.37, "learning_rate": 1.444792648278132e-05, "loss": 0.5754, "step": 2152 }, { "epoch": 0.37, "learning_rate": 1.4442900392874928e-05, "loss": 0.576, "step": 2153 }, { "epoch": 0.37, "learning_rate": 1.4437872904201542e-05, "loss": 0.5796, "step": 2154 }, { "epoch": 0.37, "learning_rate": 1.4432844018343981e-05, "loss": 0.5755, "step": 2155 }, { "epoch": 0.37, "learning_rate": 1.4427813736885489e-05, "loss": 0.5846, "step": 2156 }, { "epoch": 0.37, "learning_rate": 1.442278206140977e-05, "loss": 0.5843, "step": 2157 }, { "epoch": 0.37, "learning_rate": 1.441774899350095e-05, "loss": 0.5939, "step": 2158 }, { "epoch": 0.37, "learning_rate": 1.4412714534743599e-05, "loss": 0.5792, "step": 2159 }, { "epoch": 0.37, "learning_rate": 1.440767868672273e-05, "loss": 0.5805, "step": 2160 }, { "epoch": 0.37, "learning_rate": 1.4402641451023783e-05, "loss": 0.5966, "step": 2161 }, { "epoch": 0.37, "learning_rate": 1.4397602829232646e-05, "loss": 0.5856, "step": 2162 }, { "epoch": 0.37, "learning_rate": 1.4392562822935632e-05, "loss": 0.5785, "step": 2163 }, { "epoch": 0.37, "learning_rate": 1.4387521433719505e-05, "loss": 0.5837, "step": 2164 }, { "epoch": 0.38, "learning_rate": 1.4382478663171449e-05, "loss": 0.589, "step": 2165 }, { "epoch": 0.38, "learning_rate": 1.4377434512879087e-05, "loss": 0.5853, "step": 2166 }, { "epoch": 0.38, "learning_rate": 1.4372388984430486e-05, "loss": 0.5689, "step": 2167 }, { "epoch": 0.38, "learning_rate": 1.4367342079414135e-05, "loss": 0.5755, "step": 2168 }, { "epoch": 0.38, "learning_rate": 1.4362293799418961e-05, "loss": 0.5804, "step": 2169 }, { "epoch": 0.38, "learning_rate": 1.4357244146034325e-05, "loss": 0.5811, "step": 2170 }, { "epoch": 0.38, "learning_rate": 1.435219312085002e-05, "loss": 0.5744, "step": 2171 }, { "epoch": 0.38, "learning_rate": 1.4347140725456265e-05, "loss": 0.5804, "step": 2172 }, { "epoch": 0.38, "learning_rate": 1.4342086961443725e-05, "loss": 0.5637, "step": 2173 }, { "epoch": 0.38, "learning_rate": 1.4337031830403478e-05, "loss": 0.584, "step": 2174 }, { "epoch": 0.38, "learning_rate": 1.4331975333927042e-05, "loss": 0.5875, "step": 2175 }, { "epoch": 0.38, "learning_rate": 1.4326917473606368e-05, "loss": 0.5945, "step": 2176 }, { "epoch": 0.38, "learning_rate": 1.4321858251033827e-05, "loss": 0.5772, "step": 2177 }, { "epoch": 0.38, "learning_rate": 1.4316797667802229e-05, "loss": 0.5749, "step": 2178 }, { "epoch": 0.38, "learning_rate": 1.4311735725504803e-05, "loss": 0.5702, "step": 2179 }, { "epoch": 0.38, "learning_rate": 1.430667242573521e-05, "loss": 0.5893, "step": 2180 }, { "epoch": 0.38, "learning_rate": 1.4301607770087542e-05, "loss": 0.5821, "step": 2181 }, { "epoch": 0.38, "learning_rate": 1.429654176015631e-05, "loss": 0.568, "step": 2182 }, { "epoch": 0.38, "learning_rate": 1.4291474397536463e-05, "loss": 0.5948, "step": 2183 }, { "epoch": 0.38, "learning_rate": 1.4286405683823359e-05, "loss": 0.5759, "step": 2184 }, { "epoch": 0.38, "learning_rate": 1.42813356206128e-05, "loss": 0.5888, "step": 2185 }, { "epoch": 0.38, "learning_rate": 1.4276264209500998e-05, "loss": 0.5807, "step": 2186 }, { "epoch": 0.38, "learning_rate": 1.4271191452084598e-05, "loss": 0.5723, "step": 2187 }, { "epoch": 0.38, "learning_rate": 1.4266117349960661e-05, "loss": 0.552, "step": 2188 }, { "epoch": 0.38, "learning_rate": 1.4261041904726687e-05, "loss": 0.5933, "step": 2189 }, { "epoch": 0.38, "learning_rate": 1.425596511798058e-05, "loss": 0.5877, "step": 2190 }, { "epoch": 0.38, "learning_rate": 1.4250886991320676e-05, "loss": 0.5773, "step": 2191 }, { "epoch": 0.38, "learning_rate": 1.4245807526345732e-05, "loss": 0.5846, "step": 2192 }, { "epoch": 0.38, "learning_rate": 1.4240726724654925e-05, "loss": 0.5938, "step": 2193 }, { "epoch": 0.38, "learning_rate": 1.4235644587847857e-05, "loss": 0.5792, "step": 2194 }, { "epoch": 0.38, "learning_rate": 1.4230561117524545e-05, "loss": 0.5858, "step": 2195 }, { "epoch": 0.38, "learning_rate": 1.4225476315285427e-05, "loss": 0.5721, "step": 2196 }, { "epoch": 0.38, "learning_rate": 1.4220390182731361e-05, "loss": 0.5819, "step": 2197 }, { "epoch": 0.38, "learning_rate": 1.4215302721463624e-05, "loss": 0.5621, "step": 2198 }, { "epoch": 0.38, "learning_rate": 1.4210213933083916e-05, "loss": 0.5653, "step": 2199 }, { "epoch": 0.38, "learning_rate": 1.4205123819194344e-05, "loss": 0.5767, "step": 2200 }, { "epoch": 0.38, "learning_rate": 1.4200032381397439e-05, "loss": 0.5765, "step": 2201 }, { "epoch": 0.38, "learning_rate": 1.4194939621296153e-05, "loss": 0.57, "step": 2202 }, { "epoch": 0.38, "learning_rate": 1.4189845540493845e-05, "loss": 0.5681, "step": 2203 }, { "epoch": 0.38, "learning_rate": 1.41847501405943e-05, "loss": 0.567, "step": 2204 }, { "epoch": 0.38, "learning_rate": 1.4179653423201705e-05, "loss": 0.5731, "step": 2205 }, { "epoch": 0.38, "learning_rate": 1.4174555389920675e-05, "loss": 0.5854, "step": 2206 }, { "epoch": 0.38, "learning_rate": 1.4169456042356233e-05, "loss": 0.5846, "step": 2207 }, { "epoch": 0.38, "learning_rate": 1.4164355382113813e-05, "loss": 0.5855, "step": 2208 }, { "epoch": 0.38, "learning_rate": 1.4159253410799272e-05, "loss": 0.5764, "step": 2209 }, { "epoch": 0.38, "learning_rate": 1.4154150130018867e-05, "loss": 0.5846, "step": 2210 }, { "epoch": 0.38, "learning_rate": 1.4149045541379276e-05, "loss": 0.5831, "step": 2211 }, { "epoch": 0.38, "learning_rate": 1.414393964648759e-05, "loss": 0.5904, "step": 2212 }, { "epoch": 0.38, "learning_rate": 1.4138832446951305e-05, "loss": 0.5725, "step": 2213 }, { "epoch": 0.38, "learning_rate": 1.413372394437833e-05, "loss": 0.5893, "step": 2214 }, { "epoch": 0.38, "learning_rate": 1.4128614140376985e-05, "loss": 0.5712, "step": 2215 }, { "epoch": 0.38, "learning_rate": 1.4123503036556003e-05, "loss": 0.5924, "step": 2216 }, { "epoch": 0.38, "learning_rate": 1.411839063452452e-05, "loss": 0.5681, "step": 2217 }, { "epoch": 0.38, "learning_rate": 1.4113276935892079e-05, "loss": 0.5828, "step": 2218 }, { "epoch": 0.38, "learning_rate": 1.4108161942268643e-05, "loss": 0.574, "step": 2219 }, { "epoch": 0.38, "learning_rate": 1.4103045655264576e-05, "loss": 0.5678, "step": 2220 }, { "epoch": 0.38, "learning_rate": 1.409792807649064e-05, "loss": 0.5674, "step": 2221 }, { "epoch": 0.38, "learning_rate": 1.4092809207558022e-05, "loss": 0.5711, "step": 2222 }, { "epoch": 0.39, "learning_rate": 1.40876890500783e-05, "loss": 0.5745, "step": 2223 }, { "epoch": 0.39, "learning_rate": 1.4082567605663461e-05, "loss": 0.5852, "step": 2224 }, { "epoch": 0.39, "learning_rate": 1.4077444875925906e-05, "loss": 0.581, "step": 2225 }, { "epoch": 0.39, "learning_rate": 1.4072320862478428e-05, "loss": 0.5887, "step": 2226 }, { "epoch": 0.39, "learning_rate": 1.4067195566934235e-05, "loss": 0.5813, "step": 2227 }, { "epoch": 0.39, "learning_rate": 1.4062068990906933e-05, "loss": 0.5849, "step": 2228 }, { "epoch": 0.39, "learning_rate": 1.4056941136010526e-05, "loss": 0.5658, "step": 2229 }, { "epoch": 0.39, "learning_rate": 1.4051812003859436e-05, "loss": 0.5808, "step": 2230 }, { "epoch": 0.39, "learning_rate": 1.4046681596068468e-05, "loss": 0.5666, "step": 2231 }, { "epoch": 0.39, "learning_rate": 1.4041549914252843e-05, "loss": 0.5738, "step": 2232 }, { "epoch": 0.39, "learning_rate": 1.4036416960028181e-05, "loss": 0.5732, "step": 2233 }, { "epoch": 0.39, "learning_rate": 1.4031282735010499e-05, "loss": 0.5713, "step": 2234 }, { "epoch": 0.39, "learning_rate": 1.402614724081621e-05, "loss": 0.5678, "step": 2235 }, { "epoch": 0.39, "learning_rate": 1.4021010479062135e-05, "loss": 0.5897, "step": 2236 }, { "epoch": 0.39, "learning_rate": 1.4015872451365493e-05, "loss": 0.5674, "step": 2237 }, { "epoch": 0.39, "learning_rate": 1.4010733159343897e-05, "loss": 0.5877, "step": 2238 }, { "epoch": 0.39, "learning_rate": 1.4005592604615357e-05, "loss": 0.5597, "step": 2239 }, { "epoch": 0.39, "learning_rate": 1.400045078879829e-05, "loss": 0.5591, "step": 2240 }, { "epoch": 0.39, "learning_rate": 1.3995307713511504e-05, "loss": 0.5781, "step": 2241 }, { "epoch": 0.39, "learning_rate": 1.3990163380374195e-05, "loss": 0.5807, "step": 2242 }, { "epoch": 0.39, "learning_rate": 1.3985017791005972e-05, "loss": 0.5709, "step": 2243 }, { "epoch": 0.39, "learning_rate": 1.3979870947026828e-05, "loss": 0.5604, "step": 2244 }, { "epoch": 0.39, "learning_rate": 1.3974722850057152e-05, "loss": 0.5632, "step": 2245 }, { "epoch": 0.39, "learning_rate": 1.3969573501717732e-05, "loss": 0.5875, "step": 2246 }, { "epoch": 0.39, "learning_rate": 1.3964422903629746e-05, "loss": 0.574, "step": 2247 }, { "epoch": 0.39, "learning_rate": 1.3959271057414769e-05, "loss": 0.5714, "step": 2248 }, { "epoch": 0.39, "learning_rate": 1.3954117964694767e-05, "loss": 0.5809, "step": 2249 }, { "epoch": 0.39, "learning_rate": 1.3948963627092091e-05, "loss": 0.5716, "step": 2250 }, { "epoch": 0.39, "learning_rate": 1.3943808046229504e-05, "loss": 0.5795, "step": 2251 }, { "epoch": 0.39, "learning_rate": 1.3938651223730133e-05, "loss": 0.5974, "step": 2252 }, { "epoch": 0.39, "learning_rate": 1.3933493161217521e-05, "loss": 0.5649, "step": 2253 }, { "epoch": 0.39, "learning_rate": 1.3928333860315588e-05, "loss": 0.5822, "step": 2254 }, { "epoch": 0.39, "learning_rate": 1.3923173322648646e-05, "loss": 0.5666, "step": 2255 }, { "epoch": 0.39, "learning_rate": 1.39180115498414e-05, "loss": 0.5797, "step": 2256 }, { "epoch": 0.39, "learning_rate": 1.3912848543518936e-05, "loss": 0.5901, "step": 2257 }, { "epoch": 0.39, "learning_rate": 1.3907684305306733e-05, "loss": 0.5836, "step": 2258 }, { "epoch": 0.39, "learning_rate": 1.3902518836830664e-05, "loss": 0.5786, "step": 2259 }, { "epoch": 0.39, "learning_rate": 1.3897352139716979e-05, "loss": 0.5739, "step": 2260 }, { "epoch": 0.39, "learning_rate": 1.3892184215592323e-05, "loss": 0.5883, "step": 2261 }, { "epoch": 0.39, "learning_rate": 1.3887015066083722e-05, "loss": 0.5821, "step": 2262 }, { "epoch": 0.39, "learning_rate": 1.3881844692818587e-05, "loss": 0.5911, "step": 2263 }, { "epoch": 0.39, "learning_rate": 1.387667309742472e-05, "loss": 0.5673, "step": 2264 }, { "epoch": 0.39, "learning_rate": 1.3871500281530302e-05, "loss": 0.5801, "step": 2265 }, { "epoch": 0.39, "learning_rate": 1.3866326246763902e-05, "loss": 0.5702, "step": 2266 }, { "epoch": 0.39, "learning_rate": 1.386115099475447e-05, "loss": 0.5766, "step": 2267 }, { "epoch": 0.39, "learning_rate": 1.3855974527131344e-05, "loss": 0.5835, "step": 2268 }, { "epoch": 0.39, "learning_rate": 1.3850796845524241e-05, "loss": 0.5814, "step": 2269 }, { "epoch": 0.39, "learning_rate": 1.3845617951563255e-05, "loss": 0.5694, "step": 2270 }, { "epoch": 0.39, "learning_rate": 1.3840437846878872e-05, "loss": 0.5733, "step": 2271 }, { "epoch": 0.39, "learning_rate": 1.3835256533101955e-05, "loss": 0.5764, "step": 2272 }, { "epoch": 0.39, "learning_rate": 1.3830074011863746e-05, "loss": 0.5791, "step": 2273 }, { "epoch": 0.39, "learning_rate": 1.3824890284795867e-05, "loss": 0.5874, "step": 2274 }, { "epoch": 0.39, "learning_rate": 1.3819705353530322e-05, "loss": 0.5891, "step": 2275 }, { "epoch": 0.39, "learning_rate": 1.381451921969949e-05, "loss": 0.5631, "step": 2276 }, { "epoch": 0.39, "learning_rate": 1.3809331884936142e-05, "loss": 0.5794, "step": 2277 }, { "epoch": 0.39, "learning_rate": 1.3804143350873403e-05, "loss": 0.5656, "step": 2278 }, { "epoch": 0.39, "learning_rate": 1.3798953619144797e-05, "loss": 0.5626, "step": 2279 }, { "epoch": 0.39, "learning_rate": 1.3793762691384215e-05, "loss": 0.5644, "step": 2280 }, { "epoch": 0.4, "learning_rate": 1.3788570569225929e-05, "loss": 0.5561, "step": 2281 }, { "epoch": 0.4, "learning_rate": 1.3783377254304586e-05, "loss": 0.5794, "step": 2282 }, { "epoch": 0.4, "learning_rate": 1.3778182748255204e-05, "loss": 0.5912, "step": 2283 }, { "epoch": 0.4, "learning_rate": 1.377298705271318e-05, "loss": 0.5573, "step": 2284 }, { "epoch": 0.4, "learning_rate": 1.3767790169314286e-05, "loss": 0.5749, "step": 2285 }, { "epoch": 0.4, "learning_rate": 1.3762592099694666e-05, "loss": 0.5921, "step": 2286 }, { "epoch": 0.4, "learning_rate": 1.375739284549084e-05, "loss": 0.5674, "step": 2287 }, { "epoch": 0.4, "learning_rate": 1.3752192408339698e-05, "loss": 0.5813, "step": 2288 }, { "epoch": 0.4, "learning_rate": 1.3746990789878503e-05, "loss": 0.5847, "step": 2289 }, { "epoch": 0.4, "learning_rate": 1.3741787991744895e-05, "loss": 0.5757, "step": 2290 }, { "epoch": 0.4, "learning_rate": 1.3736584015576874e-05, "loss": 0.5881, "step": 2291 }, { "epoch": 0.4, "learning_rate": 1.373137886301282e-05, "loss": 0.5838, "step": 2292 }, { "epoch": 0.4, "learning_rate": 1.3726172535691489e-05, "loss": 0.591, "step": 2293 }, { "epoch": 0.4, "learning_rate": 1.3720965035251989e-05, "loss": 0.5968, "step": 2294 }, { "epoch": 0.4, "learning_rate": 1.3715756363333817e-05, "loss": 0.5682, "step": 2295 }, { "epoch": 0.4, "learning_rate": 1.371054652157682e-05, "loss": 0.5776, "step": 2296 }, { "epoch": 0.4, "learning_rate": 1.3705335511621229e-05, "loss": 0.6036, "step": 2297 }, { "epoch": 0.4, "learning_rate": 1.3700123335107634e-05, "loss": 0.576, "step": 2298 }, { "epoch": 0.4, "learning_rate": 1.3694909993676993e-05, "loss": 0.5866, "step": 2299 }, { "epoch": 0.4, "learning_rate": 1.3689695488970638e-05, "loss": 0.5519, "step": 2300 }, { "epoch": 0.4, "learning_rate": 1.3684479822630255e-05, "loss": 0.5678, "step": 2301 }, { "epoch": 0.4, "learning_rate": 1.3679262996297904e-05, "loss": 0.5736, "step": 2302 }, { "epoch": 0.4, "learning_rate": 1.3674045011616013e-05, "loss": 0.5781, "step": 2303 }, { "epoch": 0.4, "learning_rate": 1.3668825870227366e-05, "loss": 0.5853, "step": 2304 }, { "epoch": 0.4, "learning_rate": 1.3663605573775119e-05, "loss": 0.5719, "step": 2305 }, { "epoch": 0.4, "learning_rate": 1.3658384123902786e-05, "loss": 0.5759, "step": 2306 }, { "epoch": 0.4, "learning_rate": 1.3653161522254244e-05, "loss": 0.5795, "step": 2307 }, { "epoch": 0.4, "learning_rate": 1.3647937770473739e-05, "loss": 0.5698, "step": 2308 }, { "epoch": 0.4, "learning_rate": 1.364271287020587e-05, "loss": 0.5873, "step": 2309 }, { "epoch": 0.4, "learning_rate": 1.3637486823095608e-05, "loss": 0.5807, "step": 2310 }, { "epoch": 0.4, "learning_rate": 1.3632259630788278e-05, "loss": 0.5722, "step": 2311 }, { "epoch": 0.4, "learning_rate": 1.3627031294929564e-05, "loss": 0.5828, "step": 2312 }, { "epoch": 0.4, "learning_rate": 1.3621801817165517e-05, "loss": 0.5996, "step": 2313 }, { "epoch": 0.4, "learning_rate": 1.3616571199142542e-05, "loss": 0.5865, "step": 2314 }, { "epoch": 0.4, "learning_rate": 1.3611339442507403e-05, "loss": 0.5765, "step": 2315 }, { "epoch": 0.4, "learning_rate": 1.3606106548907228e-05, "loss": 0.5698, "step": 2316 }, { "epoch": 0.4, "learning_rate": 1.3600872519989497e-05, "loss": 0.5784, "step": 2317 }, { "epoch": 0.4, "learning_rate": 1.3595637357402049e-05, "loss": 0.5854, "step": 2318 }, { "epoch": 0.4, "learning_rate": 1.3590401062793084e-05, "loss": 0.5776, "step": 2319 }, { "epoch": 0.4, "learning_rate": 1.3585163637811148e-05, "loss": 0.5701, "step": 2320 }, { "epoch": 0.4, "learning_rate": 1.3579925084105154e-05, "loss": 0.576, "step": 2321 }, { "epoch": 0.4, "learning_rate": 1.3574685403324367e-05, "loss": 0.5819, "step": 2322 }, { "epoch": 0.4, "learning_rate": 1.3569444597118402e-05, "loss": 0.5709, "step": 2323 }, { "epoch": 0.4, "learning_rate": 1.3564202667137239e-05, "loss": 0.5807, "step": 2324 }, { "epoch": 0.4, "learning_rate": 1.35589596150312e-05, "loss": 0.5671, "step": 2325 }, { "epoch": 0.4, "learning_rate": 1.3553715442450963e-05, "loss": 0.5711, "step": 2326 }, { "epoch": 0.4, "learning_rate": 1.3548470151047567e-05, "loss": 0.5637, "step": 2327 }, { "epoch": 0.4, "learning_rate": 1.3543223742472394e-05, "loss": 0.5679, "step": 2328 }, { "epoch": 0.4, "learning_rate": 1.3537976218377182e-05, "loss": 0.5786, "step": 2329 }, { "epoch": 0.4, "learning_rate": 1.3532727580414018e-05, "loss": 0.5915, "step": 2330 }, { "epoch": 0.4, "learning_rate": 1.3527477830235343e-05, "loss": 0.6096, "step": 2331 }, { "epoch": 0.4, "learning_rate": 1.3522226969493945e-05, "loss": 0.5724, "step": 2332 }, { "epoch": 0.4, "learning_rate": 1.351697499984296e-05, "loss": 0.585, "step": 2333 }, { "epoch": 0.4, "learning_rate": 1.3511721922935884e-05, "loss": 0.5719, "step": 2334 }, { "epoch": 0.4, "learning_rate": 1.3506467740426541e-05, "loss": 0.5831, "step": 2335 }, { "epoch": 0.4, "learning_rate": 1.3501212453969124e-05, "loss": 0.5744, "step": 2336 }, { "epoch": 0.4, "learning_rate": 1.3495956065218168e-05, "loss": 0.5807, "step": 2337 }, { "epoch": 0.4, "learning_rate": 1.3490698575828543e-05, "loss": 0.5964, "step": 2338 }, { "epoch": 0.41, "learning_rate": 1.3485439987455482e-05, "loss": 0.5872, "step": 2339 }, { "epoch": 0.41, "learning_rate": 1.3480180301754553e-05, "loss": 0.5686, "step": 2340 }, { "epoch": 0.41, "learning_rate": 1.3474919520381673e-05, "loss": 0.5918, "step": 2341 }, { "epoch": 0.41, "learning_rate": 1.3469657644993109e-05, "loss": 0.5746, "step": 2342 }, { "epoch": 0.41, "learning_rate": 1.3464394677245459e-05, "loss": 0.5743, "step": 2343 }, { "epoch": 0.41, "learning_rate": 1.3459130618795678e-05, "loss": 0.582, "step": 2344 }, { "epoch": 0.41, "learning_rate": 1.3453865471301062e-05, "loss": 0.6015, "step": 2345 }, { "epoch": 0.41, "learning_rate": 1.3448599236419246e-05, "loss": 0.5876, "step": 2346 }, { "epoch": 0.41, "learning_rate": 1.344333191580821e-05, "loss": 0.5787, "step": 2347 }, { "epoch": 0.41, "learning_rate": 1.3438063511126267e-05, "loss": 0.5806, "step": 2348 }, { "epoch": 0.41, "learning_rate": 1.3432794024032088e-05, "loss": 0.5702, "step": 2349 }, { "epoch": 0.41, "learning_rate": 1.3427523456184675e-05, "loss": 0.5774, "step": 2350 }, { "epoch": 0.41, "learning_rate": 1.3422251809243368e-05, "loss": 0.5906, "step": 2351 }, { "epoch": 0.41, "learning_rate": 1.3416979084867851e-05, "loss": 0.5836, "step": 2352 }, { "epoch": 0.41, "learning_rate": 1.3411705284718148e-05, "loss": 0.5936, "step": 2353 }, { "epoch": 0.41, "learning_rate": 1.3406430410454615e-05, "loss": 0.5644, "step": 2354 }, { "epoch": 0.41, "learning_rate": 1.3401154463737957e-05, "loss": 0.5792, "step": 2355 }, { "epoch": 0.41, "learning_rate": 1.3395877446229207e-05, "loss": 0.5697, "step": 2356 }, { "epoch": 0.41, "learning_rate": 1.3390599359589737e-05, "loss": 0.5625, "step": 2357 }, { "epoch": 0.41, "learning_rate": 1.3385320205481262e-05, "loss": 0.568, "step": 2358 }, { "epoch": 0.41, "learning_rate": 1.3380039985565825e-05, "loss": 0.5744, "step": 2359 }, { "epoch": 0.41, "learning_rate": 1.3374758701505812e-05, "loss": 0.596, "step": 2360 }, { "epoch": 0.41, "learning_rate": 1.3369476354963935e-05, "loss": 0.5957, "step": 2361 }, { "epoch": 0.41, "learning_rate": 1.3364192947603247e-05, "loss": 0.5782, "step": 2362 }, { "epoch": 0.41, "learning_rate": 1.3358908481087133e-05, "loss": 0.5929, "step": 2363 }, { "epoch": 0.41, "learning_rate": 1.3353622957079316e-05, "loss": 0.5614, "step": 2364 }, { "epoch": 0.41, "learning_rate": 1.3348336377243842e-05, "loss": 0.5823, "step": 2365 }, { "epoch": 0.41, "learning_rate": 1.3343048743245098e-05, "loss": 0.5637, "step": 2366 }, { "epoch": 0.41, "learning_rate": 1.33377600567478e-05, "loss": 0.5768, "step": 2367 }, { "epoch": 0.41, "learning_rate": 1.3332470319416996e-05, "loss": 0.5733, "step": 2368 }, { "epoch": 0.41, "learning_rate": 1.3327179532918063e-05, "loss": 0.5706, "step": 2369 }, { "epoch": 0.41, "learning_rate": 1.3321887698916709e-05, "loss": 0.5708, "step": 2370 }, { "epoch": 0.41, "learning_rate": 1.3316594819078979e-05, "loss": 0.5796, "step": 2371 }, { "epoch": 0.41, "learning_rate": 1.3311300895071229e-05, "loss": 0.5536, "step": 2372 }, { "epoch": 0.41, "learning_rate": 1.3306005928560166e-05, "loss": 0.5716, "step": 2373 }, { "epoch": 0.41, "learning_rate": 1.330070992121281e-05, "loss": 0.5725, "step": 2374 }, { "epoch": 0.41, "learning_rate": 1.3295412874696512e-05, "loss": 0.5746, "step": 2375 }, { "epoch": 0.41, "learning_rate": 1.3290114790678956e-05, "loss": 0.5915, "step": 2376 }, { "epoch": 0.41, "learning_rate": 1.3284815670828144e-05, "loss": 0.5826, "step": 2377 }, { "epoch": 0.41, "learning_rate": 1.327951551681241e-05, "loss": 0.5764, "step": 2378 }, { "epoch": 0.41, "learning_rate": 1.327421433030041e-05, "loss": 0.5632, "step": 2379 }, { "epoch": 0.41, "learning_rate": 1.326891211296113e-05, "loss": 0.5869, "step": 2380 }, { "epoch": 0.41, "learning_rate": 1.3263608866463878e-05, "loss": 0.5678, "step": 2381 }, { "epoch": 0.41, "learning_rate": 1.325830459247828e-05, "loss": 0.5742, "step": 2382 }, { "epoch": 0.41, "learning_rate": 1.3252999292674292e-05, "loss": 0.5907, "step": 2383 }, { "epoch": 0.41, "learning_rate": 1.3247692968722198e-05, "loss": 0.5987, "step": 2384 }, { "epoch": 0.41, "learning_rate": 1.3242385622292593e-05, "loss": 0.5672, "step": 2385 }, { "epoch": 0.41, "learning_rate": 1.32370772550564e-05, "loss": 0.5869, "step": 2386 }, { "epoch": 0.41, "learning_rate": 1.3231767868684863e-05, "loss": 0.5777, "step": 2387 }, { "epoch": 0.41, "learning_rate": 1.3226457464849549e-05, "loss": 0.568, "step": 2388 }, { "epoch": 0.41, "learning_rate": 1.3221146045222342e-05, "loss": 0.579, "step": 2389 }, { "epoch": 0.41, "learning_rate": 1.321583361147544e-05, "loss": 0.5896, "step": 2390 }, { "epoch": 0.41, "learning_rate": 1.3210520165281376e-05, "loss": 0.575, "step": 2391 }, { "epoch": 0.41, "learning_rate": 1.3205205708312985e-05, "loss": 0.5912, "step": 2392 }, { "epoch": 0.41, "learning_rate": 1.3199890242243432e-05, "loss": 0.5676, "step": 2393 }, { "epoch": 0.41, "learning_rate": 1.3194573768746197e-05, "loss": 0.576, "step": 2394 }, { "epoch": 0.41, "learning_rate": 1.3189256289495074e-05, "loss": 0.5674, "step": 2395 }, { "epoch": 0.42, "learning_rate": 1.3183937806164174e-05, "loss": 0.5795, "step": 2396 }, { "epoch": 0.42, "learning_rate": 1.3178618320427924e-05, "loss": 0.5692, "step": 2397 }, { "epoch": 0.42, "learning_rate": 1.3173297833961074e-05, "loss": 0.586, "step": 2398 }, { "epoch": 0.42, "learning_rate": 1.3167976348438678e-05, "loss": 0.583, "step": 2399 }, { "epoch": 0.42, "learning_rate": 1.3162653865536111e-05, "loss": 0.5683, "step": 2400 }, { "epoch": 0.42, "learning_rate": 1.315733038692906e-05, "loss": 0.5774, "step": 2401 }, { "epoch": 0.42, "learning_rate": 1.3152005914293534e-05, "loss": 0.5831, "step": 2402 }, { "epoch": 0.42, "learning_rate": 1.3146680449305833e-05, "loss": 0.5713, "step": 2403 }, { "epoch": 0.42, "learning_rate": 1.3141353993642594e-05, "loss": 0.5729, "step": 2404 }, { "epoch": 0.42, "learning_rate": 1.3136026548980751e-05, "loss": 0.5619, "step": 2405 }, { "epoch": 0.42, "learning_rate": 1.3130698116997555e-05, "loss": 0.5978, "step": 2406 }, { "epoch": 0.42, "learning_rate": 1.3125368699370567e-05, "loss": 0.5768, "step": 2407 }, { "epoch": 0.42, "learning_rate": 1.3120038297777657e-05, "loss": 0.5643, "step": 2408 }, { "epoch": 0.42, "learning_rate": 1.3114706913897008e-05, "loss": 0.5616, "step": 2409 }, { "epoch": 0.42, "learning_rate": 1.3109374549407107e-05, "loss": 0.569, "step": 2410 }, { "epoch": 0.42, "learning_rate": 1.3104041205986753e-05, "loss": 0.5703, "step": 2411 }, { "epoch": 0.42, "learning_rate": 1.3098706885315057e-05, "loss": 0.5895, "step": 2412 }, { "epoch": 0.42, "learning_rate": 1.3093371589071428e-05, "loss": 0.5622, "step": 2413 }, { "epoch": 0.42, "learning_rate": 1.308803531893559e-05, "loss": 0.5791, "step": 2414 }, { "epoch": 0.42, "learning_rate": 1.3082698076587573e-05, "loss": 0.5641, "step": 2415 }, { "epoch": 0.42, "learning_rate": 1.307735986370771e-05, "loss": 0.5725, "step": 2416 }, { "epoch": 0.42, "learning_rate": 1.3072020681976639e-05, "loss": 0.5817, "step": 2417 }, { "epoch": 0.42, "learning_rate": 1.3066680533075312e-05, "loss": 0.5839, "step": 2418 }, { "epoch": 0.42, "learning_rate": 1.3061339418684967e-05, "loss": 0.5825, "step": 2419 }, { "epoch": 0.42, "learning_rate": 1.3055997340487165e-05, "loss": 0.5727, "step": 2420 }, { "epoch": 0.42, "learning_rate": 1.3050654300163763e-05, "loss": 0.5592, "step": 2421 }, { "epoch": 0.42, "learning_rate": 1.304531029939692e-05, "loss": 0.5879, "step": 2422 }, { "epoch": 0.42, "learning_rate": 1.30399653398691e-05, "loss": 0.5637, "step": 2423 }, { "epoch": 0.42, "learning_rate": 1.3034619423263063e-05, "loss": 0.5782, "step": 2424 }, { "epoch": 0.42, "learning_rate": 1.3029272551261875e-05, "loss": 0.5599, "step": 2425 }, { "epoch": 0.42, "learning_rate": 1.3023924725548902e-05, "loss": 0.5856, "step": 2426 }, { "epoch": 0.42, "learning_rate": 1.3018575947807812e-05, "loss": 0.561, "step": 2427 }, { "epoch": 0.42, "learning_rate": 1.3013226219722575e-05, "loss": 0.5835, "step": 2428 }, { "epoch": 0.42, "learning_rate": 1.3007875542977448e-05, "loss": 0.5977, "step": 2429 }, { "epoch": 0.42, "learning_rate": 1.3002523919257e-05, "loss": 0.5908, "step": 2430 }, { "epoch": 0.42, "learning_rate": 1.2997171350246095e-05, "loss": 0.5627, "step": 2431 }, { "epoch": 0.42, "learning_rate": 1.2991817837629885e-05, "loss": 0.5806, "step": 2432 }, { "epoch": 0.42, "learning_rate": 1.2986463383093838e-05, "loss": 0.5824, "step": 2433 }, { "epoch": 0.42, "learning_rate": 1.2981107988323695e-05, "loss": 0.5702, "step": 2434 }, { "epoch": 0.42, "learning_rate": 1.2975751655005512e-05, "loss": 0.5795, "step": 2435 }, { "epoch": 0.42, "learning_rate": 1.2970394384825637e-05, "loss": 0.5773, "step": 2436 }, { "epoch": 0.42, "learning_rate": 1.2965036179470707e-05, "loss": 0.5654, "step": 2437 }, { "epoch": 0.42, "learning_rate": 1.2959677040627653e-05, "loss": 0.5929, "step": 2438 }, { "epoch": 0.42, "learning_rate": 1.2954316969983704e-05, "loss": 0.5697, "step": 2439 }, { "epoch": 0.42, "learning_rate": 1.2948955969226384e-05, "loss": 0.5751, "step": 2440 }, { "epoch": 0.42, "learning_rate": 1.294359404004351e-05, "loss": 0.5835, "step": 2441 }, { "epoch": 0.42, "learning_rate": 1.293823118412318e-05, "loss": 0.5799, "step": 2442 }, { "epoch": 0.42, "learning_rate": 1.2932867403153799e-05, "loss": 0.5659, "step": 2443 }, { "epoch": 0.42, "learning_rate": 1.2927502698824057e-05, "loss": 0.6032, "step": 2444 }, { "epoch": 0.42, "learning_rate": 1.2922137072822932e-05, "loss": 0.5823, "step": 2445 }, { "epoch": 0.42, "learning_rate": 1.2916770526839693e-05, "loss": 0.5806, "step": 2446 }, { "epoch": 0.42, "learning_rate": 1.2911403062563901e-05, "loss": 0.5523, "step": 2447 }, { "epoch": 0.42, "learning_rate": 1.2906034681685409e-05, "loss": 0.5773, "step": 2448 }, { "epoch": 0.42, "learning_rate": 1.2900665385894351e-05, "loss": 0.577, "step": 2449 }, { "epoch": 0.42, "learning_rate": 1.2895295176881153e-05, "loss": 0.5688, "step": 2450 }, { "epoch": 0.42, "learning_rate": 1.2889924056336531e-05, "loss": 0.5621, "step": 2451 }, { "epoch": 0.42, "learning_rate": 1.2884552025951482e-05, "loss": 0.5637, "step": 2452 }, { "epoch": 0.42, "learning_rate": 1.2879179087417295e-05, "loss": 0.5738, "step": 2453 }, { "epoch": 0.43, "learning_rate": 1.2873805242425543e-05, "loss": 0.5754, "step": 2454 }, { "epoch": 0.43, "learning_rate": 1.2868430492668083e-05, "loss": 0.577, "step": 2455 }, { "epoch": 0.43, "learning_rate": 1.2863054839837056e-05, "loss": 0.5827, "step": 2456 }, { "epoch": 0.43, "learning_rate": 1.2857678285624892e-05, "loss": 0.5806, "step": 2457 }, { "epoch": 0.43, "learning_rate": 1.2852300831724306e-05, "loss": 0.571, "step": 2458 }, { "epoch": 0.43, "learning_rate": 1.2846922479828285e-05, "loss": 0.5855, "step": 2459 }, { "epoch": 0.43, "learning_rate": 1.2841543231630107e-05, "loss": 0.5649, "step": 2460 }, { "epoch": 0.43, "learning_rate": 1.2836163088823335e-05, "loss": 0.5721, "step": 2461 }, { "epoch": 0.43, "learning_rate": 1.2830782053101807e-05, "loss": 0.5649, "step": 2462 }, { "epoch": 0.43, "learning_rate": 1.2825400126159644e-05, "loss": 0.5851, "step": 2463 }, { "epoch": 0.43, "learning_rate": 1.2820017309691254e-05, "loss": 0.5773, "step": 2464 }, { "epoch": 0.43, "learning_rate": 1.2814633605391316e-05, "loss": 0.5697, "step": 2465 }, { "epoch": 0.43, "learning_rate": 1.2809249014954788e-05, "loss": 0.5853, "step": 2466 }, { "epoch": 0.43, "learning_rate": 1.2803863540076918e-05, "loss": 0.5761, "step": 2467 }, { "epoch": 0.43, "learning_rate": 1.2798477182453221e-05, "loss": 0.5687, "step": 2468 }, { "epoch": 0.43, "learning_rate": 1.27930899437795e-05, "loss": 0.5756, "step": 2469 }, { "epoch": 0.43, "learning_rate": 1.2787701825751821e-05, "loss": 0.5732, "step": 2470 }, { "epoch": 0.43, "learning_rate": 1.2782312830066543e-05, "loss": 0.5628, "step": 2471 }, { "epoch": 0.43, "learning_rate": 1.2776922958420295e-05, "loss": 0.57, "step": 2472 }, { "epoch": 0.43, "learning_rate": 1.2771532212509974e-05, "loss": 0.5769, "step": 2473 }, { "epoch": 0.43, "learning_rate": 1.2766140594032762e-05, "loss": 0.5766, "step": 2474 }, { "epoch": 0.43, "learning_rate": 1.2760748104686116e-05, "loss": 0.58, "step": 2475 }, { "epoch": 0.43, "learning_rate": 1.2755354746167758e-05, "loss": 0.5768, "step": 2476 }, { "epoch": 0.43, "learning_rate": 1.2749960520175696e-05, "loss": 0.5715, "step": 2477 }, { "epoch": 0.43, "learning_rate": 1.2744565428408202e-05, "loss": 0.5747, "step": 2478 }, { "epoch": 0.43, "learning_rate": 1.2739169472563822e-05, "loss": 0.5664, "step": 2479 }, { "epoch": 0.43, "learning_rate": 1.2733772654341376e-05, "loss": 0.5732, "step": 2480 }, { "epoch": 0.43, "learning_rate": 1.2728374975439954e-05, "loss": 0.5724, "step": 2481 }, { "epoch": 0.43, "learning_rate": 1.2722976437558919e-05, "loss": 0.5726, "step": 2482 }, { "epoch": 0.43, "learning_rate": 1.2717577042397904e-05, "loss": 0.5707, "step": 2483 }, { "epoch": 0.43, "learning_rate": 1.2712176791656807e-05, "loss": 0.5934, "step": 2484 }, { "epoch": 0.43, "learning_rate": 1.2706775687035808e-05, "loss": 0.589, "step": 2485 }, { "epoch": 0.43, "learning_rate": 1.270137373023534e-05, "loss": 0.569, "step": 2486 }, { "epoch": 0.43, "learning_rate": 1.2695970922956108e-05, "loss": 0.5825, "step": 2487 }, { "epoch": 0.43, "learning_rate": 1.26905672668991e-05, "loss": 0.5771, "step": 2488 }, { "epoch": 0.43, "learning_rate": 1.268516276376555e-05, "loss": 0.5756, "step": 2489 }, { "epoch": 0.43, "learning_rate": 1.2679757415256977e-05, "loss": 0.5854, "step": 2490 }, { "epoch": 0.43, "learning_rate": 1.2674351223075145e-05, "loss": 0.5667, "step": 2491 }, { "epoch": 0.43, "learning_rate": 1.2668944188922105e-05, "loss": 0.5706, "step": 2492 }, { "epoch": 0.43, "learning_rate": 1.2663536314500167e-05, "loss": 0.5717, "step": 2493 }, { "epoch": 0.43, "learning_rate": 1.2658127601511891e-05, "loss": 0.5751, "step": 2494 }, { "epoch": 0.43, "learning_rate": 1.2652718051660122e-05, "loss": 0.5691, "step": 2495 }, { "epoch": 0.43, "learning_rate": 1.264730766664796e-05, "loss": 0.5919, "step": 2496 }, { "epoch": 0.43, "learning_rate": 1.2641896448178759e-05, "loss": 0.5705, "step": 2497 }, { "epoch": 0.43, "learning_rate": 1.263648439795615e-05, "loss": 0.5778, "step": 2498 }, { "epoch": 0.43, "learning_rate": 1.2631071517684017e-05, "loss": 0.5671, "step": 2499 }, { "epoch": 0.43, "learning_rate": 1.2625657809066509e-05, "loss": 0.5777, "step": 2500 }, { "epoch": 0.43, "learning_rate": 1.2620243273808033e-05, "loss": 0.5765, "step": 2501 }, { "epoch": 0.43, "learning_rate": 1.2614827913613256e-05, "loss": 0.5859, "step": 2502 }, { "epoch": 0.43, "learning_rate": 1.2609411730187114e-05, "loss": 0.5612, "step": 2503 }, { "epoch": 0.43, "learning_rate": 1.2603994725234783e-05, "loss": 0.5766, "step": 2504 }, { "epoch": 0.43, "learning_rate": 1.2598576900461716e-05, "loss": 0.5707, "step": 2505 }, { "epoch": 0.43, "learning_rate": 1.259315825757362e-05, "loss": 0.5927, "step": 2506 }, { "epoch": 0.43, "learning_rate": 1.2587738798276457e-05, "loss": 0.5761, "step": 2507 }, { "epoch": 0.43, "learning_rate": 1.2582318524276436e-05, "loss": 0.5765, "step": 2508 }, { "epoch": 0.43, "learning_rate": 1.2576897437280042e-05, "loss": 0.5818, "step": 2509 }, { "epoch": 0.43, "learning_rate": 1.2571475538994e-05, "loss": 0.5658, "step": 2510 }, { "epoch": 0.43, "learning_rate": 1.2566052831125306e-05, "loss": 0.5862, "step": 2511 }, { "epoch": 0.44, "learning_rate": 1.2560629315381192e-05, "loss": 0.5645, "step": 2512 }, { "epoch": 0.44, "learning_rate": 1.2555204993469159e-05, "loss": 0.585, "step": 2513 }, { "epoch": 0.44, "learning_rate": 1.2549779867096956e-05, "loss": 0.5652, "step": 2514 }, { "epoch": 0.44, "learning_rate": 1.2544353937972584e-05, "loss": 0.5893, "step": 2515 }, { "epoch": 0.44, "learning_rate": 1.2538927207804306e-05, "loss": 0.5777, "step": 2516 }, { "epoch": 0.44, "learning_rate": 1.2533499678300618e-05, "loss": 0.5677, "step": 2517 }, { "epoch": 0.44, "learning_rate": 1.252807135117029e-05, "loss": 0.5763, "step": 2518 }, { "epoch": 0.44, "learning_rate": 1.252264222812233e-05, "loss": 0.5863, "step": 2519 }, { "epoch": 0.44, "learning_rate": 1.2517212310865996e-05, "loss": 0.5711, "step": 2520 }, { "epoch": 0.44, "learning_rate": 1.2511781601110804e-05, "loss": 0.5607, "step": 2521 }, { "epoch": 0.44, "learning_rate": 1.2506350100566515e-05, "loss": 0.5817, "step": 2522 }, { "epoch": 0.44, "learning_rate": 1.2500917810943134e-05, "loss": 0.566, "step": 2523 }, { "epoch": 0.44, "learning_rate": 1.2495484733950924e-05, "loss": 0.5851, "step": 2524 }, { "epoch": 0.44, "learning_rate": 1.2490050871300388e-05, "loss": 0.5803, "step": 2525 }, { "epoch": 0.44, "learning_rate": 1.2484616224702282e-05, "loss": 0.5795, "step": 2526 }, { "epoch": 0.44, "learning_rate": 1.2479180795867605e-05, "loss": 0.5805, "step": 2527 }, { "epoch": 0.44, "learning_rate": 1.2473744586507606e-05, "loss": 0.5868, "step": 2528 }, { "epoch": 0.44, "learning_rate": 1.2468307598333774e-05, "loss": 0.5832, "step": 2529 }, { "epoch": 0.44, "learning_rate": 1.246286983305785e-05, "loss": 0.5664, "step": 2530 }, { "epoch": 0.44, "learning_rate": 1.2457431292391811e-05, "loss": 0.5655, "step": 2531 }, { "epoch": 0.44, "learning_rate": 1.2451991978047891e-05, "loss": 0.5796, "step": 2532 }, { "epoch": 0.44, "learning_rate": 1.244655189173855e-05, "loss": 0.5704, "step": 2533 }, { "epoch": 0.44, "learning_rate": 1.2441111035176511e-05, "loss": 0.5803, "step": 2534 }, { "epoch": 0.44, "learning_rate": 1.2435669410074727e-05, "loss": 0.5742, "step": 2535 }, { "epoch": 0.44, "learning_rate": 1.2430227018146387e-05, "loss": 0.5801, "step": 2536 }, { "epoch": 0.44, "learning_rate": 1.2424783861104943e-05, "loss": 0.5762, "step": 2537 }, { "epoch": 0.44, "learning_rate": 1.2419339940664064e-05, "loss": 0.5686, "step": 2538 }, { "epoch": 0.44, "learning_rate": 1.2413895258537676e-05, "loss": 0.5704, "step": 2539 }, { "epoch": 0.44, "learning_rate": 1.2408449816439935e-05, "loss": 0.6157, "step": 2540 }, { "epoch": 0.44, "learning_rate": 1.2403003616085245e-05, "loss": 0.5817, "step": 2541 }, { "epoch": 0.44, "learning_rate": 1.239755665918824e-05, "loss": 0.5646, "step": 2542 }, { "epoch": 0.44, "learning_rate": 1.2392108947463802e-05, "loss": 0.5621, "step": 2543 }, { "epoch": 0.44, "learning_rate": 1.2386660482627034e-05, "loss": 0.5746, "step": 2544 }, { "epoch": 0.44, "learning_rate": 1.2381211266393294e-05, "loss": 0.5624, "step": 2545 }, { "epoch": 0.44, "learning_rate": 1.2375761300478168e-05, "loss": 0.5734, "step": 2546 }, { "epoch": 0.44, "learning_rate": 1.237031058659748e-05, "loss": 0.5772, "step": 2547 }, { "epoch": 0.44, "learning_rate": 1.2364859126467288e-05, "loss": 0.5964, "step": 2548 }, { "epoch": 0.44, "learning_rate": 1.2359406921803887e-05, "loss": 0.5695, "step": 2549 }, { "epoch": 0.44, "learning_rate": 1.2353953974323807e-05, "loss": 0.5936, "step": 2550 }, { "epoch": 0.44, "learning_rate": 1.2348500285743802e-05, "loss": 0.5796, "step": 2551 }, { "epoch": 0.44, "learning_rate": 1.2343045857780877e-05, "loss": 0.5748, "step": 2552 }, { "epoch": 0.44, "learning_rate": 1.2337590692152258e-05, "loss": 0.5874, "step": 2553 }, { "epoch": 0.44, "learning_rate": 1.2332134790575396e-05, "loss": 0.5492, "step": 2554 }, { "epoch": 0.44, "learning_rate": 1.2326678154767997e-05, "loss": 0.5731, "step": 2555 }, { "epoch": 0.44, "learning_rate": 1.2321220786447977e-05, "loss": 0.5789, "step": 2556 }, { "epoch": 0.44, "learning_rate": 1.231576268733349e-05, "loss": 0.5789, "step": 2557 }, { "epoch": 0.44, "learning_rate": 1.2310303859142923e-05, "loss": 0.5777, "step": 2558 }, { "epoch": 0.44, "learning_rate": 1.2304844303594885e-05, "loss": 0.5819, "step": 2559 }, { "epoch": 0.44, "learning_rate": 1.2299384022408223e-05, "loss": 0.557, "step": 2560 }, { "epoch": 0.44, "learning_rate": 1.2293923017302004e-05, "loss": 0.5785, "step": 2561 }, { "epoch": 0.44, "learning_rate": 1.2288461289995529e-05, "loss": 0.5668, "step": 2562 }, { "epoch": 0.44, "learning_rate": 1.2282998842208324e-05, "loss": 0.5746, "step": 2563 }, { "epoch": 0.44, "learning_rate": 1.2277535675660143e-05, "loss": 0.5699, "step": 2564 }, { "epoch": 0.44, "learning_rate": 1.2272071792070963e-05, "loss": 0.5808, "step": 2565 }, { "epoch": 0.44, "learning_rate": 1.2266607193160993e-05, "loss": 0.5826, "step": 2566 }, { "epoch": 0.44, "learning_rate": 1.226114188065066e-05, "loss": 0.5845, "step": 2567 }, { "epoch": 0.44, "learning_rate": 1.2255675856260621e-05, "loss": 0.568, "step": 2568 }, { "epoch": 0.45, "learning_rate": 1.2250209121711751e-05, "loss": 0.5737, "step": 2569 }, { "epoch": 0.45, "learning_rate": 1.2244741678725156e-05, "loss": 0.5862, "step": 2570 }, { "epoch": 0.45, "learning_rate": 1.2239273529022167e-05, "loss": 0.571, "step": 2571 }, { "epoch": 0.45, "learning_rate": 1.223380467432432e-05, "loss": 0.5722, "step": 2572 }, { "epoch": 0.45, "learning_rate": 1.2228335116353396e-05, "loss": 0.5897, "step": 2573 }, { "epoch": 0.45, "learning_rate": 1.2222864856831383e-05, "loss": 0.5686, "step": 2574 }, { "epoch": 0.45, "learning_rate": 1.221739389748049e-05, "loss": 0.5814, "step": 2575 }, { "epoch": 0.45, "learning_rate": 1.2211922240023155e-05, "loss": 0.5726, "step": 2576 }, { "epoch": 0.45, "learning_rate": 1.2206449886182025e-05, "loss": 0.5728, "step": 2577 }, { "epoch": 0.45, "learning_rate": 1.220097683767998e-05, "loss": 0.5853, "step": 2578 }, { "epoch": 0.45, "learning_rate": 1.2195503096240104e-05, "loss": 0.5812, "step": 2579 }, { "epoch": 0.45, "learning_rate": 1.2190028663585706e-05, "loss": 0.5568, "step": 2580 }, { "epoch": 0.45, "learning_rate": 1.2184553541440317e-05, "loss": 0.5708, "step": 2581 }, { "epoch": 0.45, "learning_rate": 1.2179077731527676e-05, "loss": 0.5792, "step": 2582 }, { "epoch": 0.45, "learning_rate": 1.217360123557174e-05, "loss": 0.5798, "step": 2583 }, { "epoch": 0.45, "learning_rate": 1.2168124055296693e-05, "loss": 0.5669, "step": 2584 }, { "epoch": 0.45, "learning_rate": 1.2162646192426924e-05, "loss": 0.5892, "step": 2585 }, { "epoch": 0.45, "learning_rate": 1.2157167648687034e-05, "loss": 0.5797, "step": 2586 }, { "epoch": 0.45, "learning_rate": 1.2151688425801848e-05, "loss": 0.5599, "step": 2587 }, { "epoch": 0.45, "learning_rate": 1.2146208525496399e-05, "loss": 0.5703, "step": 2588 }, { "epoch": 0.45, "learning_rate": 1.2140727949495937e-05, "loss": 0.5708, "step": 2589 }, { "epoch": 0.45, "learning_rate": 1.2135246699525918e-05, "loss": 0.5764, "step": 2590 }, { "epoch": 0.45, "learning_rate": 1.212976477731202e-05, "loss": 0.579, "step": 2591 }, { "epoch": 0.45, "learning_rate": 1.2124282184580125e-05, "loss": 0.56, "step": 2592 }, { "epoch": 0.45, "learning_rate": 1.2118798923056325e-05, "loss": 0.5791, "step": 2593 }, { "epoch": 0.45, "learning_rate": 1.211331499446693e-05, "loss": 0.5647, "step": 2594 }, { "epoch": 0.45, "learning_rate": 1.2107830400538457e-05, "loss": 0.591, "step": 2595 }, { "epoch": 0.45, "learning_rate": 1.2102345142997628e-05, "loss": 0.5946, "step": 2596 }, { "epoch": 0.45, "learning_rate": 1.2096859223571377e-05, "loss": 0.571, "step": 2597 }, { "epoch": 0.45, "learning_rate": 1.2091372643986847e-05, "loss": 0.5688, "step": 2598 }, { "epoch": 0.45, "learning_rate": 1.2085885405971394e-05, "loss": 0.5695, "step": 2599 }, { "epoch": 0.45, "learning_rate": 1.208039751125257e-05, "loss": 0.5692, "step": 2600 }, { "epoch": 0.45, "learning_rate": 1.2074908961558141e-05, "loss": 0.5723, "step": 2601 }, { "epoch": 0.45, "learning_rate": 1.2069419758616077e-05, "loss": 0.5791, "step": 2602 }, { "epoch": 0.45, "learning_rate": 1.2063929904154556e-05, "loss": 0.5946, "step": 2603 }, { "epoch": 0.45, "learning_rate": 1.2058439399901957e-05, "loss": 0.57, "step": 2604 }, { "epoch": 0.45, "learning_rate": 1.2052948247586872e-05, "loss": 0.5592, "step": 2605 }, { "epoch": 0.45, "learning_rate": 1.2047456448938088e-05, "loss": 0.5703, "step": 2606 }, { "epoch": 0.45, "learning_rate": 1.2041964005684596e-05, "loss": 0.5527, "step": 2607 }, { "epoch": 0.45, "learning_rate": 1.2036470919555593e-05, "loss": 0.5937, "step": 2608 }, { "epoch": 0.45, "learning_rate": 1.2030977192280481e-05, "loss": 0.5719, "step": 2609 }, { "epoch": 0.45, "learning_rate": 1.2025482825588861e-05, "loss": 0.5863, "step": 2610 }, { "epoch": 0.45, "learning_rate": 1.2019987821210532e-05, "loss": 0.5744, "step": 2611 }, { "epoch": 0.45, "learning_rate": 1.2014492180875502e-05, "loss": 0.5737, "step": 2612 }, { "epoch": 0.45, "learning_rate": 1.2008995906313969e-05, "loss": 0.5617, "step": 2613 }, { "epoch": 0.45, "learning_rate": 1.2003498999256336e-05, "loss": 0.5694, "step": 2614 }, { "epoch": 0.45, "learning_rate": 1.199800146143321e-05, "loss": 0.57, "step": 2615 }, { "epoch": 0.45, "learning_rate": 1.1992503294575385e-05, "loss": 0.5634, "step": 2616 }, { "epoch": 0.45, "learning_rate": 1.1987004500413862e-05, "loss": 0.5801, "step": 2617 }, { "epoch": 0.45, "learning_rate": 1.198150508067984e-05, "loss": 0.5936, "step": 2618 }, { "epoch": 0.45, "learning_rate": 1.1976005037104707e-05, "loss": 0.5784, "step": 2619 }, { "epoch": 0.45, "learning_rate": 1.197050437142006e-05, "loss": 0.5749, "step": 2620 }, { "epoch": 0.45, "learning_rate": 1.196500308535768e-05, "loss": 0.5825, "step": 2621 }, { "epoch": 0.45, "learning_rate": 1.1959501180649545e-05, "loss": 0.5788, "step": 2622 }, { "epoch": 0.45, "learning_rate": 1.1953998659027832e-05, "loss": 0.5641, "step": 2623 }, { "epoch": 0.45, "learning_rate": 1.194849552222491e-05, "loss": 0.5852, "step": 2624 }, { "epoch": 0.45, "learning_rate": 1.1942991771973347e-05, "loss": 0.5675, "step": 2625 }, { "epoch": 0.45, "learning_rate": 1.1937487410005895e-05, "loss": 0.5668, "step": 2626 }, { "epoch": 0.46, "learning_rate": 1.1931982438055506e-05, "loss": 0.5584, "step": 2627 }, { "epoch": 0.46, "learning_rate": 1.1926476857855316e-05, "loss": 0.578, "step": 2628 }, { "epoch": 0.46, "learning_rate": 1.1920970671138661e-05, "loss": 0.5742, "step": 2629 }, { "epoch": 0.46, "learning_rate": 1.1915463879639062e-05, "loss": 0.5901, "step": 2630 }, { "epoch": 0.46, "learning_rate": 1.1909956485090239e-05, "loss": 0.5709, "step": 2631 }, { "epoch": 0.46, "learning_rate": 1.190444848922609e-05, "loss": 0.5734, "step": 2632 }, { "epoch": 0.46, "learning_rate": 1.1898939893780711e-05, "loss": 0.5773, "step": 2633 }, { "epoch": 0.46, "learning_rate": 1.1893430700488384e-05, "loss": 0.5786, "step": 2634 }, { "epoch": 0.46, "learning_rate": 1.1887920911083574e-05, "loss": 0.5582, "step": 2635 }, { "epoch": 0.46, "learning_rate": 1.1882410527300945e-05, "loss": 0.5781, "step": 2636 }, { "epoch": 0.46, "learning_rate": 1.1876899550875338e-05, "loss": 0.5722, "step": 2637 }, { "epoch": 0.46, "learning_rate": 1.1871387983541789e-05, "loss": 0.571, "step": 2638 }, { "epoch": 0.46, "learning_rate": 1.1865875827035507e-05, "loss": 0.5819, "step": 2639 }, { "epoch": 0.46, "learning_rate": 1.1860363083091904e-05, "loss": 0.5852, "step": 2640 }, { "epoch": 0.46, "learning_rate": 1.1854849753446565e-05, "loss": 0.5802, "step": 2641 }, { "epoch": 0.46, "learning_rate": 1.1849335839835259e-05, "loss": 0.5664, "step": 2642 }, { "epoch": 0.46, "learning_rate": 1.1843821343993945e-05, "loss": 0.5809, "step": 2643 }, { "epoch": 0.46, "learning_rate": 1.1838306267658764e-05, "loss": 0.5757, "step": 2644 }, { "epoch": 0.46, "learning_rate": 1.1832790612566034e-05, "loss": 0.574, "step": 2645 }, { "epoch": 0.46, "learning_rate": 1.1827274380452265e-05, "loss": 0.5685, "step": 2646 }, { "epoch": 0.46, "learning_rate": 1.1821757573054138e-05, "loss": 0.5686, "step": 2647 }, { "epoch": 0.46, "learning_rate": 1.1816240192108524e-05, "loss": 0.5901, "step": 2648 }, { "epoch": 0.46, "learning_rate": 1.1810722239352468e-05, "loss": 0.5867, "step": 2649 }, { "epoch": 0.46, "learning_rate": 1.1805203716523197e-05, "loss": 0.566, "step": 2650 }, { "epoch": 0.46, "learning_rate": 1.179968462535812e-05, "loss": 0.5958, "step": 2651 }, { "epoch": 0.46, "learning_rate": 1.1794164967594827e-05, "loss": 0.5779, "step": 2652 }, { "epoch": 0.46, "learning_rate": 1.1788644744971076e-05, "loss": 0.5706, "step": 2653 }, { "epoch": 0.46, "learning_rate": 1.1783123959224816e-05, "loss": 0.5665, "step": 2654 }, { "epoch": 0.46, "learning_rate": 1.177760261209416e-05, "loss": 0.5758, "step": 2655 }, { "epoch": 0.46, "learning_rate": 1.1772080705317409e-05, "loss": 0.5743, "step": 2656 }, { "epoch": 0.46, "learning_rate": 1.1766558240633035e-05, "loss": 0.5775, "step": 2657 }, { "epoch": 0.46, "learning_rate": 1.1761035219779683e-05, "loss": 0.5939, "step": 2658 }, { "epoch": 0.46, "learning_rate": 1.1755511644496183e-05, "loss": 0.5802, "step": 2659 }, { "epoch": 0.46, "learning_rate": 1.1749987516521523e-05, "loss": 0.5904, "step": 2660 }, { "epoch": 0.46, "learning_rate": 1.1744462837594886e-05, "loss": 0.5757, "step": 2661 }, { "epoch": 0.46, "learning_rate": 1.1738937609455615e-05, "loss": 0.5692, "step": 2662 }, { "epoch": 0.46, "learning_rate": 1.173341183384322e-05, "loss": 0.5937, "step": 2663 }, { "epoch": 0.46, "learning_rate": 1.1727885512497399e-05, "loss": 0.5975, "step": 2664 }, { "epoch": 0.46, "learning_rate": 1.1722358647158013e-05, "loss": 0.5782, "step": 2665 }, { "epoch": 0.46, "learning_rate": 1.1716831239565098e-05, "loss": 0.5781, "step": 2666 }, { "epoch": 0.46, "learning_rate": 1.1711303291458856e-05, "loss": 0.5851, "step": 2667 }, { "epoch": 0.46, "learning_rate": 1.1705774804579662e-05, "loss": 0.5871, "step": 2668 }, { "epoch": 0.46, "learning_rate": 1.170024578066806e-05, "loss": 0.5931, "step": 2669 }, { "epoch": 0.46, "learning_rate": 1.1694716221464764e-05, "loss": 0.5597, "step": 2670 }, { "epoch": 0.46, "learning_rate": 1.1689186128710654e-05, "loss": 0.563, "step": 2671 }, { "epoch": 0.46, "learning_rate": 1.1683655504146785e-05, "loss": 0.5879, "step": 2672 }, { "epoch": 0.46, "learning_rate": 1.1678124349514366e-05, "loss": 0.5765, "step": 2673 }, { "epoch": 0.46, "learning_rate": 1.1672592666554787e-05, "loss": 0.5862, "step": 2674 }, { "epoch": 0.46, "learning_rate": 1.16670604570096e-05, "loss": 0.5824, "step": 2675 }, { "epoch": 0.46, "learning_rate": 1.166152772262052e-05, "loss": 0.5705, "step": 2676 }, { "epoch": 0.46, "learning_rate": 1.1655994465129425e-05, "loss": 0.5884, "step": 2677 }, { "epoch": 0.46, "learning_rate": 1.1650460686278366e-05, "loss": 0.5887, "step": 2678 }, { "epoch": 0.46, "learning_rate": 1.1644926387809548e-05, "loss": 0.575, "step": 2679 }, { "epoch": 0.46, "learning_rate": 1.163939157146535e-05, "loss": 0.5872, "step": 2680 }, { "epoch": 0.46, "learning_rate": 1.1633856238988307e-05, "loss": 0.5642, "step": 2681 }, { "epoch": 0.46, "learning_rate": 1.1628320392121118e-05, "loss": 0.5731, "step": 2682 }, { "epoch": 0.46, "learning_rate": 1.1622784032606646e-05, "loss": 0.5659, "step": 2683 }, { "epoch": 0.46, "learning_rate": 1.1617247162187912e-05, "loss": 0.587, "step": 2684 }, { "epoch": 0.47, "learning_rate": 1.1611709782608103e-05, "loss": 0.5716, "step": 2685 }, { "epoch": 0.47, "learning_rate": 1.1606171895610556e-05, "loss": 0.6019, "step": 2686 }, { "epoch": 0.47, "learning_rate": 1.1600633502938781e-05, "loss": 0.5789, "step": 2687 }, { "epoch": 0.47, "learning_rate": 1.1595094606336442e-05, "loss": 0.5667, "step": 2688 }, { "epoch": 0.47, "learning_rate": 1.1589555207547357e-05, "loss": 0.5794, "step": 2689 }, { "epoch": 0.47, "learning_rate": 1.1584015308315506e-05, "loss": 0.6048, "step": 2690 }, { "epoch": 0.47, "learning_rate": 1.1578474910385026e-05, "loss": 0.5718, "step": 2691 }, { "epoch": 0.47, "learning_rate": 1.1572934015500215e-05, "loss": 0.5838, "step": 2692 }, { "epoch": 0.47, "learning_rate": 1.156739262540552e-05, "loss": 0.5771, "step": 2693 }, { "epoch": 0.47, "learning_rate": 1.1561850741845547e-05, "loss": 0.5798, "step": 2694 }, { "epoch": 0.47, "learning_rate": 1.1556308366565058e-05, "loss": 0.5843, "step": 2695 }, { "epoch": 0.47, "learning_rate": 1.1550765501308976e-05, "loss": 0.5792, "step": 2696 }, { "epoch": 0.47, "learning_rate": 1.1545222147822368e-05, "loss": 0.5699, "step": 2697 }, { "epoch": 0.47, "learning_rate": 1.1539678307850458e-05, "loss": 0.5666, "step": 2698 }, { "epoch": 0.47, "learning_rate": 1.153413398313862e-05, "loss": 0.5743, "step": 2699 }, { "epoch": 0.47, "learning_rate": 1.152858917543239e-05, "loss": 0.5745, "step": 2700 }, { "epoch": 0.47, "learning_rate": 1.1523043886477451e-05, "loss": 0.5804, "step": 2701 }, { "epoch": 0.47, "learning_rate": 1.1517498118019637e-05, "loss": 0.5869, "step": 2702 }, { "epoch": 0.47, "learning_rate": 1.1511951871804929e-05, "loss": 0.5698, "step": 2703 }, { "epoch": 0.47, "learning_rate": 1.150640514957947e-05, "loss": 0.566, "step": 2704 }, { "epoch": 0.47, "learning_rate": 1.1500857953089535e-05, "loss": 0.5649, "step": 2705 }, { "epoch": 0.47, "learning_rate": 1.1495310284081566e-05, "loss": 0.571, "step": 2706 }, { "epoch": 0.47, "learning_rate": 1.1489762144302143e-05, "loss": 0.5858, "step": 2707 }, { "epoch": 0.47, "learning_rate": 1.1484213535498002e-05, "loss": 0.5619, "step": 2708 }, { "epoch": 0.47, "learning_rate": 1.1478664459416021e-05, "loss": 0.5638, "step": 2709 }, { "epoch": 0.47, "learning_rate": 1.1473114917803221e-05, "loss": 0.5664, "step": 2710 }, { "epoch": 0.47, "learning_rate": 1.1467564912406784e-05, "loss": 0.5743, "step": 2711 }, { "epoch": 0.47, "learning_rate": 1.1462014444974022e-05, "loss": 0.5771, "step": 2712 }, { "epoch": 0.47, "learning_rate": 1.14564635172524e-05, "loss": 0.5896, "step": 2713 }, { "epoch": 0.47, "learning_rate": 1.145091213098953e-05, "loss": 0.5615, "step": 2714 }, { "epoch": 0.47, "learning_rate": 1.1445360287933165e-05, "loss": 0.5629, "step": 2715 }, { "epoch": 0.47, "learning_rate": 1.1439807989831204e-05, "loss": 0.5721, "step": 2716 }, { "epoch": 0.47, "learning_rate": 1.1434255238431684e-05, "loss": 0.5758, "step": 2717 }, { "epoch": 0.47, "learning_rate": 1.1428702035482793e-05, "loss": 0.5746, "step": 2718 }, { "epoch": 0.47, "learning_rate": 1.1423148382732854e-05, "loss": 0.5726, "step": 2719 }, { "epoch": 0.47, "learning_rate": 1.1417594281930333e-05, "loss": 0.5963, "step": 2720 }, { "epoch": 0.47, "learning_rate": 1.1412039734823842e-05, "loss": 0.5781, "step": 2721 }, { "epoch": 0.47, "learning_rate": 1.1406484743162127e-05, "loss": 0.5791, "step": 2722 }, { "epoch": 0.47, "learning_rate": 1.1400929308694077e-05, "loss": 0.574, "step": 2723 }, { "epoch": 0.47, "learning_rate": 1.1395373433168725e-05, "loss": 0.5865, "step": 2724 }, { "epoch": 0.47, "learning_rate": 1.1389817118335231e-05, "loss": 0.5835, "step": 2725 }, { "epoch": 0.47, "learning_rate": 1.1384260365942905e-05, "loss": 0.5647, "step": 2726 }, { "epoch": 0.47, "learning_rate": 1.137870317774119e-05, "loss": 0.569, "step": 2727 }, { "epoch": 0.47, "learning_rate": 1.1373145555479663e-05, "loss": 0.5513, "step": 2728 }, { "epoch": 0.47, "learning_rate": 1.1367587500908047e-05, "loss": 0.5819, "step": 2729 }, { "epoch": 0.47, "learning_rate": 1.1362029015776185e-05, "loss": 0.5575, "step": 2730 }, { "epoch": 0.47, "learning_rate": 1.1356470101834075e-05, "loss": 0.5679, "step": 2731 }, { "epoch": 0.47, "learning_rate": 1.1350910760831843e-05, "loss": 0.5674, "step": 2732 }, { "epoch": 0.47, "learning_rate": 1.134535099451974e-05, "loss": 0.5683, "step": 2733 }, { "epoch": 0.47, "learning_rate": 1.1339790804648156e-05, "loss": 0.5696, "step": 2734 }, { "epoch": 0.47, "learning_rate": 1.1334230192967625e-05, "loss": 0.5642, "step": 2735 }, { "epoch": 0.47, "learning_rate": 1.1328669161228799e-05, "loss": 0.5875, "step": 2736 }, { "epoch": 0.47, "learning_rate": 1.1323107711182474e-05, "loss": 0.577, "step": 2737 }, { "epoch": 0.47, "learning_rate": 1.1317545844579563e-05, "loss": 0.5769, "step": 2738 }, { "epoch": 0.47, "learning_rate": 1.1311983563171127e-05, "loss": 0.5697, "step": 2739 }, { "epoch": 0.47, "learning_rate": 1.130642086870835e-05, "loss": 0.5697, "step": 2740 }, { "epoch": 0.47, "learning_rate": 1.1300857762942541e-05, "loss": 0.5757, "step": 2741 }, { "epoch": 0.47, "learning_rate": 1.1295294247625144e-05, "loss": 0.5706, "step": 2742 }, { "epoch": 0.48, "learning_rate": 1.1289730324507738e-05, "loss": 0.5723, "step": 2743 }, { "epoch": 0.48, "learning_rate": 1.1284165995342017e-05, "loss": 0.5546, "step": 2744 }, { "epoch": 0.48, "learning_rate": 1.127860126187981e-05, "loss": 0.573, "step": 2745 }, { "epoch": 0.48, "learning_rate": 1.1273036125873081e-05, "loss": 0.571, "step": 2746 }, { "epoch": 0.48, "learning_rate": 1.1267470589073901e-05, "loss": 0.5649, "step": 2747 }, { "epoch": 0.48, "learning_rate": 1.1261904653234486e-05, "loss": 0.5661, "step": 2748 }, { "epoch": 0.48, "learning_rate": 1.1256338320107168e-05, "loss": 0.5815, "step": 2749 }, { "epoch": 0.48, "learning_rate": 1.1250771591444408e-05, "loss": 0.5523, "step": 2750 }, { "epoch": 0.48, "learning_rate": 1.1245204468998787e-05, "loss": 0.5586, "step": 2751 }, { "epoch": 0.48, "learning_rate": 1.1239636954523014e-05, "loss": 0.5565, "step": 2752 }, { "epoch": 0.48, "learning_rate": 1.123406904976993e-05, "loss": 0.5807, "step": 2753 }, { "epoch": 0.48, "learning_rate": 1.1228500756492473e-05, "loss": 0.5589, "step": 2754 }, { "epoch": 0.48, "learning_rate": 1.1222932076443727e-05, "loss": 0.5769, "step": 2755 }, { "epoch": 0.48, "learning_rate": 1.1217363011376893e-05, "loss": 0.5724, "step": 2756 }, { "epoch": 0.48, "learning_rate": 1.121179356304529e-05, "loss": 0.5716, "step": 2757 }, { "epoch": 0.48, "learning_rate": 1.1206223733202357e-05, "loss": 0.5849, "step": 2758 }, { "epoch": 0.48, "learning_rate": 1.1200653523601652e-05, "loss": 0.5815, "step": 2759 }, { "epoch": 0.48, "learning_rate": 1.119508293599686e-05, "loss": 0.5679, "step": 2760 }, { "epoch": 0.48, "learning_rate": 1.1189511972141776e-05, "loss": 0.5695, "step": 2761 }, { "epoch": 0.48, "learning_rate": 1.1183940633790315e-05, "loss": 0.5741, "step": 2762 }, { "epoch": 0.48, "learning_rate": 1.117836892269652e-05, "loss": 0.5798, "step": 2763 }, { "epoch": 0.48, "learning_rate": 1.1172796840614536e-05, "loss": 0.5717, "step": 2764 }, { "epoch": 0.48, "learning_rate": 1.1167224389298639e-05, "loss": 0.5717, "step": 2765 }, { "epoch": 0.48, "learning_rate": 1.1161651570503212e-05, "loss": 0.5657, "step": 2766 }, { "epoch": 0.48, "learning_rate": 1.1156078385982754e-05, "loss": 0.5802, "step": 2767 }, { "epoch": 0.48, "learning_rate": 1.1150504837491884e-05, "loss": 0.5671, "step": 2768 }, { "epoch": 0.48, "learning_rate": 1.1144930926785334e-05, "loss": 0.5611, "step": 2769 }, { "epoch": 0.48, "learning_rate": 1.1139356655617945e-05, "loss": 0.5804, "step": 2770 }, { "epoch": 0.48, "learning_rate": 1.1133782025744685e-05, "loss": 0.5596, "step": 2771 }, { "epoch": 0.48, "learning_rate": 1.1128207038920614e-05, "loss": 0.594, "step": 2772 }, { "epoch": 0.48, "learning_rate": 1.1122631696900918e-05, "loss": 0.5876, "step": 2773 }, { "epoch": 0.48, "learning_rate": 1.1117056001440905e-05, "loss": 0.575, "step": 2774 }, { "epoch": 0.48, "learning_rate": 1.1111479954295968e-05, "loss": 0.5587, "step": 2775 }, { "epoch": 0.48, "learning_rate": 1.1105903557221629e-05, "loss": 0.591, "step": 2776 }, { "epoch": 0.48, "learning_rate": 1.1100326811973519e-05, "loss": 0.5793, "step": 2777 }, { "epoch": 0.48, "learning_rate": 1.109474972030737e-05, "loss": 0.5636, "step": 2778 }, { "epoch": 0.48, "learning_rate": 1.1089172283979035e-05, "loss": 0.5762, "step": 2779 }, { "epoch": 0.48, "learning_rate": 1.1083594504744464e-05, "loss": 0.5581, "step": 2780 }, { "epoch": 0.48, "learning_rate": 1.1078016384359725e-05, "loss": 0.5737, "step": 2781 }, { "epoch": 0.48, "learning_rate": 1.1072437924580986e-05, "loss": 0.5824, "step": 2782 }, { "epoch": 0.48, "learning_rate": 1.1066859127164521e-05, "loss": 0.5979, "step": 2783 }, { "epoch": 0.48, "learning_rate": 1.106127999386672e-05, "loss": 0.5672, "step": 2784 }, { "epoch": 0.48, "learning_rate": 1.1055700526444067e-05, "loss": 0.5719, "step": 2785 }, { "epoch": 0.48, "learning_rate": 1.105012072665316e-05, "loss": 0.5693, "step": 2786 }, { "epoch": 0.48, "learning_rate": 1.1044540596250697e-05, "loss": 0.5773, "step": 2787 }, { "epoch": 0.48, "learning_rate": 1.1038960136993482e-05, "loss": 0.5767, "step": 2788 }, { "epoch": 0.48, "learning_rate": 1.103337935063842e-05, "loss": 0.579, "step": 2789 }, { "epoch": 0.48, "learning_rate": 1.1027798238942523e-05, "loss": 0.5765, "step": 2790 }, { "epoch": 0.48, "learning_rate": 1.1022216803662902e-05, "loss": 0.5724, "step": 2791 }, { "epoch": 0.48, "learning_rate": 1.1016635046556773e-05, "loss": 0.5627, "step": 2792 }, { "epoch": 0.48, "learning_rate": 1.1011052969381448e-05, "loss": 0.5826, "step": 2793 }, { "epoch": 0.48, "learning_rate": 1.1005470573894347e-05, "loss": 0.5504, "step": 2794 }, { "epoch": 0.48, "learning_rate": 1.0999887861852985e-05, "loss": 0.5829, "step": 2795 }, { "epoch": 0.48, "learning_rate": 1.0994304835014974e-05, "loss": 0.5674, "step": 2796 }, { "epoch": 0.48, "learning_rate": 1.098872149513804e-05, "loss": 0.5675, "step": 2797 }, { "epoch": 0.48, "learning_rate": 1.0983137843979982e-05, "loss": 0.5701, "step": 2798 }, { "epoch": 0.48, "learning_rate": 1.0977553883298723e-05, "loss": 0.5737, "step": 2799 }, { "epoch": 0.49, "learning_rate": 1.0971969614852271e-05, "loss": 0.5484, "step": 2800 }, { "epoch": 0.49, "learning_rate": 1.0966385040398728e-05, "loss": 0.5911, "step": 2801 }, { "epoch": 0.49, "learning_rate": 1.0960800161696301e-05, "loss": 0.5852, "step": 2802 }, { "epoch": 0.49, "learning_rate": 1.0955214980503284e-05, "loss": 0.5769, "step": 2803 }, { "epoch": 0.49, "learning_rate": 1.0949629498578072e-05, "loss": 0.5725, "step": 2804 }, { "epoch": 0.49, "learning_rate": 1.0944043717679154e-05, "loss": 0.5869, "step": 2805 }, { "epoch": 0.49, "learning_rate": 1.093845763956511e-05, "loss": 0.5721, "step": 2806 }, { "epoch": 0.49, "learning_rate": 1.093287126599462e-05, "loss": 0.5707, "step": 2807 }, { "epoch": 0.49, "learning_rate": 1.0927284598726449e-05, "loss": 0.5551, "step": 2808 }, { "epoch": 0.49, "learning_rate": 1.092169763951946e-05, "loss": 0.5658, "step": 2809 }, { "epoch": 0.49, "learning_rate": 1.0916110390132606e-05, "loss": 0.5603, "step": 2810 }, { "epoch": 0.49, "learning_rate": 1.0910522852324931e-05, "loss": 0.5738, "step": 2811 }, { "epoch": 0.49, "learning_rate": 1.090493502785557e-05, "loss": 0.5575, "step": 2812 }, { "epoch": 0.49, "learning_rate": 1.0899346918483753e-05, "loss": 0.5676, "step": 2813 }, { "epoch": 0.49, "learning_rate": 1.0893758525968789e-05, "loss": 0.5644, "step": 2814 }, { "epoch": 0.49, "learning_rate": 1.088816985207009e-05, "loss": 0.564, "step": 2815 }, { "epoch": 0.49, "learning_rate": 1.0882580898547142e-05, "loss": 0.572, "step": 2816 }, { "epoch": 0.49, "learning_rate": 1.0876991667159527e-05, "loss": 0.5643, "step": 2817 }, { "epoch": 0.49, "learning_rate": 1.087140215966692e-05, "loss": 0.5717, "step": 2818 }, { "epoch": 0.49, "learning_rate": 1.0865812377829066e-05, "loss": 0.5667, "step": 2819 }, { "epoch": 0.49, "learning_rate": 1.0860222323405817e-05, "loss": 0.5483, "step": 2820 }, { "epoch": 0.49, "learning_rate": 1.0854631998157097e-05, "loss": 0.5819, "step": 2821 }, { "epoch": 0.49, "learning_rate": 1.0849041403842915e-05, "loss": 0.5703, "step": 2822 }, { "epoch": 0.49, "learning_rate": 1.084345054222338e-05, "loss": 0.5665, "step": 2823 }, { "epoch": 0.49, "learning_rate": 1.0837859415058658e-05, "loss": 0.5531, "step": 2824 }, { "epoch": 0.49, "learning_rate": 1.0832268024109025e-05, "loss": 0.575, "step": 2825 }, { "epoch": 0.49, "learning_rate": 1.082667637113483e-05, "loss": 0.564, "step": 2826 }, { "epoch": 0.49, "learning_rate": 1.0821084457896497e-05, "loss": 0.5676, "step": 2827 }, { "epoch": 0.49, "learning_rate": 1.0815492286154548e-05, "loss": 0.5763, "step": 2828 }, { "epoch": 0.49, "learning_rate": 1.0809899857669569e-05, "loss": 0.5839, "step": 2829 }, { "epoch": 0.49, "learning_rate": 1.080430717420224e-05, "loss": 0.553, "step": 2830 }, { "epoch": 0.49, "learning_rate": 1.0798714237513318e-05, "loss": 0.5713, "step": 2831 }, { "epoch": 0.49, "learning_rate": 1.0793121049363631e-05, "loss": 0.5797, "step": 2832 }, { "epoch": 0.49, "learning_rate": 1.0787527611514096e-05, "loss": 0.5682, "step": 2833 }, { "epoch": 0.49, "learning_rate": 1.0781933925725711e-05, "loss": 0.5673, "step": 2834 }, { "epoch": 0.49, "learning_rate": 1.0776339993759544e-05, "loss": 0.5882, "step": 2835 }, { "epoch": 0.49, "learning_rate": 1.0770745817376741e-05, "loss": 0.5687, "step": 2836 }, { "epoch": 0.49, "learning_rate": 1.0765151398338532e-05, "loss": 0.5575, "step": 2837 }, { "epoch": 0.49, "learning_rate": 1.0759556738406214e-05, "loss": 0.5647, "step": 2838 }, { "epoch": 0.49, "learning_rate": 1.0753961839341169e-05, "loss": 0.5738, "step": 2839 }, { "epoch": 0.49, "learning_rate": 1.0748366702904847e-05, "loss": 0.5911, "step": 2840 }, { "epoch": 0.49, "learning_rate": 1.0742771330858776e-05, "loss": 0.5726, "step": 2841 }, { "epoch": 0.49, "learning_rate": 1.0737175724964559e-05, "loss": 0.5803, "step": 2842 }, { "epoch": 0.49, "learning_rate": 1.0731579886983871e-05, "loss": 0.5675, "step": 2843 }, { "epoch": 0.49, "learning_rate": 1.0725983818678466e-05, "loss": 0.5664, "step": 2844 }, { "epoch": 0.49, "learning_rate": 1.0720387521810154e-05, "loss": 0.5743, "step": 2845 }, { "epoch": 0.49, "learning_rate": 1.0714790998140834e-05, "loss": 0.5879, "step": 2846 }, { "epoch": 0.49, "learning_rate": 1.0709194249432471e-05, "loss": 0.5585, "step": 2847 }, { "epoch": 0.49, "learning_rate": 1.07035972774471e-05, "loss": 0.5663, "step": 2848 }, { "epoch": 0.49, "learning_rate": 1.0698000083946826e-05, "loss": 0.5748, "step": 2849 }, { "epoch": 0.49, "learning_rate": 1.0692402670693825e-05, "loss": 0.5522, "step": 2850 }, { "epoch": 0.49, "learning_rate": 1.0686805039450344e-05, "loss": 0.5711, "step": 2851 }, { "epoch": 0.49, "learning_rate": 1.068120719197869e-05, "loss": 0.5777, "step": 2852 }, { "epoch": 0.49, "learning_rate": 1.0675609130041246e-05, "loss": 0.5678, "step": 2853 }, { "epoch": 0.49, "learning_rate": 1.0670010855400466e-05, "loss": 0.578, "step": 2854 }, { "epoch": 0.49, "learning_rate": 1.0664412369818859e-05, "loss": 0.5757, "step": 2855 }, { "epoch": 0.49, "learning_rate": 1.0658813675059012e-05, "loss": 0.5718, "step": 2856 }, { "epoch": 0.49, "learning_rate": 1.0653214772883571e-05, "loss": 0.5775, "step": 2857 }, { "epoch": 0.5, "learning_rate": 1.0647615665055251e-05, "loss": 0.5799, "step": 2858 }, { "epoch": 0.5, "learning_rate": 1.0642016353336825e-05, "loss": 0.5683, "step": 2859 }, { "epoch": 0.5, "learning_rate": 1.0636416839491141e-05, "loss": 0.5875, "step": 2860 }, { "epoch": 0.5, "learning_rate": 1.06308171252811e-05, "loss": 0.5368, "step": 2861 }, { "epoch": 0.5, "learning_rate": 1.0625217212469678e-05, "loss": 0.5771, "step": 2862 }, { "epoch": 0.5, "learning_rate": 1.0619617102819897e-05, "loss": 0.5839, "step": 2863 }, { "epoch": 0.5, "learning_rate": 1.0614016798094857e-05, "loss": 0.5658, "step": 2864 }, { "epoch": 0.5, "learning_rate": 1.0608416300057715e-05, "loss": 0.5613, "step": 2865 }, { "epoch": 0.5, "learning_rate": 1.0602815610471679e-05, "loss": 0.583, "step": 2866 }, { "epoch": 0.5, "learning_rate": 1.0597214731100031e-05, "loss": 0.5755, "step": 2867 }, { "epoch": 0.5, "learning_rate": 1.0591613663706104e-05, "loss": 0.582, "step": 2868 }, { "epoch": 0.5, "learning_rate": 1.0586012410053293e-05, "loss": 0.5809, "step": 2869 }, { "epoch": 0.5, "learning_rate": 1.0580410971905053e-05, "loss": 0.5694, "step": 2870 }, { "epoch": 0.5, "learning_rate": 1.0574809351024892e-05, "loss": 0.5741, "step": 2871 }, { "epoch": 0.5, "learning_rate": 1.0569207549176388e-05, "loss": 0.5691, "step": 2872 }, { "epoch": 0.5, "learning_rate": 1.0563605568123158e-05, "loss": 0.57, "step": 2873 }, { "epoch": 0.5, "learning_rate": 1.0558003409628887e-05, "loss": 0.5853, "step": 2874 }, { "epoch": 0.5, "learning_rate": 1.0552401075457319e-05, "loss": 0.5629, "step": 2875 }, { "epoch": 0.5, "learning_rate": 1.0546798567372238e-05, "loss": 0.5897, "step": 2876 }, { "epoch": 0.5, "learning_rate": 1.05411958871375e-05, "loss": 0.5563, "step": 2877 }, { "epoch": 0.5, "learning_rate": 1.0535593036517007e-05, "loss": 0.5895, "step": 2878 }, { "epoch": 0.5, "learning_rate": 1.0529990017274714e-05, "loss": 0.5825, "step": 2879 }, { "epoch": 0.5, "learning_rate": 1.0524386831174629e-05, "loss": 0.5717, "step": 2880 }, { "epoch": 0.5, "learning_rate": 1.0518783479980818e-05, "loss": 0.5673, "step": 2881 }, { "epoch": 0.5, "learning_rate": 1.051317996545739e-05, "loss": 0.5759, "step": 2882 }, { "epoch": 0.5, "learning_rate": 1.0507576289368513e-05, "loss": 0.5586, "step": 2883 }, { "epoch": 0.5, "learning_rate": 1.0501972453478405e-05, "loss": 0.5892, "step": 2884 }, { "epoch": 0.5, "learning_rate": 1.049636845955133e-05, "loss": 0.5824, "step": 2885 }, { "epoch": 0.5, "learning_rate": 1.0490764309351604e-05, "loss": 0.5681, "step": 2886 }, { "epoch": 0.5, "learning_rate": 1.0485160004643594e-05, "loss": 0.5801, "step": 2887 }, { "epoch": 0.5, "learning_rate": 1.0479555547191713e-05, "loss": 0.5715, "step": 2888 }, { "epoch": 0.5, "learning_rate": 1.0473950938760424e-05, "loss": 0.5749, "step": 2889 }, { "epoch": 0.5, "learning_rate": 1.0468346181114237e-05, "loss": 0.587, "step": 2890 }, { "epoch": 0.5, "learning_rate": 1.0462741276017711e-05, "loss": 0.5805, "step": 2891 }, { "epoch": 0.5, "learning_rate": 1.0457136225235444e-05, "loss": 0.6027, "step": 2892 }, { "epoch": 0.5, "learning_rate": 1.045153103053209e-05, "loss": 0.5551, "step": 2893 }, { "epoch": 0.5, "learning_rate": 1.0445925693672341e-05, "loss": 0.5768, "step": 2894 }, { "epoch": 0.5, "learning_rate": 1.0440320216420937e-05, "loss": 0.5653, "step": 2895 }, { "epoch": 0.5, "learning_rate": 1.043471460054266e-05, "loss": 0.584, "step": 2896 }, { "epoch": 0.5, "learning_rate": 1.0429108847802334e-05, "loss": 0.5537, "step": 2897 }, { "epoch": 0.5, "learning_rate": 1.0423502959964838e-05, "loss": 0.5767, "step": 2898 }, { "epoch": 0.5, "learning_rate": 1.0417896938795075e-05, "loss": 0.5449, "step": 2899 }, { "epoch": 0.5, "learning_rate": 1.0412290786058008e-05, "loss": 0.5823, "step": 2900 }, { "epoch": 0.5, "learning_rate": 1.040668450351863e-05, "loss": 0.5635, "step": 2901 }, { "epoch": 0.5, "learning_rate": 1.0401078092941972e-05, "loss": 0.5617, "step": 2902 }, { "epoch": 0.5, "learning_rate": 1.0395471556093118e-05, "loss": 0.574, "step": 2903 }, { "epoch": 0.5, "learning_rate": 1.0389864894737182e-05, "loss": 0.5881, "step": 2904 }, { "epoch": 0.5, "learning_rate": 1.0384258110639323e-05, "loss": 0.5696, "step": 2905 }, { "epoch": 0.5, "learning_rate": 1.0378651205564732e-05, "loss": 0.5807, "step": 2906 }, { "epoch": 0.5, "learning_rate": 1.0373044181278646e-05, "loss": 0.5718, "step": 2907 }, { "epoch": 0.5, "learning_rate": 1.0367437039546331e-05, "loss": 0.5813, "step": 2908 }, { "epoch": 0.5, "learning_rate": 1.0361829782133097e-05, "loss": 0.5917, "step": 2909 }, { "epoch": 0.5, "learning_rate": 1.0356222410804286e-05, "loss": 0.5677, "step": 2910 }, { "epoch": 0.5, "learning_rate": 1.035061492732528e-05, "loss": 0.5699, "step": 2911 }, { "epoch": 0.5, "learning_rate": 1.0345007333461493e-05, "loss": 0.5755, "step": 2912 }, { "epoch": 0.5, "learning_rate": 1.0339399630978373e-05, "loss": 0.5588, "step": 2913 }, { "epoch": 0.5, "learning_rate": 1.033379182164141e-05, "loss": 0.569, "step": 2914 }, { "epoch": 0.5, "learning_rate": 1.0328183907216114e-05, "loss": 0.5766, "step": 2915 }, { "epoch": 0.51, "learning_rate": 1.0322575889468036e-05, "loss": 0.5559, "step": 2916 }, { "epoch": 0.51, "learning_rate": 1.0316967770162765e-05, "loss": 0.5666, "step": 2917 }, { "epoch": 0.51, "learning_rate": 1.0311359551065912e-05, "loss": 0.5826, "step": 2918 }, { "epoch": 0.51, "learning_rate": 1.0305751233943124e-05, "loss": 0.5692, "step": 2919 }, { "epoch": 0.51, "learning_rate": 1.0300142820560079e-05, "loss": 0.5715, "step": 2920 }, { "epoch": 0.51, "learning_rate": 1.0294534312682488e-05, "loss": 0.5562, "step": 2921 }, { "epoch": 0.51, "learning_rate": 1.0288925712076081e-05, "loss": 0.5761, "step": 2922 }, { "epoch": 0.51, "learning_rate": 1.0283317020506629e-05, "loss": 0.5789, "step": 2923 }, { "epoch": 0.51, "learning_rate": 1.0277708239739925e-05, "loss": 0.5709, "step": 2924 }, { "epoch": 0.51, "learning_rate": 1.02720993715418e-05, "loss": 0.5681, "step": 2925 }, { "epoch": 0.51, "learning_rate": 1.0266490417678094e-05, "loss": 0.5814, "step": 2926 }, { "epoch": 0.51, "learning_rate": 1.0260881379914694e-05, "loss": 0.5692, "step": 2927 }, { "epoch": 0.51, "learning_rate": 1.0255272260017502e-05, "loss": 0.5717, "step": 2928 }, { "epoch": 0.51, "learning_rate": 1.0249663059752442e-05, "loss": 0.5735, "step": 2929 }, { "epoch": 0.51, "learning_rate": 1.0244053780885477e-05, "loss": 0.5811, "step": 2930 }, { "epoch": 0.51, "learning_rate": 1.0238444425182581e-05, "loss": 0.5759, "step": 2931 }, { "epoch": 0.51, "learning_rate": 1.0232834994409766e-05, "loss": 0.5801, "step": 2932 }, { "epoch": 0.51, "learning_rate": 1.0227225490333053e-05, "loss": 0.5704, "step": 2933 }, { "epoch": 0.51, "learning_rate": 1.0221615914718494e-05, "loss": 0.5559, "step": 2934 }, { "epoch": 0.51, "learning_rate": 1.021600626933217e-05, "loss": 0.5575, "step": 2935 }, { "epoch": 0.51, "learning_rate": 1.0210396555940167e-05, "loss": 0.5736, "step": 2936 }, { "epoch": 0.51, "learning_rate": 1.0204786776308606e-05, "loss": 0.5646, "step": 2937 }, { "epoch": 0.51, "learning_rate": 1.0199176932203628e-05, "loss": 0.5704, "step": 2938 }, { "epoch": 0.51, "learning_rate": 1.0193567025391385e-05, "loss": 0.5697, "step": 2939 }, { "epoch": 0.51, "learning_rate": 1.018795705763806e-05, "loss": 0.5578, "step": 2940 }, { "epoch": 0.51, "learning_rate": 1.0182347030709845e-05, "loss": 0.5608, "step": 2941 }, { "epoch": 0.51, "learning_rate": 1.0176736946372963e-05, "loss": 0.5608, "step": 2942 }, { "epoch": 0.51, "learning_rate": 1.0171126806393642e-05, "loss": 0.5695, "step": 2943 }, { "epoch": 0.51, "learning_rate": 1.0165516612538134e-05, "loss": 0.5709, "step": 2944 }, { "epoch": 0.51, "learning_rate": 1.0159906366572709e-05, "loss": 0.5651, "step": 2945 }, { "epoch": 0.51, "learning_rate": 1.0154296070263649e-05, "loss": 0.548, "step": 2946 }, { "epoch": 0.51, "learning_rate": 1.0148685725377256e-05, "loss": 0.5557, "step": 2947 }, { "epoch": 0.51, "learning_rate": 1.014307533367985e-05, "loss": 0.5833, "step": 2948 }, { "epoch": 0.51, "learning_rate": 1.0137464896937756e-05, "loss": 0.5717, "step": 2949 }, { "epoch": 0.51, "learning_rate": 1.0131854416917319e-05, "loss": 0.5773, "step": 2950 }, { "epoch": 0.51, "learning_rate": 1.0126243895384902e-05, "loss": 0.5844, "step": 2951 }, { "epoch": 0.51, "learning_rate": 1.012063333410687e-05, "loss": 0.5678, "step": 2952 }, { "epoch": 0.51, "learning_rate": 1.0115022734849614e-05, "loss": 0.5676, "step": 2953 }, { "epoch": 0.51, "learning_rate": 1.010941209937952e-05, "loss": 0.5687, "step": 2954 }, { "epoch": 0.51, "learning_rate": 1.0103801429463005e-05, "loss": 0.5724, "step": 2955 }, { "epoch": 0.51, "learning_rate": 1.0098190726866486e-05, "loss": 0.5588, "step": 2956 }, { "epoch": 0.51, "learning_rate": 1.0092579993356386e-05, "loss": 0.5925, "step": 2957 }, { "epoch": 0.51, "learning_rate": 1.0086969230699146e-05, "loss": 0.5739, "step": 2958 }, { "epoch": 0.51, "learning_rate": 1.0081358440661215e-05, "loss": 0.5818, "step": 2959 }, { "epoch": 0.51, "learning_rate": 1.0075747625009045e-05, "loss": 0.5954, "step": 2960 }, { "epoch": 0.51, "learning_rate": 1.0070136785509106e-05, "loss": 0.5706, "step": 2961 }, { "epoch": 0.51, "learning_rate": 1.0064525923927864e-05, "loss": 0.5446, "step": 2962 }, { "epoch": 0.51, "learning_rate": 1.00589150420318e-05, "loss": 0.5759, "step": 2963 }, { "epoch": 0.51, "learning_rate": 1.0053304141587398e-05, "loss": 0.5693, "step": 2964 }, { "epoch": 0.51, "learning_rate": 1.004769322436115e-05, "loss": 0.5758, "step": 2965 }, { "epoch": 0.51, "learning_rate": 1.0042082292119553e-05, "loss": 0.5565, "step": 2966 }, { "epoch": 0.51, "learning_rate": 1.0036471346629101e-05, "loss": 0.5706, "step": 2967 }, { "epoch": 0.51, "learning_rate": 1.0030860389656306e-05, "loss": 0.5798, "step": 2968 }, { "epoch": 0.51, "learning_rate": 1.0025249422967675e-05, "loss": 0.5712, "step": 2969 }, { "epoch": 0.51, "learning_rate": 1.001963844832972e-05, "loss": 0.5544, "step": 2970 }, { "epoch": 0.51, "learning_rate": 1.0014027467508954e-05, "loss": 0.5708, "step": 2971 }, { "epoch": 0.51, "learning_rate": 1.0008416482271893e-05, "loss": 0.5773, "step": 2972 }, { "epoch": 0.51, "learning_rate": 1.0002805494385051e-05, "loss": 0.5794, "step": 2973 }, { "epoch": 0.52, "learning_rate": 9.997194505614952e-06, "loss": 0.5709, "step": 2974 }, { "epoch": 0.52, "learning_rate": 9.991583517728114e-06, "loss": 0.5681, "step": 2975 }, { "epoch": 0.52, "learning_rate": 9.985972532491051e-06, "loss": 0.5577, "step": 2976 }, { "epoch": 0.52, "learning_rate": 9.980361551670285e-06, "loss": 0.5683, "step": 2977 }, { "epoch": 0.52, "learning_rate": 9.974750577032327e-06, "loss": 0.5604, "step": 2978 }, { "epoch": 0.52, "learning_rate": 9.969139610343696e-06, "loss": 0.5833, "step": 2979 }, { "epoch": 0.52, "learning_rate": 9.9635286533709e-06, "loss": 0.5597, "step": 2980 }, { "epoch": 0.52, "learning_rate": 9.957917707880452e-06, "loss": 0.5656, "step": 2981 }, { "epoch": 0.52, "learning_rate": 9.952306775638852e-06, "loss": 0.5957, "step": 2982 }, { "epoch": 0.52, "learning_rate": 9.946695858412603e-06, "loss": 0.5758, "step": 2983 }, { "epoch": 0.52, "learning_rate": 9.941084957968202e-06, "loss": 0.5637, "step": 2984 }, { "epoch": 0.52, "learning_rate": 9.935474076072136e-06, "loss": 0.5848, "step": 2985 }, { "epoch": 0.52, "learning_rate": 9.929863214490894e-06, "loss": 0.5754, "step": 2986 }, { "epoch": 0.52, "learning_rate": 9.924252374990955e-06, "loss": 0.5765, "step": 2987 }, { "epoch": 0.52, "learning_rate": 9.91864155933879e-06, "loss": 0.5742, "step": 2988 }, { "epoch": 0.52, "learning_rate": 9.913030769300858e-06, "loss": 0.5681, "step": 2989 }, { "epoch": 0.52, "learning_rate": 9.907420006643619e-06, "loss": 0.5778, "step": 2990 }, { "epoch": 0.52, "learning_rate": 9.901809273133518e-06, "loss": 0.5647, "step": 2991 }, { "epoch": 0.52, "learning_rate": 9.896198570536998e-06, "loss": 0.5747, "step": 2992 }, { "epoch": 0.52, "learning_rate": 9.890587900620483e-06, "loss": 0.5734, "step": 2993 }, { "epoch": 0.52, "learning_rate": 9.884977265150391e-06, "loss": 0.5778, "step": 2994 }, { "epoch": 0.52, "learning_rate": 9.879366665893133e-06, "loss": 0.577, "step": 2995 }, { "epoch": 0.52, "learning_rate": 9.873756104615101e-06, "loss": 0.5627, "step": 2996 }, { "epoch": 0.52, "learning_rate": 9.868145583082683e-06, "loss": 0.5704, "step": 2997 }, { "epoch": 0.52, "learning_rate": 9.862535103062246e-06, "loss": 0.5752, "step": 2998 }, { "epoch": 0.52, "learning_rate": 9.85692466632015e-06, "loss": 0.5897, "step": 2999 }, { "epoch": 0.52, "learning_rate": 9.851314274622744e-06, "loss": 0.568, "step": 3000 }, { "epoch": 0.52, "learning_rate": 9.845703929736351e-06, "loss": 0.5675, "step": 3001 }, { "epoch": 0.52, "learning_rate": 9.840093633427296e-06, "loss": 0.5737, "step": 3002 }, { "epoch": 0.52, "learning_rate": 9.834483387461871e-06, "loss": 0.5732, "step": 3003 }, { "epoch": 0.52, "learning_rate": 9.828873193606363e-06, "loss": 0.5679, "step": 3004 }, { "epoch": 0.52, "learning_rate": 9.823263053627042e-06, "loss": 0.5775, "step": 3005 }, { "epoch": 0.52, "learning_rate": 9.817652969290157e-06, "loss": 0.5626, "step": 3006 }, { "epoch": 0.52, "learning_rate": 9.812042942361943e-06, "loss": 0.5666, "step": 3007 }, { "epoch": 0.52, "learning_rate": 9.806432974608619e-06, "loss": 0.5683, "step": 3008 }, { "epoch": 0.52, "learning_rate": 9.800823067796375e-06, "loss": 0.5652, "step": 3009 }, { "epoch": 0.52, "learning_rate": 9.795213223691396e-06, "loss": 0.5586, "step": 3010 }, { "epoch": 0.52, "learning_rate": 9.789603444059835e-06, "loss": 0.5616, "step": 3011 }, { "epoch": 0.52, "learning_rate": 9.783993730667833e-06, "loss": 0.5713, "step": 3012 }, { "epoch": 0.52, "learning_rate": 9.778384085281505e-06, "loss": 0.5704, "step": 3013 }, { "epoch": 0.52, "learning_rate": 9.772774509666949e-06, "loss": 0.5698, "step": 3014 }, { "epoch": 0.52, "learning_rate": 9.767165005590237e-06, "loss": 0.5712, "step": 3015 }, { "epoch": 0.52, "learning_rate": 9.761555574817424e-06, "loss": 0.589, "step": 3016 }, { "epoch": 0.52, "learning_rate": 9.755946219114528e-06, "loss": 0.5824, "step": 3017 }, { "epoch": 0.52, "learning_rate": 9.750336940247562e-06, "loss": 0.5667, "step": 3018 }, { "epoch": 0.52, "learning_rate": 9.744727739982505e-06, "loss": 0.5657, "step": 3019 }, { "epoch": 0.52, "learning_rate": 9.739118620085308e-06, "loss": 0.5715, "step": 3020 }, { "epoch": 0.52, "learning_rate": 9.733509582321908e-06, "loss": 0.5703, "step": 3021 }, { "epoch": 0.52, "learning_rate": 9.727900628458203e-06, "loss": 0.5586, "step": 3022 }, { "epoch": 0.52, "learning_rate": 9.722291760260077e-06, "loss": 0.5637, "step": 3023 }, { "epoch": 0.52, "learning_rate": 9.716682979493375e-06, "loss": 0.5574, "step": 3024 }, { "epoch": 0.52, "learning_rate": 9.711074287923922e-06, "loss": 0.5792, "step": 3025 }, { "epoch": 0.52, "learning_rate": 9.705465687317517e-06, "loss": 0.5647, "step": 3026 }, { "epoch": 0.52, "learning_rate": 9.699857179439923e-06, "loss": 0.5693, "step": 3027 }, { "epoch": 0.52, "learning_rate": 9.694248766056878e-06, "loss": 0.5582, "step": 3028 }, { "epoch": 0.52, "learning_rate": 9.68864044893409e-06, "loss": 0.5792, "step": 3029 }, { "epoch": 0.52, "learning_rate": 9.68303222983724e-06, "loss": 0.5721, "step": 3030 }, { "epoch": 0.53, "learning_rate": 9.677424110531966e-06, "loss": 0.5854, "step": 3031 }, { "epoch": 0.53, "learning_rate": 9.671816092783892e-06, "loss": 0.5488, "step": 3032 }, { "epoch": 0.53, "learning_rate": 9.666208178358595e-06, "loss": 0.5807, "step": 3033 }, { "epoch": 0.53, "learning_rate": 9.66060036902163e-06, "loss": 0.5721, "step": 3034 }, { "epoch": 0.53, "learning_rate": 9.65499266653851e-06, "loss": 0.5749, "step": 3035 }, { "epoch": 0.53, "learning_rate": 9.649385072674724e-06, "loss": 0.5665, "step": 3036 }, { "epoch": 0.53, "learning_rate": 9.643777589195716e-06, "loss": 0.5748, "step": 3037 }, { "epoch": 0.53, "learning_rate": 9.638170217866906e-06, "loss": 0.5626, "step": 3038 }, { "epoch": 0.53, "learning_rate": 9.632562960453674e-06, "loss": 0.5789, "step": 3039 }, { "epoch": 0.53, "learning_rate": 9.626955818721357e-06, "loss": 0.5683, "step": 3040 }, { "epoch": 0.53, "learning_rate": 9.62134879443527e-06, "loss": 0.5722, "step": 3041 }, { "epoch": 0.53, "learning_rate": 9.615741889360678e-06, "loss": 0.5695, "step": 3042 }, { "epoch": 0.53, "learning_rate": 9.610135105262818e-06, "loss": 0.5658, "step": 3043 }, { "epoch": 0.53, "learning_rate": 9.604528443906887e-06, "loss": 0.5658, "step": 3044 }, { "epoch": 0.53, "learning_rate": 9.598921907058033e-06, "loss": 0.5915, "step": 3045 }, { "epoch": 0.53, "learning_rate": 9.593315496481377e-06, "loss": 0.5705, "step": 3046 }, { "epoch": 0.53, "learning_rate": 9.587709213941997e-06, "loss": 0.5708, "step": 3047 }, { "epoch": 0.53, "learning_rate": 9.582103061204928e-06, "loss": 0.5664, "step": 3048 }, { "epoch": 0.53, "learning_rate": 9.576497040035165e-06, "loss": 0.575, "step": 3049 }, { "epoch": 0.53, "learning_rate": 9.570891152197667e-06, "loss": 0.5559, "step": 3050 }, { "epoch": 0.53, "learning_rate": 9.565285399457345e-06, "loss": 0.5789, "step": 3051 }, { "epoch": 0.53, "learning_rate": 9.559679783579068e-06, "loss": 0.5833, "step": 3052 }, { "epoch": 0.53, "learning_rate": 9.55407430632766e-06, "loss": 0.5603, "step": 3053 }, { "epoch": 0.53, "learning_rate": 9.548468969467912e-06, "loss": 0.5558, "step": 3054 }, { "epoch": 0.53, "learning_rate": 9.542863774764557e-06, "loss": 0.5655, "step": 3055 }, { "epoch": 0.53, "learning_rate": 9.53725872398229e-06, "loss": 0.5792, "step": 3056 }, { "epoch": 0.53, "learning_rate": 9.531653818885763e-06, "loss": 0.5622, "step": 3057 }, { "epoch": 0.53, "learning_rate": 9.52604906123958e-06, "loss": 0.5566, "step": 3058 }, { "epoch": 0.53, "learning_rate": 9.52044445280829e-06, "loss": 0.58, "step": 3059 }, { "epoch": 0.53, "learning_rate": 9.514839995356411e-06, "loss": 0.5587, "step": 3060 }, { "epoch": 0.53, "learning_rate": 9.509235690648401e-06, "loss": 0.5814, "step": 3061 }, { "epoch": 0.53, "learning_rate": 9.503631540448674e-06, "loss": 0.5679, "step": 3062 }, { "epoch": 0.53, "learning_rate": 9.4980275465216e-06, "loss": 0.5612, "step": 3063 }, { "epoch": 0.53, "learning_rate": 9.492423710631488e-06, "loss": 0.59, "step": 3064 }, { "epoch": 0.53, "learning_rate": 9.486820034542614e-06, "loss": 0.5664, "step": 3065 }, { "epoch": 0.53, "learning_rate": 9.481216520019185e-06, "loss": 0.5803, "step": 3066 }, { "epoch": 0.53, "learning_rate": 9.475613168825374e-06, "loss": 0.5636, "step": 3067 }, { "epoch": 0.53, "learning_rate": 9.470009982725288e-06, "loss": 0.5574, "step": 3068 }, { "epoch": 0.53, "learning_rate": 9.464406963482993e-06, "loss": 0.5664, "step": 3069 }, { "epoch": 0.53, "learning_rate": 9.4588041128625e-06, "loss": 0.587, "step": 3070 }, { "epoch": 0.53, "learning_rate": 9.45320143262776e-06, "loss": 0.5524, "step": 3071 }, { "epoch": 0.53, "learning_rate": 9.447598924542686e-06, "loss": 0.5835, "step": 3072 }, { "epoch": 0.53, "learning_rate": 9.441996590371117e-06, "loss": 0.5729, "step": 3073 }, { "epoch": 0.53, "learning_rate": 9.436394431876847e-06, "loss": 0.5601, "step": 3074 }, { "epoch": 0.53, "learning_rate": 9.430792450823616e-06, "loss": 0.5673, "step": 3075 }, { "epoch": 0.53, "learning_rate": 9.42519064897511e-06, "loss": 0.5559, "step": 3076 }, { "epoch": 0.53, "learning_rate": 9.419589028094952e-06, "loss": 0.5622, "step": 3077 }, { "epoch": 0.53, "learning_rate": 9.41398758994671e-06, "loss": 0.5691, "step": 3078 }, { "epoch": 0.53, "learning_rate": 9.4083863362939e-06, "loss": 0.5869, "step": 3079 }, { "epoch": 0.53, "learning_rate": 9.40278526889997e-06, "loss": 0.5896, "step": 3080 }, { "epoch": 0.53, "learning_rate": 9.397184389528323e-06, "loss": 0.5735, "step": 3081 }, { "epoch": 0.53, "learning_rate": 9.391583699942286e-06, "loss": 0.5685, "step": 3082 }, { "epoch": 0.53, "learning_rate": 9.385983201905143e-06, "loss": 0.5614, "step": 3083 }, { "epoch": 0.53, "learning_rate": 9.380382897180103e-06, "loss": 0.5792, "step": 3084 }, { "epoch": 0.53, "learning_rate": 9.374782787530326e-06, "loss": 0.5615, "step": 3085 }, { "epoch": 0.53, "learning_rate": 9.369182874718904e-06, "loss": 0.5732, "step": 3086 }, { "epoch": 0.53, "learning_rate": 9.363583160508864e-06, "loss": 0.5858, "step": 3087 }, { "epoch": 0.53, "learning_rate": 9.357983646663178e-06, "loss": 0.5649, "step": 3088 }, { "epoch": 0.54, "learning_rate": 9.352384334944754e-06, "loss": 0.5673, "step": 3089 }, { "epoch": 0.54, "learning_rate": 9.346785227116432e-06, "loss": 0.5692, "step": 3090 }, { "epoch": 0.54, "learning_rate": 9.341186324940991e-06, "loss": 0.5663, "step": 3091 }, { "epoch": 0.54, "learning_rate": 9.335587630181142e-06, "loss": 0.5716, "step": 3092 }, { "epoch": 0.54, "learning_rate": 9.329989144599536e-06, "loss": 0.5606, "step": 3093 }, { "epoch": 0.54, "learning_rate": 9.324390869958756e-06, "loss": 0.5656, "step": 3094 }, { "epoch": 0.54, "learning_rate": 9.318792808021313e-06, "loss": 0.5768, "step": 3095 }, { "epoch": 0.54, "learning_rate": 9.313194960549661e-06, "loss": 0.5667, "step": 3096 }, { "epoch": 0.54, "learning_rate": 9.307597329306175e-06, "loss": 0.5661, "step": 3097 }, { "epoch": 0.54, "learning_rate": 9.301999916053175e-06, "loss": 0.5684, "step": 3098 }, { "epoch": 0.54, "learning_rate": 9.2964027225529e-06, "loss": 0.5601, "step": 3099 }, { "epoch": 0.54, "learning_rate": 9.290805750567532e-06, "loss": 0.5818, "step": 3100 }, { "epoch": 0.54, "learning_rate": 9.28520900185917e-06, "loss": 0.5749, "step": 3101 }, { "epoch": 0.54, "learning_rate": 9.279612478189853e-06, "loss": 0.5721, "step": 3102 }, { "epoch": 0.54, "learning_rate": 9.27401618132154e-06, "loss": 0.5575, "step": 3103 }, { "epoch": 0.54, "learning_rate": 9.268420113016132e-06, "loss": 0.5682, "step": 3104 }, { "epoch": 0.54, "learning_rate": 9.262824275035443e-06, "loss": 0.5738, "step": 3105 }, { "epoch": 0.54, "learning_rate": 9.257228669141226e-06, "loss": 0.5848, "step": 3106 }, { "epoch": 0.54, "learning_rate": 9.251633297095158e-06, "loss": 0.5731, "step": 3107 }, { "epoch": 0.54, "learning_rate": 9.246038160658833e-06, "loss": 0.5778, "step": 3108 }, { "epoch": 0.54, "learning_rate": 9.24044326159379e-06, "loss": 0.575, "step": 3109 }, { "epoch": 0.54, "learning_rate": 9.23484860166147e-06, "loss": 0.5572, "step": 3110 }, { "epoch": 0.54, "learning_rate": 9.22925418262326e-06, "loss": 0.5769, "step": 3111 }, { "epoch": 0.54, "learning_rate": 9.223660006240458e-06, "loss": 0.575, "step": 3112 }, { "epoch": 0.54, "learning_rate": 9.218066074274289e-06, "loss": 0.5551, "step": 3113 }, { "epoch": 0.54, "learning_rate": 9.212472388485907e-06, "loss": 0.5839, "step": 3114 }, { "epoch": 0.54, "learning_rate": 9.206878950636376e-06, "loss": 0.5803, "step": 3115 }, { "epoch": 0.54, "learning_rate": 9.201285762486687e-06, "loss": 0.5699, "step": 3116 }, { "epoch": 0.54, "learning_rate": 9.195692825797764e-06, "loss": 0.5737, "step": 3117 }, { "epoch": 0.54, "learning_rate": 9.190100142330433e-06, "loss": 0.5628, "step": 3118 }, { "epoch": 0.54, "learning_rate": 9.184507713845455e-06, "loss": 0.5599, "step": 3119 }, { "epoch": 0.54, "learning_rate": 9.178915542103504e-06, "loss": 0.5758, "step": 3120 }, { "epoch": 0.54, "learning_rate": 9.173323628865173e-06, "loss": 0.5592, "step": 3121 }, { "epoch": 0.54, "learning_rate": 9.167731975890977e-06, "loss": 0.5657, "step": 3122 }, { "epoch": 0.54, "learning_rate": 9.162140584941344e-06, "loss": 0.5641, "step": 3123 }, { "epoch": 0.54, "learning_rate": 9.156549457776624e-06, "loss": 0.579, "step": 3124 }, { "epoch": 0.54, "learning_rate": 9.150958596157085e-06, "loss": 0.5738, "step": 3125 }, { "epoch": 0.54, "learning_rate": 9.145368001842905e-06, "loss": 0.5553, "step": 3126 }, { "epoch": 0.54, "learning_rate": 9.139777676594184e-06, "loss": 0.5838, "step": 3127 }, { "epoch": 0.54, "learning_rate": 9.134187622170939e-06, "loss": 0.5708, "step": 3128 }, { "epoch": 0.54, "learning_rate": 9.128597840333087e-06, "loss": 0.5634, "step": 3129 }, { "epoch": 0.54, "learning_rate": 9.123008332840478e-06, "loss": 0.5611, "step": 3130 }, { "epoch": 0.54, "learning_rate": 9.117419101452864e-06, "loss": 0.5724, "step": 3131 }, { "epoch": 0.54, "learning_rate": 9.111830147929915e-06, "loss": 0.5797, "step": 3132 }, { "epoch": 0.54, "learning_rate": 9.106241474031213e-06, "loss": 0.5631, "step": 3133 }, { "epoch": 0.54, "learning_rate": 9.100653081516249e-06, "loss": 0.5852, "step": 3134 }, { "epoch": 0.54, "learning_rate": 9.095064972144432e-06, "loss": 0.556, "step": 3135 }, { "epoch": 0.54, "learning_rate": 9.089477147675072e-06, "loss": 0.5539, "step": 3136 }, { "epoch": 0.54, "learning_rate": 9.083889609867396e-06, "loss": 0.5585, "step": 3137 }, { "epoch": 0.54, "learning_rate": 9.078302360480544e-06, "loss": 0.5667, "step": 3138 }, { "epoch": 0.54, "learning_rate": 9.072715401273553e-06, "loss": 0.5637, "step": 3139 }, { "epoch": 0.54, "learning_rate": 9.067128734005382e-06, "loss": 0.5865, "step": 3140 }, { "epoch": 0.54, "learning_rate": 9.06154236043489e-06, "loss": 0.5562, "step": 3141 }, { "epoch": 0.54, "learning_rate": 9.05595628232085e-06, "loss": 0.5597, "step": 3142 }, { "epoch": 0.54, "learning_rate": 9.050370501421931e-06, "loss": 0.5715, "step": 3143 }, { "epoch": 0.54, "learning_rate": 9.04478501949672e-06, "loss": 0.5763, "step": 3144 }, { "epoch": 0.54, "learning_rate": 9.039199838303702e-06, "loss": 0.5732, "step": 3145 }, { "epoch": 0.54, "learning_rate": 9.033614959601274e-06, "loss": 0.5756, "step": 3146 }, { "epoch": 0.55, "learning_rate": 9.02803038514773e-06, "loss": 0.5491, "step": 3147 }, { "epoch": 0.55, "learning_rate": 9.022446116701278e-06, "loss": 0.5771, "step": 3148 }, { "epoch": 0.55, "learning_rate": 9.01686215602002e-06, "loss": 0.5561, "step": 3149 }, { "epoch": 0.55, "learning_rate": 9.011278504861963e-06, "loss": 0.5583, "step": 3150 }, { "epoch": 0.55, "learning_rate": 9.005695164985027e-06, "loss": 0.5675, "step": 3151 }, { "epoch": 0.55, "learning_rate": 9.000112138147018e-06, "loss": 0.567, "step": 3152 }, { "epoch": 0.55, "learning_rate": 8.994529426105656e-06, "loss": 0.5635, "step": 3153 }, { "epoch": 0.55, "learning_rate": 8.988947030618554e-06, "loss": 0.5868, "step": 3154 }, { "epoch": 0.55, "learning_rate": 8.983364953443227e-06, "loss": 0.5738, "step": 3155 }, { "epoch": 0.55, "learning_rate": 8.977783196337102e-06, "loss": 0.5634, "step": 3156 }, { "epoch": 0.55, "learning_rate": 8.972201761057482e-06, "loss": 0.5699, "step": 3157 }, { "epoch": 0.55, "learning_rate": 8.966620649361584e-06, "loss": 0.5562, "step": 3158 }, { "epoch": 0.55, "learning_rate": 8.961039863006523e-06, "loss": 0.5615, "step": 3159 }, { "epoch": 0.55, "learning_rate": 8.955459403749305e-06, "loss": 0.5798, "step": 3160 }, { "epoch": 0.55, "learning_rate": 8.949879273346844e-06, "loss": 0.5518, "step": 3161 }, { "epoch": 0.55, "learning_rate": 8.944299473555935e-06, "loss": 0.5709, "step": 3162 }, { "epoch": 0.55, "learning_rate": 8.938720006133282e-06, "loss": 0.5667, "step": 3163 }, { "epoch": 0.55, "learning_rate": 8.933140872835482e-06, "loss": 0.574, "step": 3164 }, { "epoch": 0.55, "learning_rate": 8.927562075419018e-06, "loss": 0.5732, "step": 3165 }, { "epoch": 0.55, "learning_rate": 8.921983615640277e-06, "loss": 0.5535, "step": 3166 }, { "epoch": 0.55, "learning_rate": 8.916405495255536e-06, "loss": 0.5783, "step": 3167 }, { "epoch": 0.55, "learning_rate": 8.910827716020965e-06, "loss": 0.5747, "step": 3168 }, { "epoch": 0.55, "learning_rate": 8.905250279692631e-06, "loss": 0.5495, "step": 3169 }, { "epoch": 0.55, "learning_rate": 8.899673188026488e-06, "loss": 0.5674, "step": 3170 }, { "epoch": 0.55, "learning_rate": 8.894096442778375e-06, "loss": 0.5749, "step": 3171 }, { "epoch": 0.55, "learning_rate": 8.888520045704039e-06, "loss": 0.5721, "step": 3172 }, { "epoch": 0.55, "learning_rate": 8.8829439985591e-06, "loss": 0.5643, "step": 3173 }, { "epoch": 0.55, "learning_rate": 8.877368303099083e-06, "loss": 0.5734, "step": 3174 }, { "epoch": 0.55, "learning_rate": 8.871792961079391e-06, "loss": 0.5537, "step": 3175 }, { "epoch": 0.55, "learning_rate": 8.86621797425532e-06, "loss": 0.5736, "step": 3176 }, { "epoch": 0.55, "learning_rate": 8.860643344382057e-06, "loss": 0.5716, "step": 3177 }, { "epoch": 0.55, "learning_rate": 8.855069073214668e-06, "loss": 0.5486, "step": 3178 }, { "epoch": 0.55, "learning_rate": 8.84949516250812e-06, "loss": 0.5792, "step": 3179 }, { "epoch": 0.55, "learning_rate": 8.843921614017247e-06, "loss": 0.5802, "step": 3180 }, { "epoch": 0.55, "learning_rate": 8.83834842949679e-06, "loss": 0.5818, "step": 3181 }, { "epoch": 0.55, "learning_rate": 8.832775610701363e-06, "loss": 0.5476, "step": 3182 }, { "epoch": 0.55, "learning_rate": 8.827203159385464e-06, "loss": 0.5625, "step": 3183 }, { "epoch": 0.55, "learning_rate": 8.821631077303485e-06, "loss": 0.5743, "step": 3184 }, { "epoch": 0.55, "learning_rate": 8.81605936620969e-06, "loss": 0.5813, "step": 3185 }, { "epoch": 0.55, "learning_rate": 8.810488027858231e-06, "loss": 0.5606, "step": 3186 }, { "epoch": 0.55, "learning_rate": 8.804917064003145e-06, "loss": 0.5461, "step": 3187 }, { "epoch": 0.55, "learning_rate": 8.799346476398351e-06, "loss": 0.5651, "step": 3188 }, { "epoch": 0.55, "learning_rate": 8.793776266797646e-06, "loss": 0.5812, "step": 3189 }, { "epoch": 0.55, "learning_rate": 8.788206436954712e-06, "loss": 0.5802, "step": 3190 }, { "epoch": 0.55, "learning_rate": 8.782636988623108e-06, "loss": 0.5686, "step": 3191 }, { "epoch": 0.55, "learning_rate": 8.777067923556274e-06, "loss": 0.5662, "step": 3192 }, { "epoch": 0.55, "learning_rate": 8.77149924350753e-06, "loss": 0.5573, "step": 3193 }, { "epoch": 0.55, "learning_rate": 8.765930950230075e-06, "loss": 0.563, "step": 3194 }, { "epoch": 0.55, "learning_rate": 8.760363045476986e-06, "loss": 0.5706, "step": 3195 }, { "epoch": 0.55, "learning_rate": 8.754795531001215e-06, "loss": 0.5689, "step": 3196 }, { "epoch": 0.55, "learning_rate": 8.749228408555596e-06, "loss": 0.5532, "step": 3197 }, { "epoch": 0.55, "learning_rate": 8.743661679892832e-06, "loss": 0.5731, "step": 3198 }, { "epoch": 0.55, "learning_rate": 8.738095346765519e-06, "loss": 0.5643, "step": 3199 }, { "epoch": 0.55, "learning_rate": 8.732529410926102e-06, "loss": 0.5995, "step": 3200 }, { "epoch": 0.55, "learning_rate": 8.726963874126924e-06, "loss": 0.554, "step": 3201 }, { "epoch": 0.55, "learning_rate": 8.721398738120191e-06, "loss": 0.5687, "step": 3202 }, { "epoch": 0.55, "learning_rate": 8.715834004657987e-06, "loss": 0.5721, "step": 3203 }, { "epoch": 0.55, "learning_rate": 8.710269675492264e-06, "loss": 0.5581, "step": 3204 }, { "epoch": 0.56, "learning_rate": 8.704705752374857e-06, "loss": 0.5885, "step": 3205 }, { "epoch": 0.56, "learning_rate": 8.699142237057464e-06, "loss": 0.5603, "step": 3206 }, { "epoch": 0.56, "learning_rate": 8.693579131291654e-06, "loss": 0.5826, "step": 3207 }, { "epoch": 0.56, "learning_rate": 8.688016436828876e-06, "loss": 0.5702, "step": 3208 }, { "epoch": 0.56, "learning_rate": 8.682454155420438e-06, "loss": 0.5653, "step": 3209 }, { "epoch": 0.56, "learning_rate": 8.676892288817531e-06, "loss": 0.5583, "step": 3210 }, { "epoch": 0.56, "learning_rate": 8.671330838771201e-06, "loss": 0.5767, "step": 3211 }, { "epoch": 0.56, "learning_rate": 8.665769807032375e-06, "loss": 0.5501, "step": 3212 }, { "epoch": 0.56, "learning_rate": 8.660209195351846e-06, "loss": 0.5694, "step": 3213 }, { "epoch": 0.56, "learning_rate": 8.654649005480265e-06, "loss": 0.5667, "step": 3214 }, { "epoch": 0.56, "learning_rate": 8.64908923916816e-06, "loss": 0.5618, "step": 3215 }, { "epoch": 0.56, "learning_rate": 8.643529898165926e-06, "loss": 0.5486, "step": 3216 }, { "epoch": 0.56, "learning_rate": 8.637970984223817e-06, "loss": 0.5897, "step": 3217 }, { "epoch": 0.56, "learning_rate": 8.632412499091958e-06, "loss": 0.5697, "step": 3218 }, { "epoch": 0.56, "learning_rate": 8.62685444452034e-06, "loss": 0.5536, "step": 3219 }, { "epoch": 0.56, "learning_rate": 8.621296822258813e-06, "loss": 0.5705, "step": 3220 }, { "epoch": 0.56, "learning_rate": 8.615739634057098e-06, "loss": 0.5682, "step": 3221 }, { "epoch": 0.56, "learning_rate": 8.61018288166477e-06, "loss": 0.564, "step": 3222 }, { "epoch": 0.56, "learning_rate": 8.604626566831279e-06, "loss": 0.5705, "step": 3223 }, { "epoch": 0.56, "learning_rate": 8.599070691305925e-06, "loss": 0.5694, "step": 3224 }, { "epoch": 0.56, "learning_rate": 8.593515256837875e-06, "loss": 0.57, "step": 3225 }, { "epoch": 0.56, "learning_rate": 8.58796026517616e-06, "loss": 0.5679, "step": 3226 }, { "epoch": 0.56, "learning_rate": 8.582405718069672e-06, "loss": 0.5567, "step": 3227 }, { "epoch": 0.56, "learning_rate": 8.576851617267151e-06, "loss": 0.5709, "step": 3228 }, { "epoch": 0.56, "learning_rate": 8.571297964517212e-06, "loss": 0.5763, "step": 3229 }, { "epoch": 0.56, "learning_rate": 8.56574476156832e-06, "loss": 0.566, "step": 3230 }, { "epoch": 0.56, "learning_rate": 8.560192010168798e-06, "loss": 0.5616, "step": 3231 }, { "epoch": 0.56, "learning_rate": 8.554639712066837e-06, "loss": 0.5729, "step": 3232 }, { "epoch": 0.56, "learning_rate": 8.549087869010471e-06, "loss": 0.5795, "step": 3233 }, { "epoch": 0.56, "learning_rate": 8.543536482747603e-06, "loss": 0.5666, "step": 3234 }, { "epoch": 0.56, "learning_rate": 8.537985555025982e-06, "loss": 0.5754, "step": 3235 }, { "epoch": 0.56, "learning_rate": 8.532435087593221e-06, "loss": 0.5524, "step": 3236 }, { "epoch": 0.56, "learning_rate": 8.52688508219678e-06, "loss": 0.581, "step": 3237 }, { "epoch": 0.56, "learning_rate": 8.52133554058398e-06, "loss": 0.5536, "step": 3238 }, { "epoch": 0.56, "learning_rate": 8.515786464501998e-06, "loss": 0.5643, "step": 3239 }, { "epoch": 0.56, "learning_rate": 8.510237855697855e-06, "loss": 0.5642, "step": 3240 }, { "epoch": 0.56, "learning_rate": 8.504689715918439e-06, "loss": 0.5644, "step": 3241 }, { "epoch": 0.56, "learning_rate": 8.499142046910471e-06, "loss": 0.5638, "step": 3242 }, { "epoch": 0.56, "learning_rate": 8.493594850420537e-06, "loss": 0.5803, "step": 3243 }, { "epoch": 0.56, "learning_rate": 8.488048128195073e-06, "loss": 0.5613, "step": 3244 }, { "epoch": 0.56, "learning_rate": 8.482501881980368e-06, "loss": 0.5655, "step": 3245 }, { "epoch": 0.56, "learning_rate": 8.47695611352255e-06, "loss": 0.5756, "step": 3246 }, { "epoch": 0.56, "learning_rate": 8.471410824567611e-06, "loss": 0.5646, "step": 3247 }, { "epoch": 0.56, "learning_rate": 8.465866016861383e-06, "loss": 0.5777, "step": 3248 }, { "epoch": 0.56, "learning_rate": 8.460321692149546e-06, "loss": 0.5764, "step": 3249 }, { "epoch": 0.56, "learning_rate": 8.454777852177635e-06, "loss": 0.5454, "step": 3250 }, { "epoch": 0.56, "learning_rate": 8.449234498691024e-06, "loss": 0.5714, "step": 3251 }, { "epoch": 0.56, "learning_rate": 8.443691633434942e-06, "loss": 0.5745, "step": 3252 }, { "epoch": 0.56, "learning_rate": 8.438149258154455e-06, "loss": 0.5433, "step": 3253 }, { "epoch": 0.56, "learning_rate": 8.432607374594484e-06, "loss": 0.5904, "step": 3254 }, { "epoch": 0.56, "learning_rate": 8.427065984499791e-06, "loss": 0.5666, "step": 3255 }, { "epoch": 0.56, "learning_rate": 8.421525089614977e-06, "loss": 0.5413, "step": 3256 }, { "epoch": 0.56, "learning_rate": 8.415984691684498e-06, "loss": 0.5711, "step": 3257 }, { "epoch": 0.56, "learning_rate": 8.410444792452648e-06, "loss": 0.5394, "step": 3258 }, { "epoch": 0.56, "learning_rate": 8.40490539366356e-06, "loss": 0.5746, "step": 3259 }, { "epoch": 0.56, "learning_rate": 8.39936649706122e-06, "loss": 0.5677, "step": 3260 }, { "epoch": 0.56, "learning_rate": 8.393828104389446e-06, "loss": 0.5882, "step": 3261 }, { "epoch": 0.57, "learning_rate": 8.388290217391902e-06, "loss": 0.5598, "step": 3262 }, { "epoch": 0.57, "learning_rate": 8.38275283781209e-06, "loss": 0.5639, "step": 3263 }, { "epoch": 0.57, "learning_rate": 8.377215967393355e-06, "loss": 0.5564, "step": 3264 }, { "epoch": 0.57, "learning_rate": 8.371679607878884e-06, "loss": 0.5673, "step": 3265 }, { "epoch": 0.57, "learning_rate": 8.366143761011695e-06, "loss": 0.5743, "step": 3266 }, { "epoch": 0.57, "learning_rate": 8.360608428534652e-06, "loss": 0.5591, "step": 3267 }, { "epoch": 0.57, "learning_rate": 8.355073612190452e-06, "loss": 0.558, "step": 3268 }, { "epoch": 0.57, "learning_rate": 8.349539313721639e-06, "loss": 0.5581, "step": 3269 }, { "epoch": 0.57, "learning_rate": 8.344005534870578e-06, "loss": 0.58, "step": 3270 }, { "epoch": 0.57, "learning_rate": 8.338472277379485e-06, "loss": 0.5588, "step": 3271 }, { "epoch": 0.57, "learning_rate": 8.332939542990401e-06, "loss": 0.5751, "step": 3272 }, { "epoch": 0.57, "learning_rate": 8.327407333445214e-06, "loss": 0.5902, "step": 3273 }, { "epoch": 0.57, "learning_rate": 8.321875650485636e-06, "loss": 0.5767, "step": 3274 }, { "epoch": 0.57, "learning_rate": 8.316344495853218e-06, "loss": 0.5771, "step": 3275 }, { "epoch": 0.57, "learning_rate": 8.310813871289349e-06, "loss": 0.5542, "step": 3276 }, { "epoch": 0.57, "learning_rate": 8.30528377853524e-06, "loss": 0.5659, "step": 3277 }, { "epoch": 0.57, "learning_rate": 8.299754219331944e-06, "loss": 0.5609, "step": 3278 }, { "epoch": 0.57, "learning_rate": 8.29422519542034e-06, "loss": 0.5485, "step": 3279 }, { "epoch": 0.57, "learning_rate": 8.288696708541146e-06, "loss": 0.5508, "step": 3280 }, { "epoch": 0.57, "learning_rate": 8.283168760434904e-06, "loss": 0.5799, "step": 3281 }, { "epoch": 0.57, "learning_rate": 8.277641352841985e-06, "loss": 0.551, "step": 3282 }, { "epoch": 0.57, "learning_rate": 8.272114487502604e-06, "loss": 0.5771, "step": 3283 }, { "epoch": 0.57, "learning_rate": 8.266588166156785e-06, "loss": 0.5679, "step": 3284 }, { "epoch": 0.57, "learning_rate": 8.26106239054439e-06, "loss": 0.5455, "step": 3285 }, { "epoch": 0.57, "learning_rate": 8.255537162405117e-06, "loss": 0.5828, "step": 3286 }, { "epoch": 0.57, "learning_rate": 8.250012483478478e-06, "loss": 0.5592, "step": 3287 }, { "epoch": 0.57, "learning_rate": 8.244488355503822e-06, "loss": 0.5678, "step": 3288 }, { "epoch": 0.57, "learning_rate": 8.23896478022032e-06, "loss": 0.576, "step": 3289 }, { "epoch": 0.57, "learning_rate": 8.233441759366969e-06, "loss": 0.5505, "step": 3290 }, { "epoch": 0.57, "learning_rate": 8.227919294682595e-06, "loss": 0.5691, "step": 3291 }, { "epoch": 0.57, "learning_rate": 8.222397387905841e-06, "loss": 0.5694, "step": 3292 }, { "epoch": 0.57, "learning_rate": 8.216876040775185e-06, "loss": 0.5683, "step": 3293 }, { "epoch": 0.57, "learning_rate": 8.211355255028924e-06, "loss": 0.5826, "step": 3294 }, { "epoch": 0.57, "learning_rate": 8.205835032405174e-06, "loss": 0.5617, "step": 3295 }, { "epoch": 0.57, "learning_rate": 8.20031537464188e-06, "loss": 0.5622, "step": 3296 }, { "epoch": 0.57, "learning_rate": 8.194796283476808e-06, "loss": 0.5649, "step": 3297 }, { "epoch": 0.57, "learning_rate": 8.189277760647537e-06, "loss": 0.569, "step": 3298 }, { "epoch": 0.57, "learning_rate": 8.183759807891481e-06, "loss": 0.556, "step": 3299 }, { "epoch": 0.57, "learning_rate": 8.178242426945867e-06, "loss": 0.5704, "step": 3300 }, { "epoch": 0.57, "learning_rate": 8.172725619547738e-06, "loss": 0.5654, "step": 3301 }, { "epoch": 0.57, "learning_rate": 8.167209387433968e-06, "loss": 0.5559, "step": 3302 }, { "epoch": 0.57, "learning_rate": 8.161693732341238e-06, "loss": 0.5683, "step": 3303 }, { "epoch": 0.57, "learning_rate": 8.156178656006057e-06, "loss": 0.5545, "step": 3304 }, { "epoch": 0.57, "learning_rate": 8.150664160164743e-06, "loss": 0.5623, "step": 3305 }, { "epoch": 0.57, "learning_rate": 8.145150246553437e-06, "loss": 0.5695, "step": 3306 }, { "epoch": 0.57, "learning_rate": 8.139636916908098e-06, "loss": 0.5724, "step": 3307 }, { "epoch": 0.57, "learning_rate": 8.134124172964493e-06, "loss": 0.5706, "step": 3308 }, { "epoch": 0.57, "learning_rate": 8.128612016458214e-06, "loss": 0.5785, "step": 3309 }, { "epoch": 0.57, "learning_rate": 8.123100449124662e-06, "loss": 0.5663, "step": 3310 }, { "epoch": 0.57, "learning_rate": 8.117589472699059e-06, "loss": 0.5633, "step": 3311 }, { "epoch": 0.57, "learning_rate": 8.11207908891643e-06, "loss": 0.5568, "step": 3312 }, { "epoch": 0.57, "learning_rate": 8.106569299511622e-06, "loss": 0.5712, "step": 3313 }, { "epoch": 0.57, "learning_rate": 8.101060106219292e-06, "loss": 0.5726, "step": 3314 }, { "epoch": 0.57, "learning_rate": 8.095551510773913e-06, "loss": 0.5774, "step": 3315 }, { "epoch": 0.57, "learning_rate": 8.090043514909763e-06, "loss": 0.5815, "step": 3316 }, { "epoch": 0.57, "learning_rate": 8.08453612036094e-06, "loss": 0.5628, "step": 3317 }, { "epoch": 0.57, "learning_rate": 8.079029328861342e-06, "loss": 0.5865, "step": 3318 }, { "epoch": 0.57, "learning_rate": 8.073523142144686e-06, "loss": 0.5689, "step": 3319 }, { "epoch": 0.58, "learning_rate": 8.0680175619445e-06, "loss": 0.5724, "step": 3320 }, { "epoch": 0.58, "learning_rate": 8.062512589994105e-06, "loss": 0.5683, "step": 3321 }, { "epoch": 0.58, "learning_rate": 8.057008228026654e-06, "loss": 0.5561, "step": 3322 }, { "epoch": 0.58, "learning_rate": 8.05150447777509e-06, "loss": 0.5804, "step": 3323 }, { "epoch": 0.58, "learning_rate": 8.046001340972168e-06, "loss": 0.5769, "step": 3324 }, { "epoch": 0.58, "learning_rate": 8.04049881935046e-06, "loss": 0.5566, "step": 3325 }, { "epoch": 0.58, "learning_rate": 8.034996914642325e-06, "loss": 0.5656, "step": 3326 }, { "epoch": 0.58, "learning_rate": 8.029495628579941e-06, "loss": 0.5594, "step": 3327 }, { "epoch": 0.58, "learning_rate": 8.023994962895294e-06, "loss": 0.5634, "step": 3328 }, { "epoch": 0.58, "learning_rate": 8.018494919320163e-06, "loss": 0.5593, "step": 3329 }, { "epoch": 0.58, "learning_rate": 8.012995499586142e-06, "loss": 0.5762, "step": 3330 }, { "epoch": 0.58, "learning_rate": 8.00749670542462e-06, "loss": 0.571, "step": 3331 }, { "epoch": 0.58, "learning_rate": 8.001998538566794e-06, "loss": 0.5714, "step": 3332 }, { "epoch": 0.58, "learning_rate": 7.996501000743668e-06, "loss": 0.5523, "step": 3333 }, { "epoch": 0.58, "learning_rate": 7.991004093686035e-06, "loss": 0.5757, "step": 3334 }, { "epoch": 0.58, "learning_rate": 7.985507819124502e-06, "loss": 0.5714, "step": 3335 }, { "epoch": 0.58, "learning_rate": 7.980012178789467e-06, "loss": 0.5637, "step": 3336 }, { "epoch": 0.58, "learning_rate": 7.974517174411139e-06, "loss": 0.5597, "step": 3337 }, { "epoch": 0.58, "learning_rate": 7.969022807719519e-06, "loss": 0.5648, "step": 3338 }, { "epoch": 0.58, "learning_rate": 7.963529080444412e-06, "loss": 0.5583, "step": 3339 }, { "epoch": 0.58, "learning_rate": 7.958035994315409e-06, "loss": 0.5706, "step": 3340 }, { "epoch": 0.58, "learning_rate": 7.952543551061917e-06, "loss": 0.5548, "step": 3341 }, { "epoch": 0.58, "learning_rate": 7.947051752413131e-06, "loss": 0.5831, "step": 3342 }, { "epoch": 0.58, "learning_rate": 7.941560600098045e-06, "loss": 0.5817, "step": 3343 }, { "epoch": 0.58, "learning_rate": 7.936070095845447e-06, "loss": 0.554, "step": 3344 }, { "epoch": 0.58, "learning_rate": 7.930580241383924e-06, "loss": 0.5876, "step": 3345 }, { "epoch": 0.58, "learning_rate": 7.925091038441864e-06, "loss": 0.5694, "step": 3346 }, { "epoch": 0.58, "learning_rate": 7.919602488747433e-06, "loss": 0.5586, "step": 3347 }, { "epoch": 0.58, "learning_rate": 7.91411459402861e-06, "loss": 0.5739, "step": 3348 }, { "epoch": 0.58, "learning_rate": 7.908627356013153e-06, "loss": 0.563, "step": 3349 }, { "epoch": 0.58, "learning_rate": 7.903140776428624e-06, "loss": 0.581, "step": 3350 }, { "epoch": 0.58, "learning_rate": 7.897654857002374e-06, "loss": 0.5671, "step": 3351 }, { "epoch": 0.58, "learning_rate": 7.892169599461545e-06, "loss": 0.5526, "step": 3352 }, { "epoch": 0.58, "learning_rate": 7.886685005533073e-06, "loss": 0.5492, "step": 3353 }, { "epoch": 0.58, "learning_rate": 7.881201076943678e-06, "loss": 0.5714, "step": 3354 }, { "epoch": 0.58, "learning_rate": 7.87571781541988e-06, "loss": 0.5671, "step": 3355 }, { "epoch": 0.58, "learning_rate": 7.870235222687983e-06, "loss": 0.5877, "step": 3356 }, { "epoch": 0.58, "learning_rate": 7.864753300474084e-06, "loss": 0.5654, "step": 3357 }, { "epoch": 0.58, "learning_rate": 7.859272050504065e-06, "loss": 0.562, "step": 3358 }, { "epoch": 0.58, "learning_rate": 7.853791474503603e-06, "loss": 0.5746, "step": 3359 }, { "epoch": 0.58, "learning_rate": 7.848311574198154e-06, "loss": 0.595, "step": 3360 }, { "epoch": 0.58, "learning_rate": 7.842832351312971e-06, "loss": 0.5517, "step": 3361 }, { "epoch": 0.58, "learning_rate": 7.83735380757308e-06, "loss": 0.5751, "step": 3362 }, { "epoch": 0.58, "learning_rate": 7.831875944703307e-06, "loss": 0.5769, "step": 3363 }, { "epoch": 0.58, "learning_rate": 7.826398764428261e-06, "loss": 0.5709, "step": 3364 }, { "epoch": 0.58, "learning_rate": 7.820922268472326e-06, "loss": 0.5747, "step": 3365 }, { "epoch": 0.58, "learning_rate": 7.815446458559686e-06, "loss": 0.5763, "step": 3366 }, { "epoch": 0.58, "learning_rate": 7.809971336414297e-06, "loss": 0.5822, "step": 3367 }, { "epoch": 0.58, "learning_rate": 7.8044969037599e-06, "loss": 0.5815, "step": 3368 }, { "epoch": 0.58, "learning_rate": 7.799023162320023e-06, "loss": 0.553, "step": 3369 }, { "epoch": 0.58, "learning_rate": 7.793550113817976e-06, "loss": 0.5762, "step": 3370 }, { "epoch": 0.58, "learning_rate": 7.788077759976849e-06, "loss": 0.5638, "step": 3371 }, { "epoch": 0.58, "learning_rate": 7.782606102519513e-06, "loss": 0.5797, "step": 3372 }, { "epoch": 0.58, "learning_rate": 7.777135143168622e-06, "loss": 0.5522, "step": 3373 }, { "epoch": 0.58, "learning_rate": 7.771664883646607e-06, "loss": 0.5759, "step": 3374 }, { "epoch": 0.58, "learning_rate": 7.766195325675681e-06, "loss": 0.574, "step": 3375 }, { "epoch": 0.58, "learning_rate": 7.760726470977837e-06, "loss": 0.5699, "step": 3376 }, { "epoch": 0.58, "learning_rate": 7.755258321274845e-06, "loss": 0.5658, "step": 3377 }, { "epoch": 0.59, "learning_rate": 7.74979087828825e-06, "loss": 0.5786, "step": 3378 }, { "epoch": 0.59, "learning_rate": 7.744324143739384e-06, "loss": 0.556, "step": 3379 }, { "epoch": 0.59, "learning_rate": 7.738858119349341e-06, "loss": 0.5755, "step": 3380 }, { "epoch": 0.59, "learning_rate": 7.733392806839012e-06, "loss": 0.5653, "step": 3381 }, { "epoch": 0.59, "learning_rate": 7.727928207929039e-06, "loss": 0.5636, "step": 3382 }, { "epoch": 0.59, "learning_rate": 7.722464324339862e-06, "loss": 0.5575, "step": 3383 }, { "epoch": 0.59, "learning_rate": 7.71700115779168e-06, "loss": 0.5741, "step": 3384 }, { "epoch": 0.59, "learning_rate": 7.711538710004476e-06, "loss": 0.5674, "step": 3385 }, { "epoch": 0.59, "learning_rate": 7.706076982698e-06, "loss": 0.5906, "step": 3386 }, { "epoch": 0.59, "learning_rate": 7.700615977591782e-06, "loss": 0.5541, "step": 3387 }, { "epoch": 0.59, "learning_rate": 7.69515569640512e-06, "loss": 0.5689, "step": 3388 }, { "epoch": 0.59, "learning_rate": 7.68969614085708e-06, "loss": 0.5593, "step": 3389 }, { "epoch": 0.59, "learning_rate": 7.684237312666513e-06, "loss": 0.567, "step": 3390 }, { "epoch": 0.59, "learning_rate": 7.678779213552025e-06, "loss": 0.556, "step": 3391 }, { "epoch": 0.59, "learning_rate": 7.673321845232005e-06, "loss": 0.58, "step": 3392 }, { "epoch": 0.59, "learning_rate": 7.667865209424603e-06, "loss": 0.5649, "step": 3393 }, { "epoch": 0.59, "learning_rate": 7.662409307847744e-06, "loss": 0.5899, "step": 3394 }, { "epoch": 0.59, "learning_rate": 7.656954142219126e-06, "loss": 0.5613, "step": 3395 }, { "epoch": 0.59, "learning_rate": 7.651499714256201e-06, "loss": 0.5643, "step": 3396 }, { "epoch": 0.59, "learning_rate": 7.646046025676198e-06, "loss": 0.5498, "step": 3397 }, { "epoch": 0.59, "learning_rate": 7.640593078196117e-06, "loss": 0.5699, "step": 3398 }, { "epoch": 0.59, "learning_rate": 7.635140873532714e-06, "loss": 0.58, "step": 3399 }, { "epoch": 0.59, "learning_rate": 7.629689413402522e-06, "loss": 0.566, "step": 3400 }, { "epoch": 0.59, "learning_rate": 7.6242386995218355e-06, "loss": 0.565, "step": 3401 }, { "epoch": 0.59, "learning_rate": 7.618788733606708e-06, "loss": 0.5809, "step": 3402 }, { "epoch": 0.59, "learning_rate": 7.61333951737297e-06, "loss": 0.5635, "step": 3403 }, { "epoch": 0.59, "learning_rate": 7.607891052536203e-06, "loss": 0.5625, "step": 3404 }, { "epoch": 0.59, "learning_rate": 7.602443340811761e-06, "loss": 0.569, "step": 3405 }, { "epoch": 0.59, "learning_rate": 7.596996383914756e-06, "loss": 0.5479, "step": 3406 }, { "epoch": 0.59, "learning_rate": 7.591550183560064e-06, "loss": 0.5683, "step": 3407 }, { "epoch": 0.59, "learning_rate": 7.586104741462326e-06, "loss": 0.559, "step": 3408 }, { "epoch": 0.59, "learning_rate": 7.58066005933594e-06, "loss": 0.5646, "step": 3409 }, { "epoch": 0.59, "learning_rate": 7.575216138895062e-06, "loss": 0.5594, "step": 3410 }, { "epoch": 0.59, "learning_rate": 7.569772981853616e-06, "loss": 0.5737, "step": 3411 }, { "epoch": 0.59, "learning_rate": 7.56433058992528e-06, "loss": 0.5744, "step": 3412 }, { "epoch": 0.59, "learning_rate": 7.5588889648234905e-06, "loss": 0.572, "step": 3413 }, { "epoch": 0.59, "learning_rate": 7.553448108261452e-06, "loss": 0.5733, "step": 3414 }, { "epoch": 0.59, "learning_rate": 7.548008021952113e-06, "loss": 0.5712, "step": 3415 }, { "epoch": 0.59, "learning_rate": 7.542568707608192e-06, "loss": 0.5627, "step": 3416 }, { "epoch": 0.59, "learning_rate": 7.537130166942154e-06, "loss": 0.5519, "step": 3417 }, { "epoch": 0.59, "learning_rate": 7.53169240166623e-06, "loss": 0.5667, "step": 3418 }, { "epoch": 0.59, "learning_rate": 7.526255413492396e-06, "loss": 0.5759, "step": 3419 }, { "epoch": 0.59, "learning_rate": 7.5208192041323945e-06, "loss": 0.5639, "step": 3420 }, { "epoch": 0.59, "learning_rate": 7.5153837752977196e-06, "loss": 0.5809, "step": 3421 }, { "epoch": 0.59, "learning_rate": 7.509949128699613e-06, "loss": 0.5853, "step": 3422 }, { "epoch": 0.59, "learning_rate": 7.5045152660490805e-06, "loss": 0.566, "step": 3423 }, { "epoch": 0.59, "learning_rate": 7.499082189056871e-06, "loss": 0.5725, "step": 3424 }, { "epoch": 0.59, "learning_rate": 7.493649899433491e-06, "loss": 0.5635, "step": 3425 }, { "epoch": 0.59, "learning_rate": 7.488218398889198e-06, "loss": 0.5801, "step": 3426 }, { "epoch": 0.59, "learning_rate": 7.482787689134007e-06, "loss": 0.5541, "step": 3427 }, { "epoch": 0.59, "learning_rate": 7.4773577718776735e-06, "loss": 0.5784, "step": 3428 }, { "epoch": 0.59, "learning_rate": 7.471928648829714e-06, "loss": 0.5618, "step": 3429 }, { "epoch": 0.59, "learning_rate": 7.4665003216993835e-06, "loss": 0.5672, "step": 3430 }, { "epoch": 0.59, "learning_rate": 7.4610727921956985e-06, "loss": 0.5508, "step": 3431 }, { "epoch": 0.59, "learning_rate": 7.4556460620274174e-06, "loss": 0.5672, "step": 3432 }, { "epoch": 0.59, "learning_rate": 7.4502201329030456e-06, "loss": 0.5714, "step": 3433 }, { "epoch": 0.59, "learning_rate": 7.444795006530844e-06, "loss": 0.5746, "step": 3434 }, { "epoch": 0.6, "learning_rate": 7.4393706846188095e-06, "loss": 0.5694, "step": 3435 }, { "epoch": 0.6, "learning_rate": 7.433947168874698e-06, "loss": 0.5566, "step": 3436 }, { "epoch": 0.6, "learning_rate": 7.4285244610060036e-06, "loss": 0.5817, "step": 3437 }, { "epoch": 0.6, "learning_rate": 7.4231025627199635e-06, "loss": 0.5719, "step": 3438 }, { "epoch": 0.6, "learning_rate": 7.4176814757235685e-06, "loss": 0.565, "step": 3439 }, { "epoch": 0.6, "learning_rate": 7.4122612017235494e-06, "loss": 0.5598, "step": 3440 }, { "epoch": 0.6, "learning_rate": 7.40684174242638e-06, "loss": 0.5642, "step": 3441 }, { "epoch": 0.6, "learning_rate": 7.401423099538285e-06, "loss": 0.5786, "step": 3442 }, { "epoch": 0.6, "learning_rate": 7.396005274765219e-06, "loss": 0.5736, "step": 3443 }, { "epoch": 0.6, "learning_rate": 7.3905882698128905e-06, "loss": 0.5819, "step": 3444 }, { "epoch": 0.6, "learning_rate": 7.385172086386745e-06, "loss": 0.5763, "step": 3445 }, { "epoch": 0.6, "learning_rate": 7.379756726191969e-06, "loss": 0.5682, "step": 3446 }, { "epoch": 0.6, "learning_rate": 7.3743421909334945e-06, "loss": 0.5605, "step": 3447 }, { "epoch": 0.6, "learning_rate": 7.3689284823159844e-06, "loss": 0.5781, "step": 3448 }, { "epoch": 0.6, "learning_rate": 7.363515602043852e-06, "loss": 0.5672, "step": 3449 }, { "epoch": 0.6, "learning_rate": 7.358103551821241e-06, "loss": 0.5664, "step": 3450 }, { "epoch": 0.6, "learning_rate": 7.352692333352046e-06, "loss": 0.5652, "step": 3451 }, { "epoch": 0.6, "learning_rate": 7.34728194833988e-06, "loss": 0.5605, "step": 3452 }, { "epoch": 0.6, "learning_rate": 7.341872398488112e-06, "loss": 0.5802, "step": 3453 }, { "epoch": 0.6, "learning_rate": 7.336463685499839e-06, "loss": 0.5467, "step": 3454 }, { "epoch": 0.6, "learning_rate": 7.331055811077897e-06, "loss": 0.5633, "step": 3455 }, { "epoch": 0.6, "learning_rate": 7.325648776924857e-06, "loss": 0.5649, "step": 3456 }, { "epoch": 0.6, "learning_rate": 7.320242584743027e-06, "loss": 0.5702, "step": 3457 }, { "epoch": 0.6, "learning_rate": 7.314837236234451e-06, "loss": 0.5509, "step": 3458 }, { "epoch": 0.6, "learning_rate": 7.309432733100901e-06, "loss": 0.5763, "step": 3459 }, { "epoch": 0.6, "learning_rate": 7.304029077043893e-06, "loss": 0.5634, "step": 3460 }, { "epoch": 0.6, "learning_rate": 7.2986262697646634e-06, "loss": 0.5674, "step": 3461 }, { "epoch": 0.6, "learning_rate": 7.293224312964194e-06, "loss": 0.5624, "step": 3462 }, { "epoch": 0.6, "learning_rate": 7.287823208343192e-06, "loss": 0.5691, "step": 3463 }, { "epoch": 0.6, "learning_rate": 7.2824229576020965e-06, "loss": 0.5376, "step": 3464 }, { "epoch": 0.6, "learning_rate": 7.2770235624410844e-06, "loss": 0.5816, "step": 3465 }, { "epoch": 0.6, "learning_rate": 7.2716250245600515e-06, "loss": 0.5778, "step": 3466 }, { "epoch": 0.6, "learning_rate": 7.266227345658629e-06, "loss": 0.5672, "step": 3467 }, { "epoch": 0.6, "learning_rate": 7.260830527436183e-06, "loss": 0.567, "step": 3468 }, { "epoch": 0.6, "learning_rate": 7.255434571591802e-06, "loss": 0.5701, "step": 3469 }, { "epoch": 0.6, "learning_rate": 7.250039479824306e-06, "loss": 0.5496, "step": 3470 }, { "epoch": 0.6, "learning_rate": 7.2446452538322435e-06, "loss": 0.5774, "step": 3471 }, { "epoch": 0.6, "learning_rate": 7.239251895313887e-06, "loss": 0.5741, "step": 3472 }, { "epoch": 0.6, "learning_rate": 7.233859405967241e-06, "loss": 0.5558, "step": 3473 }, { "epoch": 0.6, "learning_rate": 7.2284677874900285e-06, "loss": 0.5642, "step": 3474 }, { "epoch": 0.6, "learning_rate": 7.223077041579707e-06, "loss": 0.5777, "step": 3475 }, { "epoch": 0.6, "learning_rate": 7.217687169933458e-06, "loss": 0.5725, "step": 3476 }, { "epoch": 0.6, "learning_rate": 7.212298174248179e-06, "loss": 0.5543, "step": 3477 }, { "epoch": 0.6, "learning_rate": 7.206910056220504e-06, "loss": 0.5596, "step": 3478 }, { "epoch": 0.6, "learning_rate": 7.201522817546783e-06, "loss": 0.5709, "step": 3479 }, { "epoch": 0.6, "learning_rate": 7.196136459923086e-06, "loss": 0.5714, "step": 3480 }, { "epoch": 0.6, "learning_rate": 7.1907509850452165e-06, "loss": 0.5715, "step": 3481 }, { "epoch": 0.6, "learning_rate": 7.18536639460869e-06, "loss": 0.5695, "step": 3482 }, { "epoch": 0.6, "learning_rate": 7.179982690308749e-06, "loss": 0.5837, "step": 3483 }, { "epoch": 0.6, "learning_rate": 7.174599873840358e-06, "loss": 0.5656, "step": 3484 }, { "epoch": 0.6, "learning_rate": 7.169217946898197e-06, "loss": 0.5757, "step": 3485 }, { "epoch": 0.6, "learning_rate": 7.1638369111766696e-06, "loss": 0.5583, "step": 3486 }, { "epoch": 0.6, "learning_rate": 7.158456768369895e-06, "loss": 0.5744, "step": 3487 }, { "epoch": 0.6, "learning_rate": 7.153077520171718e-06, "loss": 0.5542, "step": 3488 }, { "epoch": 0.6, "learning_rate": 7.147699168275697e-06, "loss": 0.5526, "step": 3489 }, { "epoch": 0.6, "learning_rate": 7.142321714375107e-06, "loss": 0.5748, "step": 3490 }, { "epoch": 0.6, "learning_rate": 7.136945160162944e-06, "loss": 0.5717, "step": 3491 }, { "epoch": 0.6, "learning_rate": 7.131569507331919e-06, "loss": 0.5686, "step": 3492 }, { "epoch": 0.61, "learning_rate": 7.12619475757446e-06, "loss": 0.5803, "step": 3493 }, { "epoch": 0.61, "learning_rate": 7.120820912582706e-06, "loss": 0.5807, "step": 3494 }, { "epoch": 0.61, "learning_rate": 7.115447974048522e-06, "loss": 0.5673, "step": 3495 }, { "epoch": 0.61, "learning_rate": 7.110075943663473e-06, "loss": 0.5696, "step": 3496 }, { "epoch": 0.61, "learning_rate": 7.104704823118851e-06, "loss": 0.5573, "step": 3497 }, { "epoch": 0.61, "learning_rate": 7.099334614105652e-06, "loss": 0.5743, "step": 3498 }, { "epoch": 0.61, "learning_rate": 7.093965318314595e-06, "loss": 0.561, "step": 3499 }, { "epoch": 0.61, "learning_rate": 7.0885969374361e-06, "loss": 0.5589, "step": 3500 }, { "epoch": 0.61, "learning_rate": 7.083229473160309e-06, "loss": 0.5731, "step": 3501 }, { "epoch": 0.61, "learning_rate": 7.0778629271770726e-06, "loss": 0.5558, "step": 3502 }, { "epoch": 0.61, "learning_rate": 7.072497301175945e-06, "loss": 0.5546, "step": 3503 }, { "epoch": 0.61, "learning_rate": 7.067132596846203e-06, "loss": 0.5598, "step": 3504 }, { "epoch": 0.61, "learning_rate": 7.061768815876821e-06, "loss": 0.5548, "step": 3505 }, { "epoch": 0.61, "learning_rate": 7.056405959956492e-06, "loss": 0.572, "step": 3506 }, { "epoch": 0.61, "learning_rate": 7.051044030773619e-06, "loss": 0.5635, "step": 3507 }, { "epoch": 0.61, "learning_rate": 7.045683030016299e-06, "loss": 0.6217, "step": 3508 }, { "epoch": 0.61, "learning_rate": 7.040322959372352e-06, "loss": 0.5628, "step": 3509 }, { "epoch": 0.61, "learning_rate": 7.034963820529299e-06, "loss": 0.5521, "step": 3510 }, { "epoch": 0.61, "learning_rate": 7.029605615174365e-06, "loss": 0.5431, "step": 3511 }, { "epoch": 0.61, "learning_rate": 7.02424834499449e-06, "loss": 0.5652, "step": 3512 }, { "epoch": 0.61, "learning_rate": 7.018892011676308e-06, "loss": 0.5719, "step": 3513 }, { "epoch": 0.61, "learning_rate": 7.0135366169061665e-06, "loss": 0.5723, "step": 3514 }, { "epoch": 0.61, "learning_rate": 7.0081821623701165e-06, "loss": 0.5645, "step": 3515 }, { "epoch": 0.61, "learning_rate": 7.002828649753908e-06, "loss": 0.5537, "step": 3516 }, { "epoch": 0.61, "learning_rate": 6.997476080743002e-06, "loss": 0.5713, "step": 3517 }, { "epoch": 0.61, "learning_rate": 6.9921244570225536e-06, "loss": 0.5624, "step": 3518 }, { "epoch": 0.61, "learning_rate": 6.986773780277427e-06, "loss": 0.5695, "step": 3519 }, { "epoch": 0.61, "learning_rate": 6.981424052192187e-06, "loss": 0.5793, "step": 3520 }, { "epoch": 0.61, "learning_rate": 6.976075274451102e-06, "loss": 0.559, "step": 3521 }, { "epoch": 0.61, "learning_rate": 6.97072744873813e-06, "loss": 0.5759, "step": 3522 }, { "epoch": 0.61, "learning_rate": 6.965380576736943e-06, "loss": 0.5712, "step": 3523 }, { "epoch": 0.61, "learning_rate": 6.960034660130904e-06, "loss": 0.5686, "step": 3524 }, { "epoch": 0.61, "learning_rate": 6.9546897006030815e-06, "loss": 0.5567, "step": 3525 }, { "epoch": 0.61, "learning_rate": 6.949345699836239e-06, "loss": 0.5555, "step": 3526 }, { "epoch": 0.61, "learning_rate": 6.9440026595128365e-06, "loss": 0.5686, "step": 3527 }, { "epoch": 0.61, "learning_rate": 6.938660581315036e-06, "loss": 0.5822, "step": 3528 }, { "epoch": 0.61, "learning_rate": 6.933319466924693e-06, "loss": 0.5749, "step": 3529 }, { "epoch": 0.61, "learning_rate": 6.927979318023364e-06, "loss": 0.5632, "step": 3530 }, { "epoch": 0.61, "learning_rate": 6.922640136292293e-06, "loss": 0.5601, "step": 3531 }, { "epoch": 0.61, "learning_rate": 6.917301923412427e-06, "loss": 0.567, "step": 3532 }, { "epoch": 0.61, "learning_rate": 6.911964681064411e-06, "loss": 0.5594, "step": 3533 }, { "epoch": 0.61, "learning_rate": 6.906628410928573e-06, "loss": 0.5697, "step": 3534 }, { "epoch": 0.61, "learning_rate": 6.901293114684949e-06, "loss": 0.5654, "step": 3535 }, { "epoch": 0.61, "learning_rate": 6.895958794013251e-06, "loss": 0.5649, "step": 3536 }, { "epoch": 0.61, "learning_rate": 6.890625450592897e-06, "loss": 0.5702, "step": 3537 }, { "epoch": 0.61, "learning_rate": 6.8852930861029955e-06, "loss": 0.5678, "step": 3538 }, { "epoch": 0.61, "learning_rate": 6.8799617022223465e-06, "loss": 0.5642, "step": 3539 }, { "epoch": 0.61, "learning_rate": 6.874631300629435e-06, "loss": 0.5569, "step": 3540 }, { "epoch": 0.61, "learning_rate": 6.8693018830024485e-06, "loss": 0.5597, "step": 3541 }, { "epoch": 0.61, "learning_rate": 6.863973451019252e-06, "loss": 0.5676, "step": 3542 }, { "epoch": 0.61, "learning_rate": 6.85864600635741e-06, "loss": 0.5627, "step": 3543 }, { "epoch": 0.61, "learning_rate": 6.853319550694168e-06, "loss": 0.5613, "step": 3544 }, { "epoch": 0.61, "learning_rate": 6.847994085706469e-06, "loss": 0.5675, "step": 3545 }, { "epoch": 0.61, "learning_rate": 6.84266961307094e-06, "loss": 0.556, "step": 3546 }, { "epoch": 0.61, "learning_rate": 6.837346134463889e-06, "loss": 0.577, "step": 3547 }, { "epoch": 0.61, "learning_rate": 6.832023651561324e-06, "loss": 0.575, "step": 3548 }, { "epoch": 0.61, "learning_rate": 6.826702166038932e-06, "loss": 0.5648, "step": 3549 }, { "epoch": 0.61, "learning_rate": 6.821381679572079e-06, "loss": 0.5796, "step": 3550 }, { "epoch": 0.62, "learning_rate": 6.81606219383583e-06, "loss": 0.5549, "step": 3551 }, { "epoch": 0.62, "learning_rate": 6.810743710504931e-06, "loss": 0.5816, "step": 3552 }, { "epoch": 0.62, "learning_rate": 6.805426231253806e-06, "loss": 0.5692, "step": 3553 }, { "epoch": 0.62, "learning_rate": 6.80010975775657e-06, "loss": 0.5672, "step": 3554 }, { "epoch": 0.62, "learning_rate": 6.794794291687018e-06, "loss": 0.5726, "step": 3555 }, { "epoch": 0.62, "learning_rate": 6.789479834718627e-06, "loss": 0.5666, "step": 3556 }, { "epoch": 0.62, "learning_rate": 6.784166388524562e-06, "loss": 0.5779, "step": 3557 }, { "epoch": 0.62, "learning_rate": 6.778853954777662e-06, "loss": 0.5781, "step": 3558 }, { "epoch": 0.62, "learning_rate": 6.773542535150453e-06, "loss": 0.5567, "step": 3559 }, { "epoch": 0.62, "learning_rate": 6.768232131315136e-06, "loss": 0.5655, "step": 3560 }, { "epoch": 0.62, "learning_rate": 6.762922744943601e-06, "loss": 0.5574, "step": 3561 }, { "epoch": 0.62, "learning_rate": 6.757614377707409e-06, "loss": 0.5621, "step": 3562 }, { "epoch": 0.62, "learning_rate": 6.752307031277807e-06, "loss": 0.5666, "step": 3563 }, { "epoch": 0.62, "learning_rate": 6.7470007073257105e-06, "loss": 0.5738, "step": 3564 }, { "epoch": 0.62, "learning_rate": 6.741695407521727e-06, "loss": 0.561, "step": 3565 }, { "epoch": 0.62, "learning_rate": 6.736391133536127e-06, "loss": 0.5635, "step": 3566 }, { "epoch": 0.62, "learning_rate": 6.731087887038874e-06, "loss": 0.5515, "step": 3567 }, { "epoch": 0.62, "learning_rate": 6.725785669699593e-06, "loss": 0.5818, "step": 3568 }, { "epoch": 0.62, "learning_rate": 6.720484483187592e-06, "loss": 0.5616, "step": 3569 }, { "epoch": 0.62, "learning_rate": 6.71518432917186e-06, "loss": 0.5686, "step": 3570 }, { "epoch": 0.62, "learning_rate": 6.709885209321047e-06, "loss": 0.5702, "step": 3571 }, { "epoch": 0.62, "learning_rate": 6.70458712530349e-06, "loss": 0.563, "step": 3572 }, { "epoch": 0.62, "learning_rate": 6.699290078787193e-06, "loss": 0.5539, "step": 3573 }, { "epoch": 0.62, "learning_rate": 6.693994071439836e-06, "loss": 0.5701, "step": 3574 }, { "epoch": 0.62, "learning_rate": 6.688699104928771e-06, "loss": 0.5591, "step": 3575 }, { "epoch": 0.62, "learning_rate": 6.683405180921023e-06, "loss": 0.5738, "step": 3576 }, { "epoch": 0.62, "learning_rate": 6.678112301083292e-06, "loss": 0.5499, "step": 3577 }, { "epoch": 0.62, "learning_rate": 6.672820467081942e-06, "loss": 0.573, "step": 3578 }, { "epoch": 0.62, "learning_rate": 6.667529680583008e-06, "loss": 0.564, "step": 3579 }, { "epoch": 0.62, "learning_rate": 6.662239943252204e-06, "loss": 0.5735, "step": 3580 }, { "epoch": 0.62, "learning_rate": 6.6569512567549064e-06, "loss": 0.5576, "step": 3581 }, { "epoch": 0.62, "learning_rate": 6.651663622756161e-06, "loss": 0.5617, "step": 3582 }, { "epoch": 0.62, "learning_rate": 6.646377042920688e-06, "loss": 0.572, "step": 3583 }, { "epoch": 0.62, "learning_rate": 6.6410915189128675e-06, "loss": 0.5615, "step": 3584 }, { "epoch": 0.62, "learning_rate": 6.635807052396757e-06, "loss": 0.5673, "step": 3585 }, { "epoch": 0.62, "learning_rate": 6.630523645036066e-06, "loss": 0.5708, "step": 3586 }, { "epoch": 0.62, "learning_rate": 6.625241298494191e-06, "loss": 0.5589, "step": 3587 }, { "epoch": 0.62, "learning_rate": 6.619960014434175e-06, "loss": 0.5625, "step": 3588 }, { "epoch": 0.62, "learning_rate": 6.614679794518737e-06, "loss": 0.5568, "step": 3589 }, { "epoch": 0.62, "learning_rate": 6.609400640410264e-06, "loss": 0.5585, "step": 3590 }, { "epoch": 0.62, "learning_rate": 6.604122553770799e-06, "loss": 0.5713, "step": 3591 }, { "epoch": 0.62, "learning_rate": 6.598845536262046e-06, "loss": 0.5844, "step": 3592 }, { "epoch": 0.62, "learning_rate": 6.593569589545389e-06, "loss": 0.5567, "step": 3593 }, { "epoch": 0.62, "learning_rate": 6.588294715281857e-06, "loss": 0.5735, "step": 3594 }, { "epoch": 0.62, "learning_rate": 6.5830209151321525e-06, "loss": 0.5726, "step": 3595 }, { "epoch": 0.62, "learning_rate": 6.577748190756636e-06, "loss": 0.5528, "step": 3596 }, { "epoch": 0.62, "learning_rate": 6.572476543815328e-06, "loss": 0.5702, "step": 3597 }, { "epoch": 0.62, "learning_rate": 6.5672059759679145e-06, "loss": 0.581, "step": 3598 }, { "epoch": 0.62, "learning_rate": 6.561936488873735e-06, "loss": 0.5505, "step": 3599 }, { "epoch": 0.62, "learning_rate": 6.5566680841917955e-06, "loss": 0.5596, "step": 3600 }, { "epoch": 0.62, "learning_rate": 6.551400763580756e-06, "loss": 0.5701, "step": 3601 }, { "epoch": 0.62, "learning_rate": 6.5461345286989375e-06, "loss": 0.5742, "step": 3602 }, { "epoch": 0.62, "learning_rate": 6.540869381204322e-06, "loss": 0.553, "step": 3603 }, { "epoch": 0.62, "learning_rate": 6.535605322754541e-06, "loss": 0.5881, "step": 3604 }, { "epoch": 0.62, "learning_rate": 6.530342355006897e-06, "loss": 0.5572, "step": 3605 }, { "epoch": 0.62, "learning_rate": 6.525080479618331e-06, "loss": 0.5753, "step": 3606 }, { "epoch": 0.62, "learning_rate": 6.5198196982454506e-06, "loss": 0.5701, "step": 3607 }, { "epoch": 0.62, "learning_rate": 6.514560012544521e-06, "loss": 0.5915, "step": 3608 }, { "epoch": 0.63, "learning_rate": 6.5093014241714595e-06, "loss": 0.5764, "step": 3609 }, { "epoch": 0.63, "learning_rate": 6.504043934781836e-06, "loss": 0.5499, "step": 3610 }, { "epoch": 0.63, "learning_rate": 6.4987875460308784e-06, "loss": 0.584, "step": 3611 }, { "epoch": 0.63, "learning_rate": 6.493532259573461e-06, "loss": 0.5636, "step": 3612 }, { "epoch": 0.63, "learning_rate": 6.488278077064121e-06, "loss": 0.5566, "step": 3613 }, { "epoch": 0.63, "learning_rate": 6.483025000157041e-06, "loss": 0.5641, "step": 3614 }, { "epoch": 0.63, "learning_rate": 6.477773030506058e-06, "loss": 0.5582, "step": 3615 }, { "epoch": 0.63, "learning_rate": 6.47252216976466e-06, "loss": 0.5674, "step": 3616 }, { "epoch": 0.63, "learning_rate": 6.467272419585984e-06, "loss": 0.5645, "step": 3617 }, { "epoch": 0.63, "learning_rate": 6.4620237816228215e-06, "loss": 0.5662, "step": 3618 }, { "epoch": 0.63, "learning_rate": 6.456776257527611e-06, "loss": 0.5492, "step": 3619 }, { "epoch": 0.63, "learning_rate": 6.451529848952438e-06, "loss": 0.5504, "step": 3620 }, { "epoch": 0.63, "learning_rate": 6.446284557549039e-06, "loss": 0.5473, "step": 3621 }, { "epoch": 0.63, "learning_rate": 6.441040384968806e-06, "loss": 0.5656, "step": 3622 }, { "epoch": 0.63, "learning_rate": 6.435797332862765e-06, "loss": 0.5727, "step": 3623 }, { "epoch": 0.63, "learning_rate": 6.430555402881601e-06, "loss": 0.5502, "step": 3624 }, { "epoch": 0.63, "learning_rate": 6.425314596675637e-06, "loss": 0.5644, "step": 3625 }, { "epoch": 0.63, "learning_rate": 6.420074915894848e-06, "loss": 0.5614, "step": 3626 }, { "epoch": 0.63, "learning_rate": 6.4148363621888565e-06, "loss": 0.5571, "step": 3627 }, { "epoch": 0.63, "learning_rate": 6.40959893720692e-06, "loss": 0.5686, "step": 3628 }, { "epoch": 0.63, "learning_rate": 6.404362642597954e-06, "loss": 0.5805, "step": 3629 }, { "epoch": 0.63, "learning_rate": 6.399127480010505e-06, "loss": 0.568, "step": 3630 }, { "epoch": 0.63, "learning_rate": 6.393893451092772e-06, "loss": 0.5745, "step": 3631 }, { "epoch": 0.63, "learning_rate": 6.388660557492598e-06, "loss": 0.5714, "step": 3632 }, { "epoch": 0.63, "learning_rate": 6.38342880085746e-06, "loss": 0.566, "step": 3633 }, { "epoch": 0.63, "learning_rate": 6.378198182834487e-06, "loss": 0.5628, "step": 3634 }, { "epoch": 0.63, "learning_rate": 6.372968705070441e-06, "loss": 0.5932, "step": 3635 }, { "epoch": 0.63, "learning_rate": 6.367740369211727e-06, "loss": 0.5594, "step": 3636 }, { "epoch": 0.63, "learning_rate": 6.3625131769043966e-06, "loss": 0.5675, "step": 3637 }, { "epoch": 0.63, "learning_rate": 6.357287129794133e-06, "loss": 0.5769, "step": 3638 }, { "epoch": 0.63, "learning_rate": 6.352062229526266e-06, "loss": 0.5798, "step": 3639 }, { "epoch": 0.63, "learning_rate": 6.3468384777457605e-06, "loss": 0.583, "step": 3640 }, { "epoch": 0.63, "learning_rate": 6.341615876097218e-06, "loss": 0.5745, "step": 3641 }, { "epoch": 0.63, "learning_rate": 6.336394426224885e-06, "loss": 0.5443, "step": 3642 }, { "epoch": 0.63, "learning_rate": 6.331174129772635e-06, "loss": 0.5558, "step": 3643 }, { "epoch": 0.63, "learning_rate": 6.325954988383988e-06, "loss": 0.5602, "step": 3644 }, { "epoch": 0.63, "learning_rate": 6.320737003702098e-06, "loss": 0.5668, "step": 3645 }, { "epoch": 0.63, "learning_rate": 6.315520177369747e-06, "loss": 0.5522, "step": 3646 }, { "epoch": 0.63, "learning_rate": 6.310304511029366e-06, "loss": 0.5659, "step": 3647 }, { "epoch": 0.63, "learning_rate": 6.305090006323011e-06, "loss": 0.5628, "step": 3648 }, { "epoch": 0.63, "learning_rate": 6.29987666489237e-06, "loss": 0.5829, "step": 3649 }, { "epoch": 0.63, "learning_rate": 6.294664488378777e-06, "loss": 0.5706, "step": 3650 }, { "epoch": 0.63, "learning_rate": 6.289453478423183e-06, "loss": 0.565, "step": 3651 }, { "epoch": 0.63, "learning_rate": 6.2842436366661866e-06, "loss": 0.5592, "step": 3652 }, { "epoch": 0.63, "learning_rate": 6.279034964748012e-06, "loss": 0.5734, "step": 3653 }, { "epoch": 0.63, "learning_rate": 6.273827464308513e-06, "loss": 0.5648, "step": 3654 }, { "epoch": 0.63, "learning_rate": 6.2686211369871805e-06, "loss": 0.5524, "step": 3655 }, { "epoch": 0.63, "learning_rate": 6.263415984423129e-06, "loss": 0.5717, "step": 3656 }, { "epoch": 0.63, "learning_rate": 6.258212008255109e-06, "loss": 0.5743, "step": 3657 }, { "epoch": 0.63, "learning_rate": 6.253009210121499e-06, "loss": 0.5536, "step": 3658 }, { "epoch": 0.63, "learning_rate": 6.247807591660304e-06, "loss": 0.5638, "step": 3659 }, { "epoch": 0.63, "learning_rate": 6.242607154509163e-06, "loss": 0.5575, "step": 3660 }, { "epoch": 0.63, "learning_rate": 6.237407900305334e-06, "loss": 0.5815, "step": 3661 }, { "epoch": 0.63, "learning_rate": 6.232209830685718e-06, "loss": 0.5732, "step": 3662 }, { "epoch": 0.63, "learning_rate": 6.2270129472868235e-06, "loss": 0.5552, "step": 3663 }, { "epoch": 0.63, "learning_rate": 6.221817251744801e-06, "loss": 0.5796, "step": 3664 }, { "epoch": 0.63, "learning_rate": 6.216622745695417e-06, "loss": 0.5566, "step": 3665 }, { "epoch": 0.64, "learning_rate": 6.211429430774073e-06, "loss": 0.5735, "step": 3666 }, { "epoch": 0.64, "learning_rate": 6.2062373086157855e-06, "loss": 0.5712, "step": 3667 }, { "epoch": 0.64, "learning_rate": 6.2010463808552065e-06, "loss": 0.563, "step": 3668 }, { "epoch": 0.64, "learning_rate": 6.195856649126599e-06, "loss": 0.5561, "step": 3669 }, { "epoch": 0.64, "learning_rate": 6.190668115063862e-06, "loss": 0.5722, "step": 3670 }, { "epoch": 0.64, "learning_rate": 6.18548078030051e-06, "loss": 0.5543, "step": 3671 }, { "epoch": 0.64, "learning_rate": 6.18029464646968e-06, "loss": 0.5715, "step": 3672 }, { "epoch": 0.64, "learning_rate": 6.175109715204136e-06, "loss": 0.5841, "step": 3673 }, { "epoch": 0.64, "learning_rate": 6.169925988136256e-06, "loss": 0.5578, "step": 3674 }, { "epoch": 0.64, "learning_rate": 6.164743466898046e-06, "loss": 0.5575, "step": 3675 }, { "epoch": 0.64, "learning_rate": 6.1595621531211305e-06, "loss": 0.5614, "step": 3676 }, { "epoch": 0.64, "learning_rate": 6.154382048436749e-06, "loss": 0.5756, "step": 3677 }, { "epoch": 0.64, "learning_rate": 6.149203154475764e-06, "loss": 0.568, "step": 3678 }, { "epoch": 0.64, "learning_rate": 6.14402547286866e-06, "loss": 0.5615, "step": 3679 }, { "epoch": 0.64, "learning_rate": 6.1388490052455305e-06, "loss": 0.5649, "step": 3680 }, { "epoch": 0.64, "learning_rate": 6.133673753236102e-06, "loss": 0.5586, "step": 3681 }, { "epoch": 0.64, "learning_rate": 6.1284997184697e-06, "loss": 0.5611, "step": 3682 }, { "epoch": 0.64, "learning_rate": 6.123326902575282e-06, "loss": 0.5604, "step": 3683 }, { "epoch": 0.64, "learning_rate": 6.118155307181415e-06, "loss": 0.5564, "step": 3684 }, { "epoch": 0.64, "learning_rate": 6.11298493391628e-06, "loss": 0.5675, "step": 3685 }, { "epoch": 0.64, "learning_rate": 6.107815784407678e-06, "loss": 0.5714, "step": 3686 }, { "epoch": 0.64, "learning_rate": 6.102647860283021e-06, "loss": 0.5615, "step": 3687 }, { "epoch": 0.64, "learning_rate": 6.097481163169336e-06, "loss": 0.5485, "step": 3688 }, { "epoch": 0.64, "learning_rate": 6.092315694693267e-06, "loss": 0.5558, "step": 3689 }, { "epoch": 0.64, "learning_rate": 6.087151456481071e-06, "loss": 0.5652, "step": 3690 }, { "epoch": 0.64, "learning_rate": 6.081988450158605e-06, "loss": 0.5623, "step": 3691 }, { "epoch": 0.64, "learning_rate": 6.076826677351357e-06, "loss": 0.5675, "step": 3692 }, { "epoch": 0.64, "learning_rate": 6.071666139684414e-06, "loss": 0.5589, "step": 3693 }, { "epoch": 0.64, "learning_rate": 6.06650683878248e-06, "loss": 0.5536, "step": 3694 }, { "epoch": 0.64, "learning_rate": 6.061348776269869e-06, "loss": 0.5411, "step": 3695 }, { "epoch": 0.64, "learning_rate": 6.056191953770501e-06, "loss": 0.549, "step": 3696 }, { "epoch": 0.64, "learning_rate": 6.0510363729079104e-06, "loss": 0.571, "step": 3697 }, { "epoch": 0.64, "learning_rate": 6.045882035305237e-06, "loss": 0.5419, "step": 3698 }, { "epoch": 0.64, "learning_rate": 6.0407289425852335e-06, "loss": 0.5406, "step": 3699 }, { "epoch": 0.64, "learning_rate": 6.035577096370255e-06, "loss": 0.5566, "step": 3700 }, { "epoch": 0.64, "learning_rate": 6.030426498282269e-06, "loss": 0.5575, "step": 3701 }, { "epoch": 0.64, "learning_rate": 6.02527714994285e-06, "loss": 0.5524, "step": 3702 }, { "epoch": 0.64, "learning_rate": 6.020129052973174e-06, "loss": 0.5676, "step": 3703 }, { "epoch": 0.64, "learning_rate": 6.014982208994033e-06, "loss": 0.5672, "step": 3704 }, { "epoch": 0.64, "learning_rate": 6.0098366196258095e-06, "loss": 0.5755, "step": 3705 }, { "epoch": 0.64, "learning_rate": 6.004692286488503e-06, "loss": 0.5639, "step": 3706 }, { "epoch": 0.64, "learning_rate": 5.999549211201711e-06, "loss": 0.5578, "step": 3707 }, { "epoch": 0.64, "learning_rate": 5.994407395384645e-06, "loss": 0.5756, "step": 3708 }, { "epoch": 0.64, "learning_rate": 5.989266840656107e-06, "loss": 0.5669, "step": 3709 }, { "epoch": 0.64, "learning_rate": 5.984127548634511e-06, "loss": 0.5554, "step": 3710 }, { "epoch": 0.64, "learning_rate": 5.978989520937865e-06, "loss": 0.5559, "step": 3711 }, { "epoch": 0.64, "learning_rate": 5.973852759183793e-06, "loss": 0.5629, "step": 3712 }, { "epoch": 0.64, "learning_rate": 5.968717264989504e-06, "loss": 0.586, "step": 3713 }, { "epoch": 0.64, "learning_rate": 5.963583039971819e-06, "loss": 0.5536, "step": 3714 }, { "epoch": 0.64, "learning_rate": 5.9584500857471575e-06, "loss": 0.5632, "step": 3715 }, { "epoch": 0.64, "learning_rate": 5.953318403931533e-06, "loss": 0.5695, "step": 3716 }, { "epoch": 0.64, "learning_rate": 5.948187996140568e-06, "loss": 0.5698, "step": 3717 }, { "epoch": 0.64, "learning_rate": 5.943058863989478e-06, "loss": 0.5485, "step": 3718 }, { "epoch": 0.64, "learning_rate": 5.937931009093072e-06, "loss": 0.5678, "step": 3719 }, { "epoch": 0.64, "learning_rate": 5.932804433065767e-06, "loss": 0.5688, "step": 3720 }, { "epoch": 0.64, "learning_rate": 5.927679137521573e-06, "loss": 0.5645, "step": 3721 }, { "epoch": 0.64, "learning_rate": 5.922555124074096e-06, "loss": 0.5585, "step": 3722 }, { "epoch": 0.64, "learning_rate": 5.917432394336542e-06, "loss": 0.5564, "step": 3723 }, { "epoch": 0.65, "learning_rate": 5.912310949921705e-06, "loss": 0.5689, "step": 3724 }, { "epoch": 0.65, "learning_rate": 5.9071907924419805e-06, "loss": 0.5737, "step": 3725 }, { "epoch": 0.65, "learning_rate": 5.90207192350936e-06, "loss": 0.5597, "step": 3726 }, { "epoch": 0.65, "learning_rate": 5.896954344735426e-06, "loss": 0.552, "step": 3727 }, { "epoch": 0.65, "learning_rate": 5.891838057731358e-06, "loss": 0.5618, "step": 3728 }, { "epoch": 0.65, "learning_rate": 5.886723064107921e-06, "loss": 0.5664, "step": 3729 }, { "epoch": 0.65, "learning_rate": 5.881609365475485e-06, "loss": 0.5697, "step": 3730 }, { "epoch": 0.65, "learning_rate": 5.876496963443998e-06, "loss": 0.5649, "step": 3731 }, { "epoch": 0.65, "learning_rate": 5.871385859623018e-06, "loss": 0.5649, "step": 3732 }, { "epoch": 0.65, "learning_rate": 5.866276055621673e-06, "loss": 0.561, "step": 3733 }, { "epoch": 0.65, "learning_rate": 5.861167553048699e-06, "loss": 0.5728, "step": 3734 }, { "epoch": 0.65, "learning_rate": 5.856060353512414e-06, "loss": 0.566, "step": 3735 }, { "epoch": 0.65, "learning_rate": 5.850954458620728e-06, "loss": 0.5657, "step": 3736 }, { "epoch": 0.65, "learning_rate": 5.845849869981137e-06, "loss": 0.5671, "step": 3737 }, { "epoch": 0.65, "learning_rate": 5.840746589200732e-06, "loss": 0.5711, "step": 3738 }, { "epoch": 0.65, "learning_rate": 5.835644617886189e-06, "loss": 0.5591, "step": 3739 }, { "epoch": 0.65, "learning_rate": 5.830543957643773e-06, "loss": 0.5841, "step": 3740 }, { "epoch": 0.65, "learning_rate": 5.825444610079326e-06, "loss": 0.5775, "step": 3741 }, { "epoch": 0.65, "learning_rate": 5.820346576798297e-06, "loss": 0.5575, "step": 3742 }, { "epoch": 0.65, "learning_rate": 5.815249859405704e-06, "loss": 0.5581, "step": 3743 }, { "epoch": 0.65, "learning_rate": 5.810154459506152e-06, "loss": 0.5736, "step": 3744 }, { "epoch": 0.65, "learning_rate": 5.805060378703848e-06, "loss": 0.559, "step": 3745 }, { "epoch": 0.65, "learning_rate": 5.799967618602565e-06, "loss": 0.5638, "step": 3746 }, { "epoch": 0.65, "learning_rate": 5.794876180805659e-06, "loss": 0.5593, "step": 3747 }, { "epoch": 0.65, "learning_rate": 5.789786066916089e-06, "loss": 0.5719, "step": 3748 }, { "epoch": 0.65, "learning_rate": 5.78469727853638e-06, "loss": 0.5796, "step": 3749 }, { "epoch": 0.65, "learning_rate": 5.779609817268646e-06, "loss": 0.5573, "step": 3750 }, { "epoch": 0.65, "learning_rate": 5.774523684714577e-06, "loss": 0.5636, "step": 3751 }, { "epoch": 0.65, "learning_rate": 5.769438882475459e-06, "loss": 0.5585, "step": 3752 }, { "epoch": 0.65, "learning_rate": 5.764355412152147e-06, "loss": 0.5605, "step": 3753 }, { "epoch": 0.65, "learning_rate": 5.759273275345074e-06, "loss": 0.5613, "step": 3754 }, { "epoch": 0.65, "learning_rate": 5.7541924736542694e-06, "loss": 0.5528, "step": 3755 }, { "epoch": 0.65, "learning_rate": 5.7491130086793275e-06, "loss": 0.5442, "step": 3756 }, { "epoch": 0.65, "learning_rate": 5.7440348820194204e-06, "loss": 0.5868, "step": 3757 }, { "epoch": 0.65, "learning_rate": 5.7389580952733136e-06, "loss": 0.5761, "step": 3758 }, { "epoch": 0.65, "learning_rate": 5.733882650039338e-06, "loss": 0.5651, "step": 3759 }, { "epoch": 0.65, "learning_rate": 5.728808547915405e-06, "loss": 0.5894, "step": 3760 }, { "epoch": 0.65, "learning_rate": 5.723735790499005e-06, "loss": 0.5658, "step": 3761 }, { "epoch": 0.65, "learning_rate": 5.718664379387205e-06, "loss": 0.5791, "step": 3762 }, { "epoch": 0.65, "learning_rate": 5.71359431617664e-06, "loss": 0.5566, "step": 3763 }, { "epoch": 0.65, "learning_rate": 5.7085256024635404e-06, "loss": 0.5718, "step": 3764 }, { "epoch": 0.65, "learning_rate": 5.703458239843691e-06, "loss": 0.5575, "step": 3765 }, { "epoch": 0.65, "learning_rate": 5.698392229912463e-06, "loss": 0.5491, "step": 3766 }, { "epoch": 0.65, "learning_rate": 5.693327574264791e-06, "loss": 0.5542, "step": 3767 }, { "epoch": 0.65, "learning_rate": 5.688264274495201e-06, "loss": 0.5703, "step": 3768 }, { "epoch": 0.65, "learning_rate": 5.683202332197775e-06, "loss": 0.5588, "step": 3769 }, { "epoch": 0.65, "learning_rate": 5.6781417489661725e-06, "loss": 0.5685, "step": 3770 }, { "epoch": 0.65, "learning_rate": 5.673082526393634e-06, "loss": 0.5742, "step": 3771 }, { "epoch": 0.65, "learning_rate": 5.6680246660729584e-06, "loss": 0.5809, "step": 3772 }, { "epoch": 0.65, "learning_rate": 5.662968169596525e-06, "loss": 0.5545, "step": 3773 }, { "epoch": 0.65, "learning_rate": 5.657913038556279e-06, "loss": 0.547, "step": 3774 }, { "epoch": 0.65, "learning_rate": 5.652859274543738e-06, "loss": 0.5533, "step": 3775 }, { "epoch": 0.65, "learning_rate": 5.647806879149983e-06, "loss": 0.5559, "step": 3776 }, { "epoch": 0.65, "learning_rate": 5.642755853965678e-06, "loss": 0.5818, "step": 3777 }, { "epoch": 0.65, "learning_rate": 5.637706200581043e-06, "loss": 0.5665, "step": 3778 }, { "epoch": 0.65, "learning_rate": 5.63265792058587e-06, "loss": 0.5579, "step": 3779 }, { "epoch": 0.65, "learning_rate": 5.627611015569516e-06, "loss": 0.5554, "step": 3780 }, { "epoch": 0.65, "learning_rate": 5.622565487120915e-06, "loss": 0.5797, "step": 3781 }, { "epoch": 0.66, "learning_rate": 5.617521336828556e-06, "loss": 0.5514, "step": 3782 }, { "epoch": 0.66, "learning_rate": 5.612478566280497e-06, "loss": 0.575, "step": 3783 }, { "epoch": 0.66, "learning_rate": 5.607437177064367e-06, "loss": 0.5704, "step": 3784 }, { "epoch": 0.66, "learning_rate": 5.602397170767357e-06, "loss": 0.5634, "step": 3785 }, { "epoch": 0.66, "learning_rate": 5.597358548976221e-06, "loss": 0.5522, "step": 3786 }, { "epoch": 0.66, "learning_rate": 5.5923213132772715e-06, "loss": 0.5658, "step": 3787 }, { "epoch": 0.66, "learning_rate": 5.587285465256405e-06, "loss": 0.5572, "step": 3788 }, { "epoch": 0.66, "learning_rate": 5.5822510064990534e-06, "loss": 0.5496, "step": 3789 }, { "epoch": 0.66, "learning_rate": 5.577217938590232e-06, "loss": 0.5694, "step": 3790 }, { "epoch": 0.66, "learning_rate": 5.572186263114512e-06, "loss": 0.5564, "step": 3791 }, { "epoch": 0.66, "learning_rate": 5.567155981656025e-06, "loss": 0.5766, "step": 3792 }, { "epoch": 0.66, "learning_rate": 5.5621270957984575e-06, "loss": 0.5621, "step": 3793 }, { "epoch": 0.66, "learning_rate": 5.557099607125073e-06, "loss": 0.5685, "step": 3794 }, { "epoch": 0.66, "learning_rate": 5.5520735172186846e-06, "loss": 0.5765, "step": 3795 }, { "epoch": 0.66, "learning_rate": 5.547048827661657e-06, "loss": 0.5596, "step": 3796 }, { "epoch": 0.66, "learning_rate": 5.542025540035935e-06, "loss": 0.5482, "step": 3797 }, { "epoch": 0.66, "learning_rate": 5.537003655923003e-06, "loss": 0.5432, "step": 3798 }, { "epoch": 0.66, "learning_rate": 5.531983176903913e-06, "loss": 0.559, "step": 3799 }, { "epoch": 0.66, "learning_rate": 5.5269641045592645e-06, "loss": 0.5615, "step": 3800 }, { "epoch": 0.66, "learning_rate": 5.5219464404692346e-06, "loss": 0.5681, "step": 3801 }, { "epoch": 0.66, "learning_rate": 5.516930186213538e-06, "loss": 0.5781, "step": 3802 }, { "epoch": 0.66, "learning_rate": 5.511915343371452e-06, "loss": 0.5604, "step": 3803 }, { "epoch": 0.66, "learning_rate": 5.506901913521808e-06, "loss": 0.5787, "step": 3804 }, { "epoch": 0.66, "learning_rate": 5.501889898242996e-06, "loss": 0.5661, "step": 3805 }, { "epoch": 0.66, "learning_rate": 5.496879299112954e-06, "loss": 0.5677, "step": 3806 }, { "epoch": 0.66, "learning_rate": 5.491870117709186e-06, "loss": 0.5622, "step": 3807 }, { "epoch": 0.66, "learning_rate": 5.486862355608738e-06, "loss": 0.5678, "step": 3808 }, { "epoch": 0.66, "learning_rate": 5.481856014388211e-06, "loss": 0.5402, "step": 3809 }, { "epoch": 0.66, "learning_rate": 5.476851095623767e-06, "loss": 0.5583, "step": 3810 }, { "epoch": 0.66, "learning_rate": 5.471847600891114e-06, "loss": 0.5641, "step": 3811 }, { "epoch": 0.66, "learning_rate": 5.4668455317655086e-06, "loss": 0.5643, "step": 3812 }, { "epoch": 0.66, "learning_rate": 5.461844889821759e-06, "loss": 0.5551, "step": 3813 }, { "epoch": 0.66, "learning_rate": 5.456845676634235e-06, "loss": 0.5622, "step": 3814 }, { "epoch": 0.66, "learning_rate": 5.451847893776845e-06, "loss": 0.557, "step": 3815 }, { "epoch": 0.66, "learning_rate": 5.446851542823051e-06, "loss": 0.5689, "step": 3816 }, { "epoch": 0.66, "learning_rate": 5.441856625345863e-06, "loss": 0.5479, "step": 3817 }, { "epoch": 0.66, "learning_rate": 5.4368631429178406e-06, "loss": 0.5524, "step": 3818 }, { "epoch": 0.66, "learning_rate": 5.43187109711109e-06, "loss": 0.5599, "step": 3819 }, { "epoch": 0.66, "learning_rate": 5.4268804894972705e-06, "loss": 0.57, "step": 3820 }, { "epoch": 0.66, "learning_rate": 5.421891321647583e-06, "loss": 0.5574, "step": 3821 }, { "epoch": 0.66, "learning_rate": 5.416903595132774e-06, "loss": 0.5607, "step": 3822 }, { "epoch": 0.66, "learning_rate": 5.4119173115231475e-06, "loss": 0.5661, "step": 3823 }, { "epoch": 0.66, "learning_rate": 5.406932472388537e-06, "loss": 0.58, "step": 3824 }, { "epoch": 0.66, "learning_rate": 5.401949079298332e-06, "loss": 0.5653, "step": 3825 }, { "epoch": 0.66, "learning_rate": 5.396967133821461e-06, "loss": 0.5619, "step": 3826 }, { "epoch": 0.66, "learning_rate": 5.3919866375264055e-06, "loss": 0.5707, "step": 3827 }, { "epoch": 0.66, "learning_rate": 5.387007591981181e-06, "loss": 0.5828, "step": 3828 }, { "epoch": 0.66, "learning_rate": 5.382029998753349e-06, "loss": 0.5552, "step": 3829 }, { "epoch": 0.66, "learning_rate": 5.377053859410022e-06, "loss": 0.5822, "step": 3830 }, { "epoch": 0.66, "learning_rate": 5.372079175517838e-06, "loss": 0.5603, "step": 3831 }, { "epoch": 0.66, "learning_rate": 5.367105948642988e-06, "loss": 0.5619, "step": 3832 }, { "epoch": 0.66, "learning_rate": 5.3621341803512084e-06, "loss": 0.5721, "step": 3833 }, { "epoch": 0.66, "learning_rate": 5.357163872207769e-06, "loss": 0.5746, "step": 3834 }, { "epoch": 0.66, "learning_rate": 5.352195025777475e-06, "loss": 0.5789, "step": 3835 }, { "epoch": 0.66, "learning_rate": 5.347227642624687e-06, "loss": 0.5808, "step": 3836 }, { "epoch": 0.66, "learning_rate": 5.342261724313292e-06, "loss": 0.5625, "step": 3837 }, { "epoch": 0.66, "learning_rate": 5.337297272406717e-06, "loss": 0.5767, "step": 3838 }, { "epoch": 0.66, "learning_rate": 5.332334288467937e-06, "loss": 0.5608, "step": 3839 }, { "epoch": 0.67, "learning_rate": 5.327372774059454e-06, "loss": 0.5778, "step": 3840 }, { "epoch": 0.67, "learning_rate": 5.322412730743309e-06, "loss": 0.561, "step": 3841 }, { "epoch": 0.67, "learning_rate": 5.317454160081082e-06, "loss": 0.5562, "step": 3842 }, { "epoch": 0.67, "learning_rate": 5.312497063633897e-06, "loss": 0.5621, "step": 3843 }, { "epoch": 0.67, "learning_rate": 5.307541442962403e-06, "loss": 0.5718, "step": 3844 }, { "epoch": 0.67, "learning_rate": 5.302587299626778e-06, "loss": 0.5496, "step": 3845 }, { "epoch": 0.67, "learning_rate": 5.297634635186757e-06, "loss": 0.5746, "step": 3846 }, { "epoch": 0.67, "learning_rate": 5.2926834512015925e-06, "loss": 0.5641, "step": 3847 }, { "epoch": 0.67, "learning_rate": 5.287733749230071e-06, "loss": 0.5603, "step": 3848 }, { "epoch": 0.67, "learning_rate": 5.282785530830525e-06, "loss": 0.5587, "step": 3849 }, { "epoch": 0.67, "learning_rate": 5.277838797560808e-06, "loss": 0.5586, "step": 3850 }, { "epoch": 0.67, "learning_rate": 5.272893550978305e-06, "loss": 0.5736, "step": 3851 }, { "epoch": 0.67, "learning_rate": 5.267949792639947e-06, "loss": 0.5541, "step": 3852 }, { "epoch": 0.67, "learning_rate": 5.263007524102182e-06, "loss": 0.5672, "step": 3853 }, { "epoch": 0.67, "learning_rate": 5.258066746920993e-06, "loss": 0.5575, "step": 3854 }, { "epoch": 0.67, "learning_rate": 5.253127462651895e-06, "loss": 0.5739, "step": 3855 }, { "epoch": 0.67, "learning_rate": 5.248189672849935e-06, "loss": 0.562, "step": 3856 }, { "epoch": 0.67, "learning_rate": 5.243253379069685e-06, "loss": 0.5633, "step": 3857 }, { "epoch": 0.67, "learning_rate": 5.238318582865249e-06, "loss": 0.5427, "step": 3858 }, { "epoch": 0.67, "learning_rate": 5.233385285790258e-06, "loss": 0.589, "step": 3859 }, { "epoch": 0.67, "learning_rate": 5.228453489397871e-06, "loss": 0.5566, "step": 3860 }, { "epoch": 0.67, "learning_rate": 5.223523195240772e-06, "loss": 0.5489, "step": 3861 }, { "epoch": 0.67, "learning_rate": 5.218594404871181e-06, "loss": 0.5575, "step": 3862 }, { "epoch": 0.67, "learning_rate": 5.213667119840837e-06, "loss": 0.5561, "step": 3863 }, { "epoch": 0.67, "learning_rate": 5.2087413417010025e-06, "loss": 0.5494, "step": 3864 }, { "epoch": 0.67, "learning_rate": 5.203817072002477e-06, "loss": 0.5638, "step": 3865 }, { "epoch": 0.67, "learning_rate": 5.198894312295575e-06, "loss": 0.5444, "step": 3866 }, { "epoch": 0.67, "learning_rate": 5.193973064130136e-06, "loss": 0.5794, "step": 3867 }, { "epoch": 0.67, "learning_rate": 5.189053329055524e-06, "loss": 0.5544, "step": 3868 }, { "epoch": 0.67, "learning_rate": 5.184135108620638e-06, "loss": 0.5851, "step": 3869 }, { "epoch": 0.67, "learning_rate": 5.179218404373886e-06, "loss": 0.5596, "step": 3870 }, { "epoch": 0.67, "learning_rate": 5.174303217863199e-06, "loss": 0.5685, "step": 3871 }, { "epoch": 0.67, "learning_rate": 5.169389550636046e-06, "loss": 0.548, "step": 3872 }, { "epoch": 0.67, "learning_rate": 5.164477404239395e-06, "loss": 0.5603, "step": 3873 }, { "epoch": 0.67, "learning_rate": 5.159566780219749e-06, "loss": 0.5513, "step": 3874 }, { "epoch": 0.67, "learning_rate": 5.154657680123134e-06, "loss": 0.5686, "step": 3875 }, { "epoch": 0.67, "learning_rate": 5.149750105495088e-06, "loss": 0.5738, "step": 3876 }, { "epoch": 0.67, "learning_rate": 5.1448440578806715e-06, "loss": 0.5865, "step": 3877 }, { "epoch": 0.67, "learning_rate": 5.13993953882447e-06, "loss": 0.5556, "step": 3878 }, { "epoch": 0.67, "learning_rate": 5.135036549870578e-06, "loss": 0.5586, "step": 3879 }, { "epoch": 0.67, "learning_rate": 5.130135092562616e-06, "loss": 0.5568, "step": 3880 }, { "epoch": 0.67, "learning_rate": 5.125235168443714e-06, "loss": 0.5685, "step": 3881 }, { "epoch": 0.67, "learning_rate": 5.1203367790565314e-06, "loss": 0.5612, "step": 3882 }, { "epoch": 0.67, "learning_rate": 5.115439925943237e-06, "loss": 0.5717, "step": 3883 }, { "epoch": 0.67, "learning_rate": 5.110544610645509e-06, "loss": 0.5486, "step": 3884 }, { "epoch": 0.67, "learning_rate": 5.10565083470456e-06, "loss": 0.5623, "step": 3885 }, { "epoch": 0.67, "learning_rate": 5.100758599661104e-06, "loss": 0.5689, "step": 3886 }, { "epoch": 0.67, "learning_rate": 5.095867907055363e-06, "loss": 0.574, "step": 3887 }, { "epoch": 0.67, "learning_rate": 5.0909787584270955e-06, "loss": 0.5639, "step": 3888 }, { "epoch": 0.67, "learning_rate": 5.086091155315557e-06, "loss": 0.5826, "step": 3889 }, { "epoch": 0.67, "learning_rate": 5.081205099259516e-06, "loss": 0.5373, "step": 3890 }, { "epoch": 0.67, "learning_rate": 5.076320591797269e-06, "loss": 0.566, "step": 3891 }, { "epoch": 0.67, "learning_rate": 5.0714376344666095e-06, "loss": 0.5738, "step": 3892 }, { "epoch": 0.67, "learning_rate": 5.066556228804851e-06, "loss": 0.5613, "step": 3893 }, { "epoch": 0.67, "learning_rate": 5.061676376348808e-06, "loss": 0.5608, "step": 3894 }, { "epoch": 0.67, "learning_rate": 5.056798078634826e-06, "loss": 0.5607, "step": 3895 }, { "epoch": 0.67, "learning_rate": 5.0519213371987415e-06, "loss": 0.5497, "step": 3896 }, { "epoch": 0.68, "learning_rate": 5.047046153575906e-06, "loss": 0.5649, "step": 3897 }, { "epoch": 0.68, "learning_rate": 5.042172529301193e-06, "loss": 0.5615, "step": 3898 }, { "epoch": 0.68, "learning_rate": 5.0373004659089675e-06, "loss": 0.5714, "step": 3899 }, { "epoch": 0.68, "learning_rate": 5.032429964933112e-06, "loss": 0.5716, "step": 3900 }, { "epoch": 0.68, "learning_rate": 5.027561027907017e-06, "loss": 0.5887, "step": 3901 }, { "epoch": 0.68, "learning_rate": 5.02269365636358e-06, "loss": 0.56, "step": 3902 }, { "epoch": 0.68, "learning_rate": 5.017827851835199e-06, "loss": 0.5586, "step": 3903 }, { "epoch": 0.68, "learning_rate": 5.012963615853793e-06, "loss": 0.5523, "step": 3904 }, { "epoch": 0.68, "learning_rate": 5.008100949950776e-06, "loss": 0.5558, "step": 3905 }, { "epoch": 0.68, "learning_rate": 5.00323985565707e-06, "loss": 0.5718, "step": 3906 }, { "epoch": 0.68, "learning_rate": 4.998380334503099e-06, "loss": 0.5694, "step": 3907 }, { "epoch": 0.68, "learning_rate": 4.993522388018803e-06, "loss": 0.5691, "step": 3908 }, { "epoch": 0.68, "learning_rate": 4.988666017733615e-06, "loss": 0.5627, "step": 3909 }, { "epoch": 0.68, "learning_rate": 4.9838112251764725e-06, "loss": 0.5641, "step": 3910 }, { "epoch": 0.68, "learning_rate": 4.978958011875825e-06, "loss": 0.5677, "step": 3911 }, { "epoch": 0.68, "learning_rate": 4.974106379359618e-06, "loss": 0.5632, "step": 3912 }, { "epoch": 0.68, "learning_rate": 4.9692563291552945e-06, "loss": 0.5625, "step": 3913 }, { "epoch": 0.68, "learning_rate": 4.964407862789817e-06, "loss": 0.5476, "step": 3914 }, { "epoch": 0.68, "learning_rate": 4.9595609817896276e-06, "loss": 0.5715, "step": 3915 }, { "epoch": 0.68, "learning_rate": 4.9547156876806765e-06, "loss": 0.5553, "step": 3916 }, { "epoch": 0.68, "learning_rate": 4.949871981988429e-06, "loss": 0.5612, "step": 3917 }, { "epoch": 0.68, "learning_rate": 4.9450298662378306e-06, "loss": 0.5587, "step": 3918 }, { "epoch": 0.68, "learning_rate": 4.9401893419533355e-06, "loss": 0.5717, "step": 3919 }, { "epoch": 0.68, "learning_rate": 4.935350410658891e-06, "loss": 0.5459, "step": 3920 }, { "epoch": 0.68, "learning_rate": 4.930513073877956e-06, "loss": 0.5712, "step": 3921 }, { "epoch": 0.68, "learning_rate": 4.925677333133475e-06, "loss": 0.5674, "step": 3922 }, { "epoch": 0.68, "learning_rate": 4.920843189947888e-06, "loss": 0.5718, "step": 3923 }, { "epoch": 0.68, "learning_rate": 4.916010645843148e-06, "loss": 0.5691, "step": 3924 }, { "epoch": 0.68, "learning_rate": 4.9111797023406884e-06, "loss": 0.5747, "step": 3925 }, { "epoch": 0.68, "learning_rate": 4.9063503609614405e-06, "loss": 0.5614, "step": 3926 }, { "epoch": 0.68, "learning_rate": 4.901522623225845e-06, "loss": 0.5663, "step": 3927 }, { "epoch": 0.68, "learning_rate": 4.896696490653826e-06, "loss": 0.5661, "step": 3928 }, { "epoch": 0.68, "learning_rate": 4.891871964764794e-06, "loss": 0.5643, "step": 3929 }, { "epoch": 0.68, "learning_rate": 4.887049047077674e-06, "loss": 0.5582, "step": 3930 }, { "epoch": 0.68, "learning_rate": 4.882227739110873e-06, "loss": 0.5402, "step": 3931 }, { "epoch": 0.68, "learning_rate": 4.8774080423822866e-06, "loss": 0.5654, "step": 3932 }, { "epoch": 0.68, "learning_rate": 4.8725899584093184e-06, "loss": 0.5562, "step": 3933 }, { "epoch": 0.68, "learning_rate": 4.867773488708851e-06, "loss": 0.5702, "step": 3934 }, { "epoch": 0.68, "learning_rate": 4.8629586347972625e-06, "loss": 0.5923, "step": 3935 }, { "epoch": 0.68, "learning_rate": 4.8581453981904205e-06, "loss": 0.5345, "step": 3936 }, { "epoch": 0.68, "learning_rate": 4.853333780403691e-06, "loss": 0.5623, "step": 3937 }, { "epoch": 0.68, "learning_rate": 4.848523782951921e-06, "loss": 0.5585, "step": 3938 }, { "epoch": 0.68, "learning_rate": 4.843715407349451e-06, "loss": 0.5648, "step": 3939 }, { "epoch": 0.68, "learning_rate": 4.838908655110116e-06, "loss": 0.5568, "step": 3940 }, { "epoch": 0.68, "learning_rate": 4.834103527747233e-06, "loss": 0.5754, "step": 3941 }, { "epoch": 0.68, "learning_rate": 4.829300026773608e-06, "loss": 0.5614, "step": 3942 }, { "epoch": 0.68, "learning_rate": 4.824498153701537e-06, "loss": 0.5757, "step": 3943 }, { "epoch": 0.68, "learning_rate": 4.819697910042803e-06, "loss": 0.5366, "step": 3944 }, { "epoch": 0.68, "learning_rate": 4.814899297308673e-06, "loss": 0.5577, "step": 3945 }, { "epoch": 0.68, "learning_rate": 4.810102317009912e-06, "loss": 0.5751, "step": 3946 }, { "epoch": 0.68, "learning_rate": 4.8053069706567555e-06, "loss": 0.5696, "step": 3947 }, { "epoch": 0.68, "learning_rate": 4.8005132597589355e-06, "loss": 0.5502, "step": 3948 }, { "epoch": 0.68, "learning_rate": 4.79572118582566e-06, "loss": 0.5771, "step": 3949 }, { "epoch": 0.68, "learning_rate": 4.790930750365635e-06, "loss": 0.5682, "step": 3950 }, { "epoch": 0.68, "learning_rate": 4.786141954887037e-06, "loss": 0.5772, "step": 3951 }, { "epoch": 0.68, "learning_rate": 4.78135480089753e-06, "loss": 0.554, "step": 3952 }, { "epoch": 0.68, "learning_rate": 4.77656928990427e-06, "loss": 0.5624, "step": 3953 }, { "epoch": 0.68, "learning_rate": 4.771785423413885e-06, "loss": 0.5609, "step": 3954 }, { "epoch": 0.69, "learning_rate": 4.76700320293249e-06, "loss": 0.5544, "step": 3955 }, { "epoch": 0.69, "learning_rate": 4.762222629965678e-06, "loss": 0.5734, "step": 3956 }, { "epoch": 0.69, "learning_rate": 4.75744370601853e-06, "loss": 0.5718, "step": 3957 }, { "epoch": 0.69, "learning_rate": 4.752666432595596e-06, "loss": 0.5479, "step": 3958 }, { "epoch": 0.69, "learning_rate": 4.747890811200926e-06, "loss": 0.5594, "step": 3959 }, { "epoch": 0.69, "learning_rate": 4.74311684333803e-06, "loss": 0.5681, "step": 3960 }, { "epoch": 0.69, "learning_rate": 4.738344530509911e-06, "loss": 0.5693, "step": 3961 }, { "epoch": 0.69, "learning_rate": 4.7335738742190366e-06, "loss": 0.5652, "step": 3962 }, { "epoch": 0.69, "learning_rate": 4.728804875967372e-06, "loss": 0.5503, "step": 3963 }, { "epoch": 0.69, "learning_rate": 4.724037537256346e-06, "loss": 0.5649, "step": 3964 }, { "epoch": 0.69, "learning_rate": 4.719271859586865e-06, "loss": 0.5737, "step": 3965 }, { "epoch": 0.69, "learning_rate": 4.714507844459325e-06, "loss": 0.5466, "step": 3966 }, { "epoch": 0.69, "learning_rate": 4.709745493373585e-06, "loss": 0.5689, "step": 3967 }, { "epoch": 0.69, "learning_rate": 4.704984807828987e-06, "loss": 0.5596, "step": 3968 }, { "epoch": 0.69, "learning_rate": 4.700225789324343e-06, "loss": 0.5544, "step": 3969 }, { "epoch": 0.69, "learning_rate": 4.695468439357954e-06, "loss": 0.5751, "step": 3970 }, { "epoch": 0.69, "learning_rate": 4.690712759427571e-06, "loss": 0.5588, "step": 3971 }, { "epoch": 0.69, "learning_rate": 4.685958751030446e-06, "loss": 0.5655, "step": 3972 }, { "epoch": 0.69, "learning_rate": 4.681206415663289e-06, "loss": 0.5685, "step": 3973 }, { "epoch": 0.69, "learning_rate": 4.6764557548222854e-06, "loss": 0.5743, "step": 3974 }, { "epoch": 0.69, "learning_rate": 4.671706770003094e-06, "loss": 0.5673, "step": 3975 }, { "epoch": 0.69, "learning_rate": 4.666959462700852e-06, "loss": 0.5638, "step": 3976 }, { "epoch": 0.69, "learning_rate": 4.6622138344101605e-06, "loss": 0.5409, "step": 3977 }, { "epoch": 0.69, "learning_rate": 4.657469886625093e-06, "loss": 0.5657, "step": 3978 }, { "epoch": 0.69, "learning_rate": 4.652727620839199e-06, "loss": 0.568, "step": 3979 }, { "epoch": 0.69, "learning_rate": 4.647987038545496e-06, "loss": 0.574, "step": 3980 }, { "epoch": 0.69, "learning_rate": 4.643248141236469e-06, "loss": 0.571, "step": 3981 }, { "epoch": 0.69, "learning_rate": 4.638510930404069e-06, "loss": 0.5755, "step": 3982 }, { "epoch": 0.69, "learning_rate": 4.633775407539731e-06, "loss": 0.5919, "step": 3983 }, { "epoch": 0.69, "learning_rate": 4.629041574134346e-06, "loss": 0.5851, "step": 3984 }, { "epoch": 0.69, "learning_rate": 4.624309431678273e-06, "loss": 0.5536, "step": 3985 }, { "epoch": 0.69, "learning_rate": 4.619578981661341e-06, "loss": 0.5695, "step": 3986 }, { "epoch": 0.69, "learning_rate": 4.614850225572851e-06, "loss": 0.561, "step": 3987 }, { "epoch": 0.69, "learning_rate": 4.6101231649015606e-06, "loss": 0.5444, "step": 3988 }, { "epoch": 0.69, "learning_rate": 4.605397801135707e-06, "loss": 0.5669, "step": 3989 }, { "epoch": 0.69, "learning_rate": 4.600674135762982e-06, "loss": 0.5739, "step": 3990 }, { "epoch": 0.69, "learning_rate": 4.595952170270542e-06, "loss": 0.5691, "step": 3991 }, { "epoch": 0.69, "learning_rate": 4.591231906145022e-06, "loss": 0.5565, "step": 3992 }, { "epoch": 0.69, "learning_rate": 4.586513344872505e-06, "loss": 0.5775, "step": 3993 }, { "epoch": 0.69, "learning_rate": 4.581796487938548e-06, "loss": 0.5714, "step": 3994 }, { "epoch": 0.69, "learning_rate": 4.577081336828163e-06, "loss": 0.5596, "step": 3995 }, { "epoch": 0.69, "learning_rate": 4.572367893025839e-06, "loss": 0.5533, "step": 3996 }, { "epoch": 0.69, "learning_rate": 4.5676561580155125e-06, "loss": 0.5755, "step": 3997 }, { "epoch": 0.69, "learning_rate": 4.562946133280589e-06, "loss": 0.5588, "step": 3998 }, { "epoch": 0.69, "learning_rate": 4.558237820303937e-06, "loss": 0.5546, "step": 3999 }, { "epoch": 0.69, "learning_rate": 4.553531220567881e-06, "loss": 0.5704, "step": 4000 }, { "epoch": 0.69, "learning_rate": 4.5488263355542085e-06, "loss": 0.5719, "step": 4001 }, { "epoch": 0.69, "learning_rate": 4.5441231667441724e-06, "loss": 0.5641, "step": 4002 }, { "epoch": 0.69, "learning_rate": 4.539421715618476e-06, "loss": 0.5545, "step": 4003 }, { "epoch": 0.69, "learning_rate": 4.5347219836572855e-06, "loss": 0.5495, "step": 4004 }, { "epoch": 0.69, "learning_rate": 4.530023972340232e-06, "loss": 0.568, "step": 4005 }, { "epoch": 0.69, "learning_rate": 4.525327683146396e-06, "loss": 0.5625, "step": 4006 }, { "epoch": 0.69, "learning_rate": 4.520633117554318e-06, "loss": 0.5511, "step": 4007 }, { "epoch": 0.69, "learning_rate": 4.515940277042002e-06, "loss": 0.5616, "step": 4008 }, { "epoch": 0.69, "learning_rate": 4.511249163086901e-06, "loss": 0.5615, "step": 4009 }, { "epoch": 0.69, "learning_rate": 4.506559777165929e-06, "loss": 0.557, "step": 4010 }, { "epoch": 0.69, "learning_rate": 4.501872120755448e-06, "loss": 0.5618, "step": 4011 }, { "epoch": 0.69, "learning_rate": 4.497186195331296e-06, "loss": 0.5675, "step": 4012 }, { "epoch": 0.7, "learning_rate": 4.492502002368738e-06, "loss": 0.5585, "step": 4013 }, { "epoch": 0.7, "learning_rate": 4.487819543342511e-06, "loss": 0.5547, "step": 4014 }, { "epoch": 0.7, "learning_rate": 4.4831388197268074e-06, "loss": 0.5415, "step": 4015 }, { "epoch": 0.7, "learning_rate": 4.4784598329952675e-06, "loss": 0.5689, "step": 4016 }, { "epoch": 0.7, "learning_rate": 4.473782584620979e-06, "loss": 0.5577, "step": 4017 }, { "epoch": 0.7, "learning_rate": 4.469107076076499e-06, "loss": 0.5562, "step": 4018 }, { "epoch": 0.7, "learning_rate": 4.464433308833821e-06, "loss": 0.5579, "step": 4019 }, { "epoch": 0.7, "learning_rate": 4.459761284364394e-06, "loss": 0.5436, "step": 4020 }, { "epoch": 0.7, "learning_rate": 4.455091004139129e-06, "loss": 0.5582, "step": 4021 }, { "epoch": 0.7, "learning_rate": 4.450422469628374e-06, "loss": 0.5599, "step": 4022 }, { "epoch": 0.7, "learning_rate": 4.445755682301933e-06, "loss": 0.5643, "step": 4023 }, { "epoch": 0.7, "learning_rate": 4.441090643629057e-06, "loss": 0.5724, "step": 4024 }, { "epoch": 0.7, "learning_rate": 4.436427355078455e-06, "loss": 0.5685, "step": 4025 }, { "epoch": 0.7, "learning_rate": 4.431765818118281e-06, "loss": 0.5422, "step": 4026 }, { "epoch": 0.7, "learning_rate": 4.427106034216125e-06, "loss": 0.5577, "step": 4027 }, { "epoch": 0.7, "learning_rate": 4.422448004839044e-06, "loss": 0.5535, "step": 4028 }, { "epoch": 0.7, "learning_rate": 4.417791731453534e-06, "loss": 0.5692, "step": 4029 }, { "epoch": 0.7, "learning_rate": 4.413137215525532e-06, "loss": 0.5605, "step": 4030 }, { "epoch": 0.7, "learning_rate": 4.408484458520438e-06, "loss": 0.5603, "step": 4031 }, { "epoch": 0.7, "learning_rate": 4.403833461903084e-06, "loss": 0.5626, "step": 4032 }, { "epoch": 0.7, "learning_rate": 4.399184227137749e-06, "loss": 0.5517, "step": 4033 }, { "epoch": 0.7, "learning_rate": 4.394536755688169e-06, "loss": 0.554, "step": 4034 }, { "epoch": 0.7, "learning_rate": 4.389891049017511e-06, "loss": 0.5673, "step": 4035 }, { "epoch": 0.7, "learning_rate": 4.385247108588391e-06, "loss": 0.5589, "step": 4036 }, { "epoch": 0.7, "learning_rate": 4.380604935862869e-06, "loss": 0.5504, "step": 4037 }, { "epoch": 0.7, "learning_rate": 4.375964532302456e-06, "loss": 0.5586, "step": 4038 }, { "epoch": 0.7, "learning_rate": 4.3713258993680926e-06, "loss": 0.557, "step": 4039 }, { "epoch": 0.7, "learning_rate": 4.366689038520173e-06, "loss": 0.5655, "step": 4040 }, { "epoch": 0.7, "learning_rate": 4.3620539512185265e-06, "loss": 0.5615, "step": 4041 }, { "epoch": 0.7, "learning_rate": 4.357420638922427e-06, "loss": 0.5616, "step": 4042 }, { "epoch": 0.7, "learning_rate": 4.352789103090587e-06, "loss": 0.5574, "step": 4043 }, { "epoch": 0.7, "learning_rate": 4.348159345181168e-06, "loss": 0.5742, "step": 4044 }, { "epoch": 0.7, "learning_rate": 4.343531366651761e-06, "loss": 0.5434, "step": 4045 }, { "epoch": 0.7, "learning_rate": 4.3389051689594e-06, "loss": 0.5717, "step": 4046 }, { "epoch": 0.7, "learning_rate": 4.334280753560566e-06, "loss": 0.545, "step": 4047 }, { "epoch": 0.7, "learning_rate": 4.329658121911169e-06, "loss": 0.5629, "step": 4048 }, { "epoch": 0.7, "learning_rate": 4.325037275466562e-06, "loss": 0.5716, "step": 4049 }, { "epoch": 0.7, "learning_rate": 4.32041821568153e-06, "loss": 0.5686, "step": 4050 }, { "epoch": 0.7, "learning_rate": 4.315800944010309e-06, "loss": 0.556, "step": 4051 }, { "epoch": 0.7, "learning_rate": 4.31118546190656e-06, "loss": 0.5608, "step": 4052 }, { "epoch": 0.7, "learning_rate": 4.30657177082338e-06, "loss": 0.569, "step": 4053 }, { "epoch": 0.7, "learning_rate": 4.301959872213314e-06, "loss": 0.5664, "step": 4054 }, { "epoch": 0.7, "learning_rate": 4.297349767528336e-06, "loss": 0.5528, "step": 4055 }, { "epoch": 0.7, "learning_rate": 4.292741458219841e-06, "loss": 0.5501, "step": 4056 }, { "epoch": 0.7, "learning_rate": 4.288134945738684e-06, "loss": 0.5665, "step": 4057 }, { "epoch": 0.7, "learning_rate": 4.283530231535141e-06, "loss": 0.5671, "step": 4058 }, { "epoch": 0.7, "learning_rate": 4.278927317058916e-06, "loss": 0.5636, "step": 4059 }, { "epoch": 0.7, "learning_rate": 4.274326203759163e-06, "loss": 0.5867, "step": 4060 }, { "epoch": 0.7, "learning_rate": 4.2697268930844534e-06, "loss": 0.5767, "step": 4061 }, { "epoch": 0.7, "learning_rate": 4.2651293864828e-06, "loss": 0.555, "step": 4062 }, { "epoch": 0.7, "learning_rate": 4.2605336854016395e-06, "loss": 0.5462, "step": 4063 }, { "epoch": 0.7, "learning_rate": 4.255939791287854e-06, "loss": 0.5622, "step": 4064 }, { "epoch": 0.7, "learning_rate": 4.251347705587744e-06, "loss": 0.564, "step": 4065 }, { "epoch": 0.7, "learning_rate": 4.246757429747039e-06, "loss": 0.5709, "step": 4066 }, { "epoch": 0.7, "learning_rate": 4.242168965210915e-06, "loss": 0.552, "step": 4067 }, { "epoch": 0.7, "learning_rate": 4.2375823134239624e-06, "loss": 0.5774, "step": 4068 }, { "epoch": 0.7, "learning_rate": 4.232997475830205e-06, "loss": 0.5618, "step": 4069 }, { "epoch": 0.71, "learning_rate": 4.228414453873097e-06, "loss": 0.5772, "step": 4070 }, { "epoch": 0.71, "learning_rate": 4.223833248995519e-06, "loss": 0.562, "step": 4071 }, { "epoch": 0.71, "learning_rate": 4.2192538626397785e-06, "loss": 0.5711, "step": 4072 }, { "epoch": 0.71, "learning_rate": 4.214676296247619e-06, "loss": 0.5616, "step": 4073 }, { "epoch": 0.71, "learning_rate": 4.2101005512602015e-06, "loss": 0.5639, "step": 4074 }, { "epoch": 0.71, "learning_rate": 4.205526629118116e-06, "loss": 0.5474, "step": 4075 }, { "epoch": 0.71, "learning_rate": 4.200954531261378e-06, "loss": 0.5461, "step": 4076 }, { "epoch": 0.71, "learning_rate": 4.196384259129433e-06, "loss": 0.5716, "step": 4077 }, { "epoch": 0.71, "learning_rate": 4.191815814161149e-06, "loss": 0.5763, "step": 4078 }, { "epoch": 0.71, "learning_rate": 4.1872491977948125e-06, "loss": 0.5802, "step": 4079 }, { "epoch": 0.71, "learning_rate": 4.182684411468148e-06, "loss": 0.5654, "step": 4080 }, { "epoch": 0.71, "learning_rate": 4.178121456618294e-06, "loss": 0.5424, "step": 4081 }, { "epoch": 0.71, "learning_rate": 4.173560334681809e-06, "loss": 0.5623, "step": 4082 }, { "epoch": 0.71, "learning_rate": 4.169001047094691e-06, "loss": 0.5633, "step": 4083 }, { "epoch": 0.71, "learning_rate": 4.164443595292339e-06, "loss": 0.5468, "step": 4084 }, { "epoch": 0.71, "learning_rate": 4.159887980709584e-06, "loss": 0.5543, "step": 4085 }, { "epoch": 0.71, "learning_rate": 4.155334204780688e-06, "loss": 0.5661, "step": 4086 }, { "epoch": 0.71, "learning_rate": 4.150782268939319e-06, "loss": 0.5441, "step": 4087 }, { "epoch": 0.71, "learning_rate": 4.146232174618573e-06, "loss": 0.5498, "step": 4088 }, { "epoch": 0.71, "learning_rate": 4.1416839232509605e-06, "loss": 0.5672, "step": 4089 }, { "epoch": 0.71, "learning_rate": 4.1371375162684255e-06, "loss": 0.554, "step": 4090 }, { "epoch": 0.71, "learning_rate": 4.132592955102318e-06, "loss": 0.5619, "step": 4091 }, { "epoch": 0.71, "learning_rate": 4.128050241183407e-06, "loss": 0.5632, "step": 4092 }, { "epoch": 0.71, "learning_rate": 4.123509375941891e-06, "loss": 0.5554, "step": 4093 }, { "epoch": 0.71, "learning_rate": 4.118970360807375e-06, "loss": 0.5676, "step": 4094 }, { "epoch": 0.71, "learning_rate": 4.114433197208885e-06, "loss": 0.5648, "step": 4095 }, { "epoch": 0.71, "learning_rate": 4.10989788657487e-06, "loss": 0.5604, "step": 4096 }, { "epoch": 0.71, "learning_rate": 4.105364430333191e-06, "loss": 0.5539, "step": 4097 }, { "epoch": 0.71, "learning_rate": 4.1008328299111145e-06, "loss": 0.5543, "step": 4098 }, { "epoch": 0.71, "learning_rate": 4.0963030867353445e-06, "loss": 0.561, "step": 4099 }, { "epoch": 0.71, "learning_rate": 4.0917752022319845e-06, "loss": 0.5523, "step": 4100 }, { "epoch": 0.71, "learning_rate": 4.087249177826553e-06, "loss": 0.5521, "step": 4101 }, { "epoch": 0.71, "learning_rate": 4.0827250149439956e-06, "loss": 0.5639, "step": 4102 }, { "epoch": 0.71, "learning_rate": 4.078202715008659e-06, "loss": 0.5546, "step": 4103 }, { "epoch": 0.71, "learning_rate": 4.073682279444309e-06, "loss": 0.5407, "step": 4104 }, { "epoch": 0.71, "learning_rate": 4.0691637096741166e-06, "loss": 0.568, "step": 4105 }, { "epoch": 0.71, "learning_rate": 4.064647007120681e-06, "loss": 0.5607, "step": 4106 }, { "epoch": 0.71, "learning_rate": 4.060132173206002e-06, "loss": 0.5549, "step": 4107 }, { "epoch": 0.71, "learning_rate": 4.055619209351488e-06, "loss": 0.5637, "step": 4108 }, { "epoch": 0.71, "learning_rate": 4.0511081169779735e-06, "loss": 0.5659, "step": 4109 }, { "epoch": 0.71, "learning_rate": 4.046598897505688e-06, "loss": 0.5731, "step": 4110 }, { "epoch": 0.71, "learning_rate": 4.042091552354279e-06, "loss": 0.5485, "step": 4111 }, { "epoch": 0.71, "learning_rate": 4.037586082942805e-06, "loss": 0.5721, "step": 4112 }, { "epoch": 0.71, "learning_rate": 4.033082490689728e-06, "loss": 0.5651, "step": 4113 }, { "epoch": 0.71, "learning_rate": 4.028580777012922e-06, "loss": 0.5516, "step": 4114 }, { "epoch": 0.71, "learning_rate": 4.024080943329676e-06, "loss": 0.5528, "step": 4115 }, { "epoch": 0.71, "learning_rate": 4.0195829910566795e-06, "loss": 0.5494, "step": 4116 }, { "epoch": 0.71, "learning_rate": 4.01508692161003e-06, "loss": 0.5595, "step": 4117 }, { "epoch": 0.71, "learning_rate": 4.01059273640523e-06, "loss": 0.5715, "step": 4118 }, { "epoch": 0.71, "learning_rate": 4.006100436857201e-06, "loss": 0.5589, "step": 4119 }, { "epoch": 0.71, "learning_rate": 4.001610024380258e-06, "loss": 0.5576, "step": 4120 }, { "epoch": 0.71, "learning_rate": 3.997121500388124e-06, "loss": 0.5687, "step": 4121 }, { "epoch": 0.71, "learning_rate": 3.992634866293935e-06, "loss": 0.5627, "step": 4122 }, { "epoch": 0.71, "learning_rate": 3.988150123510224e-06, "loss": 0.5584, "step": 4123 }, { "epoch": 0.71, "learning_rate": 3.9836672734489315e-06, "loss": 0.5502, "step": 4124 }, { "epoch": 0.71, "learning_rate": 3.9791863175214015e-06, "loss": 0.5611, "step": 4125 }, { "epoch": 0.71, "learning_rate": 3.974707257138383e-06, "loss": 0.5521, "step": 4126 }, { "epoch": 0.71, "learning_rate": 3.970230093710023e-06, "loss": 0.5695, "step": 4127 }, { "epoch": 0.72, "learning_rate": 3.965754828645883e-06, "loss": 0.5544, "step": 4128 }, { "epoch": 0.72, "learning_rate": 3.961281463354916e-06, "loss": 0.5633, "step": 4129 }, { "epoch": 0.72, "learning_rate": 3.956809999245479e-06, "loss": 0.5493, "step": 4130 }, { "epoch": 0.72, "learning_rate": 3.952340437725331e-06, "loss": 0.5656, "step": 4131 }, { "epoch": 0.72, "learning_rate": 3.947872780201637e-06, "loss": 0.5751, "step": 4132 }, { "epoch": 0.72, "learning_rate": 3.943407028080956e-06, "loss": 0.546, "step": 4133 }, { "epoch": 0.72, "learning_rate": 3.9389431827692455e-06, "loss": 0.5601, "step": 4134 }, { "epoch": 0.72, "learning_rate": 3.934481245671875e-06, "loss": 0.5415, "step": 4135 }, { "epoch": 0.72, "learning_rate": 3.930021218193599e-06, "loss": 0.5589, "step": 4136 }, { "epoch": 0.72, "learning_rate": 3.92556310173858e-06, "loss": 0.5691, "step": 4137 }, { "epoch": 0.72, "learning_rate": 3.921106897710368e-06, "loss": 0.542, "step": 4138 }, { "epoch": 0.72, "learning_rate": 3.916652607511931e-06, "loss": 0.5697, "step": 4139 }, { "epoch": 0.72, "learning_rate": 3.91220023254561e-06, "loss": 0.5681, "step": 4140 }, { "epoch": 0.72, "learning_rate": 3.907749774213161e-06, "loss": 0.5584, "step": 4141 }, { "epoch": 0.72, "learning_rate": 3.903301233915731e-06, "loss": 0.5559, "step": 4142 }, { "epoch": 0.72, "learning_rate": 3.898854613053858e-06, "loss": 0.5735, "step": 4143 }, { "epoch": 0.72, "learning_rate": 3.894409913027481e-06, "loss": 0.5454, "step": 4144 }, { "epoch": 0.72, "learning_rate": 3.88996713523594e-06, "loss": 0.5596, "step": 4145 }, { "epoch": 0.72, "learning_rate": 3.8855262810779595e-06, "loss": 0.5666, "step": 4146 }, { "epoch": 0.72, "learning_rate": 3.881087351951657e-06, "loss": 0.5456, "step": 4147 }, { "epoch": 0.72, "learning_rate": 3.876650349254557e-06, "loss": 0.5661, "step": 4148 }, { "epoch": 0.72, "learning_rate": 3.872215274383567e-06, "loss": 0.5505, "step": 4149 }, { "epoch": 0.72, "learning_rate": 3.867782128734992e-06, "loss": 0.553, "step": 4150 }, { "epoch": 0.72, "learning_rate": 3.86335091370452e-06, "loss": 0.5671, "step": 4151 }, { "epoch": 0.72, "learning_rate": 3.858921630687249e-06, "loss": 0.5618, "step": 4152 }, { "epoch": 0.72, "learning_rate": 3.854494281077655e-06, "loss": 0.5564, "step": 4153 }, { "epoch": 0.72, "learning_rate": 3.85006886626961e-06, "loss": 0.5501, "step": 4154 }, { "epoch": 0.72, "learning_rate": 3.845645387656374e-06, "loss": 0.5585, "step": 4155 }, { "epoch": 0.72, "learning_rate": 3.841223846630599e-06, "loss": 0.5537, "step": 4156 }, { "epoch": 0.72, "learning_rate": 3.8368042445843265e-06, "loss": 0.5711, "step": 4157 }, { "epoch": 0.72, "learning_rate": 3.832386582908995e-06, "loss": 0.5646, "step": 4158 }, { "epoch": 0.72, "learning_rate": 3.8279708629954196e-06, "loss": 0.5656, "step": 4159 }, { "epoch": 0.72, "learning_rate": 3.823557086233808e-06, "loss": 0.5635, "step": 4160 }, { "epoch": 0.72, "learning_rate": 3.819145254013766e-06, "loss": 0.5733, "step": 4161 }, { "epoch": 0.72, "learning_rate": 3.8147353677242735e-06, "loss": 0.5523, "step": 4162 }, { "epoch": 0.72, "learning_rate": 3.810327428753706e-06, "loss": 0.563, "step": 4163 }, { "epoch": 0.72, "learning_rate": 3.805921438489819e-06, "loss": 0.537, "step": 4164 }, { "epoch": 0.72, "learning_rate": 3.8015173983197674e-06, "loss": 0.5499, "step": 4165 }, { "epoch": 0.72, "learning_rate": 3.7971153096300773e-06, "loss": 0.5842, "step": 4166 }, { "epoch": 0.72, "learning_rate": 3.7927151738066693e-06, "loss": 0.5594, "step": 4167 }, { "epoch": 0.72, "learning_rate": 3.788316992234845e-06, "loss": 0.5513, "step": 4168 }, { "epoch": 0.72, "learning_rate": 3.783920766299295e-06, "loss": 0.566, "step": 4169 }, { "epoch": 0.72, "learning_rate": 3.779526497384086e-06, "loss": 0.5548, "step": 4170 }, { "epoch": 0.72, "learning_rate": 3.775134186872682e-06, "loss": 0.5734, "step": 4171 }, { "epoch": 0.72, "learning_rate": 3.7707438361479186e-06, "loss": 0.5571, "step": 4172 }, { "epoch": 0.72, "learning_rate": 3.766355446592016e-06, "loss": 0.5607, "step": 4173 }, { "epoch": 0.72, "learning_rate": 3.7619690195865852e-06, "loss": 0.5561, "step": 4174 }, { "epoch": 0.72, "learning_rate": 3.7575845565126113e-06, "loss": 0.5643, "step": 4175 }, { "epoch": 0.72, "learning_rate": 3.7532020587504593e-06, "loss": 0.5608, "step": 4176 }, { "epoch": 0.72, "learning_rate": 3.7488215276798865e-06, "loss": 0.5712, "step": 4177 }, { "epoch": 0.72, "learning_rate": 3.7444429646800206e-06, "loss": 0.5552, "step": 4178 }, { "epoch": 0.72, "learning_rate": 3.740066371129373e-06, "loss": 0.5525, "step": 4179 }, { "epoch": 0.72, "learning_rate": 3.735691748405832e-06, "loss": 0.5724, "step": 4180 }, { "epoch": 0.72, "learning_rate": 3.7313190978866786e-06, "loss": 0.5573, "step": 4181 }, { "epoch": 0.72, "learning_rate": 3.726948420948553e-06, "loss": 0.5714, "step": 4182 }, { "epoch": 0.72, "learning_rate": 3.7225797189674828e-06, "loss": 0.5633, "step": 4183 }, { "epoch": 0.72, "learning_rate": 3.7182129933188836e-06, "loss": 0.5455, "step": 4184 }, { "epoch": 0.72, "learning_rate": 3.713848245377536e-06, "loss": 0.5727, "step": 4185 }, { "epoch": 0.73, "learning_rate": 3.709485476517597e-06, "loss": 0.5481, "step": 4186 }, { "epoch": 0.73, "learning_rate": 3.7051246881126157e-06, "loss": 0.5586, "step": 4187 }, { "epoch": 0.73, "learning_rate": 3.700765881535502e-06, "loss": 0.5504, "step": 4188 }, { "epoch": 0.73, "learning_rate": 3.696409058158544e-06, "loss": 0.5667, "step": 4189 }, { "epoch": 0.73, "learning_rate": 3.692054219353417e-06, "loss": 0.5599, "step": 4190 }, { "epoch": 0.73, "learning_rate": 3.6877013664911588e-06, "loss": 0.5672, "step": 4191 }, { "epoch": 0.73, "learning_rate": 3.6833505009421867e-06, "loss": 0.5516, "step": 4192 }, { "epoch": 0.73, "learning_rate": 3.6790016240762895e-06, "loss": 0.5632, "step": 4193 }, { "epoch": 0.73, "learning_rate": 3.6746547372626384e-06, "loss": 0.5738, "step": 4194 }, { "epoch": 0.73, "learning_rate": 3.6703098418697735e-06, "loss": 0.5736, "step": 4195 }, { "epoch": 0.73, "learning_rate": 3.6659669392655962e-06, "loss": 0.557, "step": 4196 }, { "epoch": 0.73, "learning_rate": 3.6616260308174e-06, "loss": 0.5598, "step": 4197 }, { "epoch": 0.73, "learning_rate": 3.657287117891839e-06, "loss": 0.576, "step": 4198 }, { "epoch": 0.73, "learning_rate": 3.652950201854939e-06, "loss": 0.5706, "step": 4199 }, { "epoch": 0.73, "learning_rate": 3.6486152840721046e-06, "loss": 0.5585, "step": 4200 }, { "epoch": 0.73, "learning_rate": 3.644282365908105e-06, "loss": 0.5619, "step": 4201 }, { "epoch": 0.73, "learning_rate": 3.639951448727076e-06, "loss": 0.5715, "step": 4202 }, { "epoch": 0.73, "learning_rate": 3.6356225338925367e-06, "loss": 0.5604, "step": 4203 }, { "epoch": 0.73, "learning_rate": 3.6312956227673647e-06, "loss": 0.5628, "step": 4204 }, { "epoch": 0.73, "learning_rate": 3.626970716713809e-06, "loss": 0.5713, "step": 4205 }, { "epoch": 0.73, "learning_rate": 3.6226478170934866e-06, "loss": 0.5691, "step": 4206 }, { "epoch": 0.73, "learning_rate": 3.618326925267388e-06, "loss": 0.5399, "step": 4207 }, { "epoch": 0.73, "learning_rate": 3.6140080425958667e-06, "loss": 0.5716, "step": 4208 }, { "epoch": 0.73, "learning_rate": 3.609691170438645e-06, "loss": 0.5518, "step": 4209 }, { "epoch": 0.73, "learning_rate": 3.6053763101548123e-06, "loss": 0.5452, "step": 4210 }, { "epoch": 0.73, "learning_rate": 3.601063463102823e-06, "loss": 0.5512, "step": 4211 }, { "epoch": 0.73, "learning_rate": 3.5967526306404974e-06, "loss": 0.5663, "step": 4212 }, { "epoch": 0.73, "learning_rate": 3.5924438141250297e-06, "loss": 0.5748, "step": 4213 }, { "epoch": 0.73, "learning_rate": 3.588137014912968e-06, "loss": 0.5645, "step": 4214 }, { "epoch": 0.73, "learning_rate": 3.5838322343602293e-06, "loss": 0.5653, "step": 4215 }, { "epoch": 0.73, "learning_rate": 3.579529473822102e-06, "loss": 0.564, "step": 4216 }, { "epoch": 0.73, "learning_rate": 3.5752287346532288e-06, "loss": 0.5521, "step": 4217 }, { "epoch": 0.73, "learning_rate": 3.570930018207619e-06, "loss": 0.5665, "step": 4218 }, { "epoch": 0.73, "learning_rate": 3.5666333258386443e-06, "loss": 0.5596, "step": 4219 }, { "epoch": 0.73, "learning_rate": 3.562338658899047e-06, "loss": 0.5465, "step": 4220 }, { "epoch": 0.73, "learning_rate": 3.5580460187409217e-06, "loss": 0.5608, "step": 4221 }, { "epoch": 0.73, "learning_rate": 3.553755406715724e-06, "loss": 0.5519, "step": 4222 }, { "epoch": 0.73, "learning_rate": 3.549466824174288e-06, "loss": 0.5657, "step": 4223 }, { "epoch": 0.73, "learning_rate": 3.5451802724667863e-06, "loss": 0.5708, "step": 4224 }, { "epoch": 0.73, "learning_rate": 3.54089575294276e-06, "loss": 0.561, "step": 4225 }, { "epoch": 0.73, "learning_rate": 3.536613266951122e-06, "loss": 0.5643, "step": 4226 }, { "epoch": 0.73, "learning_rate": 3.5323328158401315e-06, "loss": 0.5479, "step": 4227 }, { "epoch": 0.73, "learning_rate": 3.5280544009574068e-06, "loss": 0.5557, "step": 4228 }, { "epoch": 0.73, "learning_rate": 3.5237780236499386e-06, "loss": 0.5662, "step": 4229 }, { "epoch": 0.73, "learning_rate": 3.5195036852640617e-06, "loss": 0.5671, "step": 4230 }, { "epoch": 0.73, "learning_rate": 3.5152313871454756e-06, "loss": 0.5597, "step": 4231 }, { "epoch": 0.73, "learning_rate": 3.5109611306392322e-06, "loss": 0.5613, "step": 4232 }, { "epoch": 0.73, "learning_rate": 3.506692917089751e-06, "loss": 0.5583, "step": 4233 }, { "epoch": 0.73, "learning_rate": 3.502426747840799e-06, "loss": 0.5477, "step": 4234 }, { "epoch": 0.73, "learning_rate": 3.4981626242355003e-06, "loss": 0.551, "step": 4235 }, { "epoch": 0.73, "learning_rate": 3.4939005476163436e-06, "loss": 0.5564, "step": 4236 }, { "epoch": 0.73, "learning_rate": 3.489640519325166e-06, "loss": 0.5648, "step": 4237 }, { "epoch": 0.73, "learning_rate": 3.4853825407031503e-06, "loss": 0.5705, "step": 4238 }, { "epoch": 0.73, "learning_rate": 3.481126613090855e-06, "loss": 0.5526, "step": 4239 }, { "epoch": 0.73, "learning_rate": 3.4768727378281774e-06, "loss": 0.5636, "step": 4240 }, { "epoch": 0.73, "learning_rate": 3.472620916254372e-06, "loss": 0.5589, "step": 4241 }, { "epoch": 0.73, "learning_rate": 3.4683711497080543e-06, "loss": 0.5689, "step": 4242 }, { "epoch": 0.73, "learning_rate": 3.4641234395271827e-06, "loss": 0.5493, "step": 4243 }, { "epoch": 0.74, "learning_rate": 3.4598777870490717e-06, "loss": 0.5547, "step": 4244 }, { "epoch": 0.74, "learning_rate": 3.4556341936103853e-06, "loss": 0.5546, "step": 4245 }, { "epoch": 0.74, "learning_rate": 3.4513926605471504e-06, "loss": 0.5487, "step": 4246 }, { "epoch": 0.74, "learning_rate": 3.447153189194731e-06, "loss": 0.5479, "step": 4247 }, { "epoch": 0.74, "learning_rate": 3.442915780887848e-06, "loss": 0.5538, "step": 4248 }, { "epoch": 0.74, "learning_rate": 3.4386804369605764e-06, "loss": 0.5679, "step": 4249 }, { "epoch": 0.74, "learning_rate": 3.434447158746336e-06, "loss": 0.5598, "step": 4250 }, { "epoch": 0.74, "learning_rate": 3.4302159475778985e-06, "loss": 0.5478, "step": 4251 }, { "epoch": 0.74, "learning_rate": 3.4259868047873835e-06, "loss": 0.5855, "step": 4252 }, { "epoch": 0.74, "learning_rate": 3.4217597317062613e-06, "loss": 0.5576, "step": 4253 }, { "epoch": 0.74, "learning_rate": 3.4175347296653448e-06, "loss": 0.5534, "step": 4254 }, { "epoch": 0.74, "learning_rate": 3.4133117999948086e-06, "loss": 0.5662, "step": 4255 }, { "epoch": 0.74, "learning_rate": 3.4090909440241594e-06, "loss": 0.557, "step": 4256 }, { "epoch": 0.74, "learning_rate": 3.40487216308226e-06, "loss": 0.5433, "step": 4257 }, { "epoch": 0.74, "learning_rate": 3.400655458497313e-06, "loss": 0.5556, "step": 4258 }, { "epoch": 0.74, "learning_rate": 3.3964408315968776e-06, "loss": 0.5632, "step": 4259 }, { "epoch": 0.74, "learning_rate": 3.3922282837078524e-06, "loss": 0.564, "step": 4260 }, { "epoch": 0.74, "learning_rate": 3.388017816156476e-06, "loss": 0.5529, "step": 4261 }, { "epoch": 0.74, "learning_rate": 3.3838094302683453e-06, "loss": 0.5698, "step": 4262 }, { "epoch": 0.74, "learning_rate": 3.3796031273683926e-06, "loss": 0.544, "step": 4263 }, { "epoch": 0.74, "learning_rate": 3.3753989087808926e-06, "loss": 0.5546, "step": 4264 }, { "epoch": 0.74, "learning_rate": 3.371196775829477e-06, "loss": 0.5634, "step": 4265 }, { "epoch": 0.74, "learning_rate": 3.366996729837102e-06, "loss": 0.5531, "step": 4266 }, { "epoch": 0.74, "learning_rate": 3.3627987721260758e-06, "loss": 0.5577, "step": 4267 }, { "epoch": 0.74, "learning_rate": 3.358602904018057e-06, "loss": 0.5677, "step": 4268 }, { "epoch": 0.74, "learning_rate": 3.354409126834034e-06, "loss": 0.5432, "step": 4269 }, { "epoch": 0.74, "learning_rate": 3.3502174418943446e-06, "loss": 0.5789, "step": 4270 }, { "epoch": 0.74, "learning_rate": 3.3460278505186593e-06, "loss": 0.5536, "step": 4271 }, { "epoch": 0.74, "learning_rate": 3.3418403540260035e-06, "loss": 0.5717, "step": 4272 }, { "epoch": 0.74, "learning_rate": 3.33765495373473e-06, "loss": 0.5636, "step": 4273 }, { "epoch": 0.74, "learning_rate": 3.3334716509625354e-06, "loss": 0.5678, "step": 4274 }, { "epoch": 0.74, "learning_rate": 3.329290447026462e-06, "loss": 0.5687, "step": 4275 }, { "epoch": 0.74, "learning_rate": 3.325111343242884e-06, "loss": 0.5832, "step": 4276 }, { "epoch": 0.74, "learning_rate": 3.320934340927513e-06, "loss": 0.5491, "step": 4277 }, { "epoch": 0.74, "learning_rate": 3.3167594413954084e-06, "loss": 0.5781, "step": 4278 }, { "epoch": 0.74, "learning_rate": 3.3125866459609644e-06, "loss": 0.5567, "step": 4279 }, { "epoch": 0.74, "learning_rate": 3.308415955937898e-06, "loss": 0.5802, "step": 4280 }, { "epoch": 0.74, "learning_rate": 3.3042473726392875e-06, "loss": 0.5492, "step": 4281 }, { "epoch": 0.74, "learning_rate": 3.300080897377531e-06, "loss": 0.5643, "step": 4282 }, { "epoch": 0.74, "learning_rate": 3.295916531464367e-06, "loss": 0.5541, "step": 4283 }, { "epoch": 0.74, "learning_rate": 3.2917542762108758e-06, "loss": 0.5634, "step": 4284 }, { "epoch": 0.74, "learning_rate": 3.287594132927464e-06, "loss": 0.5486, "step": 4285 }, { "epoch": 0.74, "learning_rate": 3.2834361029238805e-06, "loss": 0.5534, "step": 4286 }, { "epoch": 0.74, "learning_rate": 3.2792801875092005e-06, "loss": 0.5643, "step": 4287 }, { "epoch": 0.74, "learning_rate": 3.275126387991847e-06, "loss": 0.565, "step": 4288 }, { "epoch": 0.74, "learning_rate": 3.270974705679565e-06, "loss": 0.5638, "step": 4289 }, { "epoch": 0.74, "learning_rate": 3.2668251418794318e-06, "loss": 0.5542, "step": 4290 }, { "epoch": 0.74, "learning_rate": 3.2626776978978723e-06, "loss": 0.5658, "step": 4291 }, { "epoch": 0.74, "learning_rate": 3.2585323750406284e-06, "loss": 0.5594, "step": 4292 }, { "epoch": 0.74, "learning_rate": 3.254389174612782e-06, "loss": 0.5486, "step": 4293 }, { "epoch": 0.74, "learning_rate": 3.2502480979187433e-06, "loss": 0.5551, "step": 4294 }, { "epoch": 0.74, "learning_rate": 3.2461091462622576e-06, "loss": 0.5625, "step": 4295 }, { "epoch": 0.74, "learning_rate": 3.2419723209463937e-06, "loss": 0.5592, "step": 4296 }, { "epoch": 0.74, "learning_rate": 3.237837623273564e-06, "loss": 0.5639, "step": 4297 }, { "epoch": 0.74, "learning_rate": 3.2337050545455006e-06, "loss": 0.5619, "step": 4298 }, { "epoch": 0.74, "learning_rate": 3.229574616063268e-06, "loss": 0.553, "step": 4299 }, { "epoch": 0.74, "learning_rate": 3.225446309127256e-06, "loss": 0.5657, "step": 4300 }, { "epoch": 0.75, "learning_rate": 3.2213201350371948e-06, "loss": 0.5617, "step": 4301 }, { "epoch": 0.75, "learning_rate": 3.2171960950921333e-06, "loss": 0.5625, "step": 4302 }, { "epoch": 0.75, "learning_rate": 3.2130741905904463e-06, "loss": 0.5615, "step": 4303 }, { "epoch": 0.75, "learning_rate": 3.2089544228298477e-06, "loss": 0.578, "step": 4304 }, { "epoch": 0.75, "learning_rate": 3.204836793107369e-06, "loss": 0.5688, "step": 4305 }, { "epoch": 0.75, "learning_rate": 3.200721302719373e-06, "loss": 0.5477, "step": 4306 }, { "epoch": 0.75, "learning_rate": 3.1966079529615447e-06, "loss": 0.5483, "step": 4307 }, { "epoch": 0.75, "learning_rate": 3.1924967451289013e-06, "loss": 0.5567, "step": 4308 }, { "epoch": 0.75, "learning_rate": 3.1883876805157766e-06, "loss": 0.5406, "step": 4309 }, { "epoch": 0.75, "learning_rate": 3.184280760415843e-06, "loss": 0.5653, "step": 4310 }, { "epoch": 0.75, "learning_rate": 3.180175986122087e-06, "loss": 0.5672, "step": 4311 }, { "epoch": 0.75, "learning_rate": 3.1760733589268223e-06, "loss": 0.5661, "step": 4312 }, { "epoch": 0.75, "learning_rate": 3.171972880121684e-06, "loss": 0.5673, "step": 4313 }, { "epoch": 0.75, "learning_rate": 3.1678745509976405e-06, "loss": 0.5656, "step": 4314 }, { "epoch": 0.75, "learning_rate": 3.1637783728449734e-06, "loss": 0.5572, "step": 4315 }, { "epoch": 0.75, "learning_rate": 3.159684346953286e-06, "loss": 0.5669, "step": 4316 }, { "epoch": 0.75, "learning_rate": 3.155592474611516e-06, "loss": 0.5777, "step": 4317 }, { "epoch": 0.75, "learning_rate": 3.1515027571079137e-06, "loss": 0.5492, "step": 4318 }, { "epoch": 0.75, "learning_rate": 3.1474151957300512e-06, "loss": 0.5726, "step": 4319 }, { "epoch": 0.75, "learning_rate": 3.1433297917648197e-06, "loss": 0.5385, "step": 4320 }, { "epoch": 0.75, "learning_rate": 3.1392465464984455e-06, "loss": 0.5607, "step": 4321 }, { "epoch": 0.75, "learning_rate": 3.1351654612164517e-06, "loss": 0.5871, "step": 4322 }, { "epoch": 0.75, "learning_rate": 3.131086537203705e-06, "loss": 0.5595, "step": 4323 }, { "epoch": 0.75, "learning_rate": 3.127009775744375e-06, "loss": 0.5597, "step": 4324 }, { "epoch": 0.75, "learning_rate": 3.1229351781219585e-06, "loss": 0.5578, "step": 4325 }, { "epoch": 0.75, "learning_rate": 3.1188627456192655e-06, "loss": 0.5498, "step": 4326 }, { "epoch": 0.75, "learning_rate": 3.1147924795184347e-06, "loss": 0.5691, "step": 4327 }, { "epoch": 0.75, "learning_rate": 3.1107243811009112e-06, "loss": 0.5564, "step": 4328 }, { "epoch": 0.75, "learning_rate": 3.106658451647462e-06, "loss": 0.562, "step": 4329 }, { "epoch": 0.75, "learning_rate": 3.1025946924381743e-06, "loss": 0.5843, "step": 4330 }, { "epoch": 0.75, "learning_rate": 3.0985331047524493e-06, "loss": 0.5592, "step": 4331 }, { "epoch": 0.75, "learning_rate": 3.094473689869002e-06, "loss": 0.546, "step": 4332 }, { "epoch": 0.75, "learning_rate": 3.0904164490658652e-06, "loss": 0.5791, "step": 4333 }, { "epoch": 0.75, "learning_rate": 3.086361383620393e-06, "loss": 0.5553, "step": 4334 }, { "epoch": 0.75, "learning_rate": 3.082308494809246e-06, "loss": 0.5493, "step": 4335 }, { "epoch": 0.75, "learning_rate": 3.078257783908404e-06, "loss": 0.5597, "step": 4336 }, { "epoch": 0.75, "learning_rate": 3.074209252193159e-06, "loss": 0.5589, "step": 4337 }, { "epoch": 0.75, "learning_rate": 3.0701629009381194e-06, "loss": 0.5532, "step": 4338 }, { "epoch": 0.75, "learning_rate": 3.0661187314172023e-06, "loss": 0.5624, "step": 4339 }, { "epoch": 0.75, "learning_rate": 3.062076744903648e-06, "loss": 0.5396, "step": 4340 }, { "epoch": 0.75, "learning_rate": 3.0580369426699986e-06, "loss": 0.5693, "step": 4341 }, { "epoch": 0.75, "learning_rate": 3.0539993259881117e-06, "loss": 0.5557, "step": 4342 }, { "epoch": 0.75, "learning_rate": 3.0499638961291623e-06, "loss": 0.5631, "step": 4343 }, { "epoch": 0.75, "learning_rate": 3.045930654363631e-06, "loss": 0.5706, "step": 4344 }, { "epoch": 0.75, "learning_rate": 3.041899601961308e-06, "loss": 0.5777, "step": 4345 }, { "epoch": 0.75, "learning_rate": 3.037870740191303e-06, "loss": 0.5695, "step": 4346 }, { "epoch": 0.75, "learning_rate": 3.033844070322027e-06, "loss": 0.5535, "step": 4347 }, { "epoch": 0.75, "learning_rate": 3.029819593621206e-06, "loss": 0.5626, "step": 4348 }, { "epoch": 0.75, "learning_rate": 3.0257973113558716e-06, "loss": 0.5786, "step": 4349 }, { "epoch": 0.75, "learning_rate": 3.0217772247923684e-06, "loss": 0.5654, "step": 4350 }, { "epoch": 0.75, "learning_rate": 3.0177593351963474e-06, "loss": 0.5546, "step": 4351 }, { "epoch": 0.75, "learning_rate": 3.013743643832765e-06, "loss": 0.5705, "step": 4352 }, { "epoch": 0.75, "learning_rate": 3.009730151965897e-06, "loss": 0.5408, "step": 4353 }, { "epoch": 0.75, "learning_rate": 3.0057188608593146e-06, "loss": 0.5626, "step": 4354 }, { "epoch": 0.75, "learning_rate": 3.001709771775897e-06, "loss": 0.5608, "step": 4355 }, { "epoch": 0.75, "learning_rate": 2.99770288597784e-06, "loss": 0.5708, "step": 4356 }, { "epoch": 0.75, "learning_rate": 2.993698204726637e-06, "loss": 0.5525, "step": 4357 }, { "epoch": 0.75, "learning_rate": 2.989695729283085e-06, "loss": 0.5424, "step": 4358 }, { "epoch": 0.76, "learning_rate": 2.9856954609072986e-06, "loss": 0.5414, "step": 4359 }, { "epoch": 0.76, "learning_rate": 2.9816974008586873e-06, "loss": 0.5576, "step": 4360 }, { "epoch": 0.76, "learning_rate": 2.9777015503959674e-06, "loss": 0.5772, "step": 4361 }, { "epoch": 0.76, "learning_rate": 2.9737079107771562e-06, "loss": 0.5728, "step": 4362 }, { "epoch": 0.76, "learning_rate": 2.9697164832595892e-06, "loss": 0.5715, "step": 4363 }, { "epoch": 0.76, "learning_rate": 2.965727269099887e-06, "loss": 0.5547, "step": 4364 }, { "epoch": 0.76, "learning_rate": 2.961740269553981e-06, "loss": 0.5601, "step": 4365 }, { "epoch": 0.76, "learning_rate": 2.957755485877112e-06, "loss": 0.5449, "step": 4366 }, { "epoch": 0.76, "learning_rate": 2.953772919323814e-06, "loss": 0.5689, "step": 4367 }, { "epoch": 0.76, "learning_rate": 2.9497925711479237e-06, "loss": 0.5536, "step": 4368 }, { "epoch": 0.76, "learning_rate": 2.945814442602587e-06, "loss": 0.5678, "step": 4369 }, { "epoch": 0.76, "learning_rate": 2.9418385349402444e-06, "loss": 0.5455, "step": 4370 }, { "epoch": 0.76, "learning_rate": 2.937864849412634e-06, "loss": 0.5524, "step": 4371 }, { "epoch": 0.76, "learning_rate": 2.9338933872708064e-06, "loss": 0.5415, "step": 4372 }, { "epoch": 0.76, "learning_rate": 2.9299241497651e-06, "loss": 0.5515, "step": 4373 }, { "epoch": 0.76, "learning_rate": 2.925957138145159e-06, "loss": 0.5618, "step": 4374 }, { "epoch": 0.76, "learning_rate": 2.9219923536599228e-06, "loss": 0.5582, "step": 4375 }, { "epoch": 0.76, "learning_rate": 2.9180297975576368e-06, "loss": 0.5583, "step": 4376 }, { "epoch": 0.76, "learning_rate": 2.9140694710858376e-06, "loss": 0.5631, "step": 4377 }, { "epoch": 0.76, "learning_rate": 2.9101113754913636e-06, "loss": 0.5541, "step": 4378 }, { "epoch": 0.76, "learning_rate": 2.906155512020349e-06, "loss": 0.5535, "step": 4379 }, { "epoch": 0.76, "learning_rate": 2.902201881918225e-06, "loss": 0.5599, "step": 4380 }, { "epoch": 0.76, "learning_rate": 2.898250486429719e-06, "loss": 0.5705, "step": 4381 }, { "epoch": 0.76, "learning_rate": 2.894301326798863e-06, "loss": 0.5663, "step": 4382 }, { "epoch": 0.76, "learning_rate": 2.8903544042689745e-06, "loss": 0.5752, "step": 4383 }, { "epoch": 0.76, "learning_rate": 2.886409720082668e-06, "loss": 0.5533, "step": 4384 }, { "epoch": 0.76, "learning_rate": 2.8824672754818617e-06, "loss": 0.5565, "step": 4385 }, { "epoch": 0.76, "learning_rate": 2.8785270717077618e-06, "loss": 0.5603, "step": 4386 }, { "epoch": 0.76, "learning_rate": 2.8745891100008683e-06, "loss": 0.5682, "step": 4387 }, { "epoch": 0.76, "learning_rate": 2.870653391600976e-06, "loss": 0.5555, "step": 4388 }, { "epoch": 0.76, "learning_rate": 2.8667199177471794e-06, "loss": 0.5639, "step": 4389 }, { "epoch": 0.76, "learning_rate": 2.86278868967786e-06, "loss": 0.5612, "step": 4390 }, { "epoch": 0.76, "learning_rate": 2.8588597086306933e-06, "loss": 0.5541, "step": 4391 }, { "epoch": 0.76, "learning_rate": 2.854932975842648e-06, "loss": 0.5561, "step": 4392 }, { "epoch": 0.76, "learning_rate": 2.851008492549986e-06, "loss": 0.5763, "step": 4393 }, { "epoch": 0.76, "learning_rate": 2.8470862599882554e-06, "loss": 0.5361, "step": 4394 }, { "epoch": 0.76, "learning_rate": 2.8431662793923075e-06, "loss": 0.5726, "step": 4395 }, { "epoch": 0.76, "learning_rate": 2.839248551996274e-06, "loss": 0.5494, "step": 4396 }, { "epoch": 0.76, "learning_rate": 2.8353330790335777e-06, "loss": 0.568, "step": 4397 }, { "epoch": 0.76, "learning_rate": 2.8314198617369403e-06, "loss": 0.5531, "step": 4398 }, { "epoch": 0.76, "learning_rate": 2.827508901338366e-06, "loss": 0.5445, "step": 4399 }, { "epoch": 0.76, "learning_rate": 2.8236001990691487e-06, "loss": 0.5733, "step": 4400 }, { "epoch": 0.76, "learning_rate": 2.8196937561598704e-06, "loss": 0.5798, "step": 4401 }, { "epoch": 0.76, "learning_rate": 2.8157895738404095e-06, "loss": 0.5784, "step": 4402 }, { "epoch": 0.76, "learning_rate": 2.811887653339924e-06, "loss": 0.5318, "step": 4403 }, { "epoch": 0.76, "learning_rate": 2.8079879958868615e-06, "loss": 0.5623, "step": 4404 }, { "epoch": 0.76, "learning_rate": 2.8040906027089674e-06, "loss": 0.5679, "step": 4405 }, { "epoch": 0.76, "learning_rate": 2.800195475033256e-06, "loss": 0.5638, "step": 4406 }, { "epoch": 0.76, "learning_rate": 2.7963026140860384e-06, "loss": 0.5585, "step": 4407 }, { "epoch": 0.76, "learning_rate": 2.7924120210929174e-06, "loss": 0.5453, "step": 4408 }, { "epoch": 0.76, "learning_rate": 2.7885236972787733e-06, "loss": 0.5589, "step": 4409 }, { "epoch": 0.76, "learning_rate": 2.7846376438677715e-06, "loss": 0.553, "step": 4410 }, { "epoch": 0.76, "learning_rate": 2.7807538620833707e-06, "loss": 0.5477, "step": 4411 }, { "epoch": 0.76, "learning_rate": 2.776872353148308e-06, "loss": 0.5777, "step": 4412 }, { "epoch": 0.76, "learning_rate": 2.772993118284606e-06, "loss": 0.5691, "step": 4413 }, { "epoch": 0.76, "learning_rate": 2.7691161587135683e-06, "loss": 0.5515, "step": 4414 }, { "epoch": 0.76, "learning_rate": 2.765241475655792e-06, "loss": 0.582, "step": 4415 }, { "epoch": 0.76, "learning_rate": 2.761369070331149e-06, "loss": 0.5708, "step": 4416 }, { "epoch": 0.77, "learning_rate": 2.7574989439587917e-06, "loss": 0.554, "step": 4417 }, { "epoch": 0.77, "learning_rate": 2.7536310977571655e-06, "loss": 0.5737, "step": 4418 }, { "epoch": 0.77, "learning_rate": 2.749765532943993e-06, "loss": 0.5528, "step": 4419 }, { "epoch": 0.77, "learning_rate": 2.7459022507362687e-06, "loss": 0.5569, "step": 4420 }, { "epoch": 0.77, "learning_rate": 2.742041252350286e-06, "loss": 0.5778, "step": 4421 }, { "epoch": 0.77, "learning_rate": 2.738182539001607e-06, "loss": 0.5558, "step": 4422 }, { "epoch": 0.77, "learning_rate": 2.7343261119050744e-06, "loss": 0.5707, "step": 4423 }, { "epoch": 0.77, "learning_rate": 2.730471972274822e-06, "loss": 0.5629, "step": 4424 }, { "epoch": 0.77, "learning_rate": 2.7266201213242526e-06, "loss": 0.5531, "step": 4425 }, { "epoch": 0.77, "learning_rate": 2.7227705602660513e-06, "loss": 0.5859, "step": 4426 }, { "epoch": 0.77, "learning_rate": 2.718923290312182e-06, "loss": 0.5802, "step": 4427 }, { "epoch": 0.77, "learning_rate": 2.715078312673891e-06, "loss": 0.5725, "step": 4428 }, { "epoch": 0.77, "learning_rate": 2.7112356285617e-06, "loss": 0.54, "step": 4429 }, { "epoch": 0.77, "learning_rate": 2.707395239185404e-06, "loss": 0.5435, "step": 4430 }, { "epoch": 0.77, "learning_rate": 2.7035571457540865e-06, "loss": 0.5434, "step": 4431 }, { "epoch": 0.77, "learning_rate": 2.6997213494761e-06, "loss": 0.555, "step": 4432 }, { "epoch": 0.77, "learning_rate": 2.6958878515590747e-06, "loss": 0.5693, "step": 4433 }, { "epoch": 0.77, "learning_rate": 2.692056653209919e-06, "loss": 0.549, "step": 4434 }, { "epoch": 0.77, "learning_rate": 2.6882277556348156e-06, "loss": 0.5552, "step": 4435 }, { "epoch": 0.77, "learning_rate": 2.6844011600392215e-06, "loss": 0.5485, "step": 4436 }, { "epoch": 0.77, "learning_rate": 2.6805768676278766e-06, "loss": 0.5432, "step": 4437 }, { "epoch": 0.77, "learning_rate": 2.676754879604788e-06, "loss": 0.5689, "step": 4438 }, { "epoch": 0.77, "learning_rate": 2.6729351971732398e-06, "loss": 0.5616, "step": 4439 }, { "epoch": 0.77, "learning_rate": 2.669117821535786e-06, "loss": 0.5659, "step": 4440 }, { "epoch": 0.77, "learning_rate": 2.6653027538942655e-06, "loss": 0.5594, "step": 4441 }, { "epoch": 0.77, "learning_rate": 2.6614899954497797e-06, "loss": 0.5643, "step": 4442 }, { "epoch": 0.77, "learning_rate": 2.657679547402704e-06, "loss": 0.5566, "step": 4443 }, { "epoch": 0.77, "learning_rate": 2.653871410952695e-06, "loss": 0.57, "step": 4444 }, { "epoch": 0.77, "learning_rate": 2.650065587298675e-06, "loss": 0.5659, "step": 4445 }, { "epoch": 0.77, "learning_rate": 2.6462620776388313e-06, "loss": 0.5978, "step": 4446 }, { "epoch": 0.77, "learning_rate": 2.6424608831706435e-06, "loss": 0.5788, "step": 4447 }, { "epoch": 0.77, "learning_rate": 2.6386620050908383e-06, "loss": 0.5429, "step": 4448 }, { "epoch": 0.77, "learning_rate": 2.6348654445954235e-06, "loss": 0.5571, "step": 4449 }, { "epoch": 0.77, "learning_rate": 2.631071202879685e-06, "loss": 0.5673, "step": 4450 }, { "epoch": 0.77, "learning_rate": 2.6272792811381663e-06, "loss": 0.5597, "step": 4451 }, { "epoch": 0.77, "learning_rate": 2.623489680564685e-06, "loss": 0.5585, "step": 4452 }, { "epoch": 0.77, "learning_rate": 2.619702402352332e-06, "loss": 0.5505, "step": 4453 }, { "epoch": 0.77, "learning_rate": 2.615917447693462e-06, "loss": 0.5631, "step": 4454 }, { "epoch": 0.77, "learning_rate": 2.6121348177797e-06, "loss": 0.561, "step": 4455 }, { "epoch": 0.77, "learning_rate": 2.608354513801934e-06, "loss": 0.5722, "step": 4456 }, { "epoch": 0.77, "learning_rate": 2.6045765369503316e-06, "loss": 0.5747, "step": 4457 }, { "epoch": 0.77, "learning_rate": 2.600800888414319e-06, "loss": 0.5666, "step": 4458 }, { "epoch": 0.77, "learning_rate": 2.597027569382585e-06, "loss": 0.5644, "step": 4459 }, { "epoch": 0.77, "learning_rate": 2.593256581043099e-06, "loss": 0.5899, "step": 4460 }, { "epoch": 0.77, "learning_rate": 2.5894879245830883e-06, "loss": 0.5792, "step": 4461 }, { "epoch": 0.77, "learning_rate": 2.585721601189036e-06, "loss": 0.5668, "step": 4462 }, { "epoch": 0.77, "learning_rate": 2.5819576120467126e-06, "loss": 0.5402, "step": 4463 }, { "epoch": 0.77, "learning_rate": 2.5781959583411375e-06, "loss": 0.5548, "step": 4464 }, { "epoch": 0.77, "learning_rate": 2.574436641256597e-06, "loss": 0.5569, "step": 4465 }, { "epoch": 0.77, "learning_rate": 2.5706796619766493e-06, "loss": 0.5841, "step": 4466 }, { "epoch": 0.77, "learning_rate": 2.5669250216841104e-06, "loss": 0.5601, "step": 4467 }, { "epoch": 0.77, "learning_rate": 2.563172721561058e-06, "loss": 0.5733, "step": 4468 }, { "epoch": 0.77, "learning_rate": 2.5594227627888356e-06, "loss": 0.5472, "step": 4469 }, { "epoch": 0.77, "learning_rate": 2.5556751465480555e-06, "loss": 0.5723, "step": 4470 }, { "epoch": 0.77, "learning_rate": 2.5519298740185837e-06, "loss": 0.5539, "step": 4471 }, { "epoch": 0.77, "learning_rate": 2.5481869463795494e-06, "loss": 0.5561, "step": 4472 }, { "epoch": 0.77, "learning_rate": 2.54444636480935e-06, "loss": 0.576, "step": 4473 }, { "epoch": 0.77, "learning_rate": 2.5407081304856384e-06, "loss": 0.5805, "step": 4474 }, { "epoch": 0.78, "learning_rate": 2.536972244585331e-06, "loss": 0.5541, "step": 4475 }, { "epoch": 0.78, "learning_rate": 2.533238708284602e-06, "loss": 0.5613, "step": 4476 }, { "epoch": 0.78, "learning_rate": 2.5295075227588908e-06, "loss": 0.5634, "step": 4477 }, { "epoch": 0.78, "learning_rate": 2.5257786891828872e-06, "loss": 0.5514, "step": 4478 }, { "epoch": 0.78, "learning_rate": 2.5220522087305556e-06, "loss": 0.5537, "step": 4479 }, { "epoch": 0.78, "learning_rate": 2.518328082575108e-06, "loss": 0.5663, "step": 4480 }, { "epoch": 0.78, "learning_rate": 2.5146063118890176e-06, "loss": 0.5765, "step": 4481 }, { "epoch": 0.78, "learning_rate": 2.510886897844014e-06, "loss": 0.5559, "step": 4482 }, { "epoch": 0.78, "learning_rate": 2.5071698416110924e-06, "loss": 0.5552, "step": 4483 }, { "epoch": 0.78, "learning_rate": 2.5034551443604995e-06, "loss": 0.5618, "step": 4484 }, { "epoch": 0.78, "learning_rate": 2.499742807261738e-06, "loss": 0.5398, "step": 4485 }, { "epoch": 0.78, "learning_rate": 2.4960328314835746e-06, "loss": 0.557, "step": 4486 }, { "epoch": 0.78, "learning_rate": 2.492325218194026e-06, "loss": 0.5445, "step": 4487 }, { "epoch": 0.78, "learning_rate": 2.4886199685603676e-06, "loss": 0.5512, "step": 4488 }, { "epoch": 0.78, "learning_rate": 2.4849170837491265e-06, "loss": 0.5647, "step": 4489 }, { "epoch": 0.78, "learning_rate": 2.481216564926099e-06, "loss": 0.5651, "step": 4490 }, { "epoch": 0.78, "learning_rate": 2.4775184132563145e-06, "loss": 0.5622, "step": 4491 }, { "epoch": 0.78, "learning_rate": 2.473822629904078e-06, "loss": 0.547, "step": 4492 }, { "epoch": 0.78, "learning_rate": 2.4701292160329373e-06, "loss": 0.5504, "step": 4493 }, { "epoch": 0.78, "learning_rate": 2.4664381728056962e-06, "loss": 0.5616, "step": 4494 }, { "epoch": 0.78, "learning_rate": 2.462749501384413e-06, "loss": 0.5531, "step": 4495 }, { "epoch": 0.78, "learning_rate": 2.4590632029304018e-06, "loss": 0.5678, "step": 4496 }, { "epoch": 0.78, "learning_rate": 2.455379278604226e-06, "loss": 0.5442, "step": 4497 }, { "epoch": 0.78, "learning_rate": 2.4516977295656997e-06, "loss": 0.5426, "step": 4498 }, { "epoch": 0.78, "learning_rate": 2.448018556973897e-06, "loss": 0.5471, "step": 4499 }, { "epoch": 0.78, "learning_rate": 2.4443417619871367e-06, "loss": 0.5637, "step": 4500 }, { "epoch": 0.78, "learning_rate": 2.4406673457629914e-06, "loss": 0.5559, "step": 4501 }, { "epoch": 0.78, "learning_rate": 2.436995309458282e-06, "loss": 0.5532, "step": 4502 }, { "epoch": 0.78, "learning_rate": 2.4333256542290883e-06, "loss": 0.5554, "step": 4503 }, { "epoch": 0.78, "learning_rate": 2.429658381230733e-06, "loss": 0.5714, "step": 4504 }, { "epoch": 0.78, "learning_rate": 2.4259934916177898e-06, "loss": 0.574, "step": 4505 }, { "epoch": 0.78, "learning_rate": 2.4223309865440823e-06, "loss": 0.5451, "step": 4506 }, { "epoch": 0.78, "learning_rate": 2.418670867162686e-06, "loss": 0.561, "step": 4507 }, { "epoch": 0.78, "learning_rate": 2.41501313462592e-06, "loss": 0.5481, "step": 4508 }, { "epoch": 0.78, "learning_rate": 2.411357790085359e-06, "loss": 0.5662, "step": 4509 }, { "epoch": 0.78, "learning_rate": 2.407704834691823e-06, "loss": 0.5551, "step": 4510 }, { "epoch": 0.78, "learning_rate": 2.404054269595374e-06, "loss": 0.5661, "step": 4511 }, { "epoch": 0.78, "learning_rate": 2.4004060959453312e-06, "loss": 0.5489, "step": 4512 }, { "epoch": 0.78, "learning_rate": 2.396760314890256e-06, "loss": 0.5657, "step": 4513 }, { "epoch": 0.78, "learning_rate": 2.393116927577953e-06, "loss": 0.5548, "step": 4514 }, { "epoch": 0.78, "learning_rate": 2.389475935155482e-06, "loss": 0.5784, "step": 4515 }, { "epoch": 0.78, "learning_rate": 2.3858373387691404e-06, "loss": 0.5679, "step": 4516 }, { "epoch": 0.78, "learning_rate": 2.382201139564476e-06, "loss": 0.5429, "step": 4517 }, { "epoch": 0.78, "learning_rate": 2.3785673386862808e-06, "loss": 0.5648, "step": 4518 }, { "epoch": 0.78, "learning_rate": 2.3749359372785884e-06, "loss": 0.5385, "step": 4519 }, { "epoch": 0.78, "learning_rate": 2.371306936484684e-06, "loss": 0.56, "step": 4520 }, { "epoch": 0.78, "learning_rate": 2.367680337447087e-06, "loss": 0.5543, "step": 4521 }, { "epoch": 0.78, "learning_rate": 2.3640561413075746e-06, "loss": 0.575, "step": 4522 }, { "epoch": 0.78, "learning_rate": 2.3604343492071545e-06, "loss": 0.5494, "step": 4523 }, { "epoch": 0.78, "learning_rate": 2.3568149622860815e-06, "loss": 0.5599, "step": 4524 }, { "epoch": 0.78, "learning_rate": 2.353197981683859e-06, "loss": 0.5445, "step": 4525 }, { "epoch": 0.78, "learning_rate": 2.3495834085392265e-06, "loss": 0.5375, "step": 4526 }, { "epoch": 0.78, "learning_rate": 2.345971243990163e-06, "loss": 0.5557, "step": 4527 }, { "epoch": 0.78, "learning_rate": 2.3423614891738977e-06, "loss": 0.5364, "step": 4528 }, { "epoch": 0.78, "learning_rate": 2.3387541452268968e-06, "loss": 0.5601, "step": 4529 }, { "epoch": 0.78, "learning_rate": 2.3351492132848665e-06, "loss": 0.5543, "step": 4530 }, { "epoch": 0.78, "learning_rate": 2.33154669448275e-06, "loss": 0.5649, "step": 4531 }, { "epoch": 0.79, "learning_rate": 2.3279465899547473e-06, "loss": 0.5577, "step": 4532 }, { "epoch": 0.79, "learning_rate": 2.3243489008342735e-06, "loss": 0.5573, "step": 4533 }, { "epoch": 0.79, "learning_rate": 2.3207536282539987e-06, "loss": 0.5533, "step": 4534 }, { "epoch": 0.79, "learning_rate": 2.317160773345836e-06, "loss": 0.5586, "step": 4535 }, { "epoch": 0.79, "learning_rate": 2.313570337240926e-06, "loss": 0.5484, "step": 4536 }, { "epoch": 0.79, "learning_rate": 2.3099823210696503e-06, "loss": 0.5643, "step": 4537 }, { "epoch": 0.79, "learning_rate": 2.306396725961638e-06, "loss": 0.5507, "step": 4538 }, { "epoch": 0.79, "learning_rate": 2.302813553045745e-06, "loss": 0.5542, "step": 4539 }, { "epoch": 0.79, "learning_rate": 2.2992328034500668e-06, "loss": 0.561, "step": 4540 }, { "epoch": 0.79, "learning_rate": 2.295654478301942e-06, "loss": 0.5522, "step": 4541 }, { "epoch": 0.79, "learning_rate": 2.292078578727939e-06, "loss": 0.5725, "step": 4542 }, { "epoch": 0.79, "learning_rate": 2.2885051058538664e-06, "loss": 0.5519, "step": 4543 }, { "epoch": 0.79, "learning_rate": 2.284934060804764e-06, "loss": 0.5573, "step": 4544 }, { "epoch": 0.79, "learning_rate": 2.2813654447049172e-06, "loss": 0.5473, "step": 4545 }, { "epoch": 0.79, "learning_rate": 2.2777992586778374e-06, "loss": 0.5545, "step": 4546 }, { "epoch": 0.79, "learning_rate": 2.2742355038462726e-06, "loss": 0.5796, "step": 4547 }, { "epoch": 0.79, "learning_rate": 2.270674181332209e-06, "loss": 0.5505, "step": 4548 }, { "epoch": 0.79, "learning_rate": 2.2671152922568617e-06, "loss": 0.5648, "step": 4549 }, { "epoch": 0.79, "learning_rate": 2.2635588377406823e-06, "loss": 0.5617, "step": 4550 }, { "epoch": 0.79, "learning_rate": 2.2600048189033618e-06, "loss": 0.5762, "step": 4551 }, { "epoch": 0.79, "learning_rate": 2.256453236863815e-06, "loss": 0.5548, "step": 4552 }, { "epoch": 0.79, "learning_rate": 2.2529040927401913e-06, "loss": 0.5569, "step": 4553 }, { "epoch": 0.79, "learning_rate": 2.24935738764988e-06, "loss": 0.5404, "step": 4554 }, { "epoch": 0.79, "learning_rate": 2.2458131227094936e-06, "loss": 0.5439, "step": 4555 }, { "epoch": 0.79, "learning_rate": 2.2422712990348815e-06, "loss": 0.5646, "step": 4556 }, { "epoch": 0.79, "learning_rate": 2.2387319177411194e-06, "loss": 0.5768, "step": 4557 }, { "epoch": 0.79, "learning_rate": 2.235194979942523e-06, "loss": 0.5771, "step": 4558 }, { "epoch": 0.79, "learning_rate": 2.2316604867526313e-06, "loss": 0.5452, "step": 4559 }, { "epoch": 0.79, "learning_rate": 2.228128439284214e-06, "loss": 0.556, "step": 4560 }, { "epoch": 0.79, "learning_rate": 2.2245988386492745e-06, "loss": 0.5798, "step": 4561 }, { "epoch": 0.79, "learning_rate": 2.221071685959043e-06, "loss": 0.5525, "step": 4562 }, { "epoch": 0.79, "learning_rate": 2.217546982323977e-06, "loss": 0.5578, "step": 4563 }, { "epoch": 0.79, "learning_rate": 2.214024728853771e-06, "loss": 0.5657, "step": 4564 }, { "epoch": 0.79, "learning_rate": 2.210504926657341e-06, "loss": 0.5748, "step": 4565 }, { "epoch": 0.79, "learning_rate": 2.2069875768428296e-06, "loss": 0.5555, "step": 4566 }, { "epoch": 0.79, "learning_rate": 2.2034726805176165e-06, "loss": 0.5604, "step": 4567 }, { "epoch": 0.79, "learning_rate": 2.199960238788301e-06, "loss": 0.5503, "step": 4568 }, { "epoch": 0.79, "learning_rate": 2.1964502527607135e-06, "loss": 0.5381, "step": 4569 }, { "epoch": 0.79, "learning_rate": 2.1929427235399037e-06, "loss": 0.5403, "step": 4570 }, { "epoch": 0.79, "learning_rate": 2.189437652230162e-06, "loss": 0.5737, "step": 4571 }, { "epoch": 0.79, "learning_rate": 2.185935039934992e-06, "loss": 0.5551, "step": 4572 }, { "epoch": 0.79, "learning_rate": 2.182434887757127e-06, "loss": 0.5561, "step": 4573 }, { "epoch": 0.79, "learning_rate": 2.178937196798534e-06, "loss": 0.5654, "step": 4574 }, { "epoch": 0.79, "learning_rate": 2.175441968160391e-06, "loss": 0.5605, "step": 4575 }, { "epoch": 0.79, "learning_rate": 2.1719492029431054e-06, "loss": 0.5475, "step": 4576 }, { "epoch": 0.79, "learning_rate": 2.1684589022463187e-06, "loss": 0.5702, "step": 4577 }, { "epoch": 0.79, "learning_rate": 2.164971067168885e-06, "loss": 0.5503, "step": 4578 }, { "epoch": 0.79, "learning_rate": 2.161485698808885e-06, "loss": 0.5761, "step": 4579 }, { "epoch": 0.79, "learning_rate": 2.1580027982636277e-06, "loss": 0.5583, "step": 4580 }, { "epoch": 0.79, "learning_rate": 2.15452236662964e-06, "loss": 0.5547, "step": 4581 }, { "epoch": 0.79, "learning_rate": 2.1510444050026722e-06, "loss": 0.5551, "step": 4582 }, { "epoch": 0.79, "learning_rate": 2.1475689144776958e-06, "loss": 0.5656, "step": 4583 }, { "epoch": 0.79, "learning_rate": 2.1440958961489112e-06, "loss": 0.5574, "step": 4584 }, { "epoch": 0.79, "learning_rate": 2.140625351109733e-06, "loss": 0.5629, "step": 4585 }, { "epoch": 0.79, "learning_rate": 2.1371572804527972e-06, "loss": 0.5536, "step": 4586 }, { "epoch": 0.79, "learning_rate": 2.1336916852699674e-06, "loss": 0.5667, "step": 4587 }, { "epoch": 0.79, "learning_rate": 2.130228566652326e-06, "loss": 0.5344, "step": 4588 }, { "epoch": 0.79, "learning_rate": 2.126767925690163e-06, "loss": 0.5539, "step": 4589 }, { "epoch": 0.8, "learning_rate": 2.123309763473007e-06, "loss": 0.5608, "step": 4590 }, { "epoch": 0.8, "learning_rate": 2.1198540810895975e-06, "loss": 0.5556, "step": 4591 }, { "epoch": 0.8, "learning_rate": 2.116400879627888e-06, "loss": 0.5529, "step": 4592 }, { "epoch": 0.8, "learning_rate": 2.1129501601750647e-06, "loss": 0.551, "step": 4593 }, { "epoch": 0.8, "learning_rate": 2.1095019238175195e-06, "loss": 0.5694, "step": 4594 }, { "epoch": 0.8, "learning_rate": 2.10605617164087e-06, "loss": 0.5525, "step": 4595 }, { "epoch": 0.8, "learning_rate": 2.1026129047299436e-06, "loss": 0.548, "step": 4596 }, { "epoch": 0.8, "learning_rate": 2.099172124168797e-06, "loss": 0.5644, "step": 4597 }, { "epoch": 0.8, "learning_rate": 2.0957338310406962e-06, "loss": 0.5429, "step": 4598 }, { "epoch": 0.8, "learning_rate": 2.0922980264281223e-06, "loss": 0.5688, "step": 4599 }, { "epoch": 0.8, "learning_rate": 2.088864711412781e-06, "loss": 0.5646, "step": 4600 }, { "epoch": 0.8, "learning_rate": 2.0854338870755873e-06, "loss": 0.5672, "step": 4601 }, { "epoch": 0.8, "learning_rate": 2.0820055544966757e-06, "loss": 0.5729, "step": 4602 }, { "epoch": 0.8, "learning_rate": 2.078579714755393e-06, "loss": 0.5685, "step": 4603 }, { "epoch": 0.8, "learning_rate": 2.0751563689303045e-06, "loss": 0.543, "step": 4604 }, { "epoch": 0.8, "learning_rate": 2.071735518099185e-06, "loss": 0.5718, "step": 4605 }, { "epoch": 0.8, "learning_rate": 2.0683171633390333e-06, "loss": 0.5536, "step": 4606 }, { "epoch": 0.8, "learning_rate": 2.064901305726055e-06, "loss": 0.5616, "step": 4607 }, { "epoch": 0.8, "learning_rate": 2.0614879463356683e-06, "loss": 0.5644, "step": 4608 }, { "epoch": 0.8, "learning_rate": 2.0580770862425093e-06, "loss": 0.5703, "step": 4609 }, { "epoch": 0.8, "learning_rate": 2.054668726520428e-06, "loss": 0.5449, "step": 4610 }, { "epoch": 0.8, "learning_rate": 2.051262868242483e-06, "loss": 0.5591, "step": 4611 }, { "epoch": 0.8, "learning_rate": 2.0478595124809453e-06, "loss": 0.5532, "step": 4612 }, { "epoch": 0.8, "learning_rate": 2.0444586603073036e-06, "loss": 0.5564, "step": 4613 }, { "epoch": 0.8, "learning_rate": 2.0410603127922548e-06, "loss": 0.5528, "step": 4614 }, { "epoch": 0.8, "learning_rate": 2.0376644710057026e-06, "loss": 0.5501, "step": 4615 }, { "epoch": 0.8, "learning_rate": 2.0342711360167755e-06, "loss": 0.555, "step": 4616 }, { "epoch": 0.8, "learning_rate": 2.0308803088937956e-06, "loss": 0.5532, "step": 4617 }, { "epoch": 0.8, "learning_rate": 2.0274919907043033e-06, "loss": 0.5386, "step": 4618 }, { "epoch": 0.8, "learning_rate": 2.0241061825150545e-06, "loss": 0.5628, "step": 4619 }, { "epoch": 0.8, "learning_rate": 2.020722885392008e-06, "loss": 0.5642, "step": 4620 }, { "epoch": 0.8, "learning_rate": 2.01734210040033e-06, "loss": 0.5612, "step": 4621 }, { "epoch": 0.8, "learning_rate": 2.013963828604406e-06, "loss": 0.5647, "step": 4622 }, { "epoch": 0.8, "learning_rate": 2.010588071067822e-06, "loss": 0.5698, "step": 4623 }, { "epoch": 0.8, "learning_rate": 2.007214828853372e-06, "loss": 0.54, "step": 4624 }, { "epoch": 0.8, "learning_rate": 2.0038441030230594e-06, "loss": 0.5508, "step": 4625 }, { "epoch": 0.8, "learning_rate": 2.000475894638101e-06, "loss": 0.5446, "step": 4626 }, { "epoch": 0.8, "learning_rate": 1.9971102047589132e-06, "loss": 0.5706, "step": 4627 }, { "epoch": 0.8, "learning_rate": 1.9937470344451215e-06, "loss": 0.5513, "step": 4628 }, { "epoch": 0.8, "learning_rate": 1.990386384755565e-06, "loss": 0.5591, "step": 4629 }, { "epoch": 0.8, "learning_rate": 1.9870282567482814e-06, "loss": 0.5469, "step": 4630 }, { "epoch": 0.8, "learning_rate": 1.9836726514805105e-06, "loss": 0.5624, "step": 4631 }, { "epoch": 0.8, "learning_rate": 1.9803195700087117e-06, "loss": 0.5501, "step": 4632 }, { "epoch": 0.8, "learning_rate": 1.9769690133885387e-06, "loss": 0.5524, "step": 4633 }, { "epoch": 0.8, "learning_rate": 1.973620982674852e-06, "loss": 0.5523, "step": 4634 }, { "epoch": 0.8, "learning_rate": 1.9702754789217237e-06, "loss": 0.5556, "step": 4635 }, { "epoch": 0.8, "learning_rate": 1.9669325031824225e-06, "loss": 0.5545, "step": 4636 }, { "epoch": 0.8, "learning_rate": 1.963592056509425e-06, "loss": 0.5614, "step": 4637 }, { "epoch": 0.8, "learning_rate": 1.9602541399544053e-06, "loss": 0.5548, "step": 4638 }, { "epoch": 0.8, "learning_rate": 1.956918754568253e-06, "loss": 0.567, "step": 4639 }, { "epoch": 0.8, "learning_rate": 1.9535859014010525e-06, "loss": 0.561, "step": 4640 }, { "epoch": 0.8, "learning_rate": 1.9502555815020897e-06, "loss": 0.5628, "step": 4641 }, { "epoch": 0.8, "learning_rate": 1.9469277959198584e-06, "loss": 0.5443, "step": 4642 }, { "epoch": 0.8, "learning_rate": 1.943602545702051e-06, "loss": 0.5533, "step": 4643 }, { "epoch": 0.8, "learning_rate": 1.940279831895563e-06, "loss": 0.5671, "step": 4644 }, { "epoch": 0.8, "learning_rate": 1.9369596555464897e-06, "loss": 0.5544, "step": 4645 }, { "epoch": 0.8, "learning_rate": 1.93364201770013e-06, "loss": 0.5607, "step": 4646 }, { "epoch": 0.8, "learning_rate": 1.930326919400979e-06, "loss": 0.5514, "step": 4647 }, { "epoch": 0.81, "learning_rate": 1.9270143616927394e-06, "loss": 0.5637, "step": 4648 }, { "epoch": 0.81, "learning_rate": 1.923704345618309e-06, "loss": 0.56, "step": 4649 }, { "epoch": 0.81, "learning_rate": 1.9203968722197876e-06, "loss": 0.5737, "step": 4650 }, { "epoch": 0.81, "learning_rate": 1.9170919425384695e-06, "loss": 0.5571, "step": 4651 }, { "epoch": 0.81, "learning_rate": 1.913789557614857e-06, "loss": 0.5615, "step": 4652 }, { "epoch": 0.81, "learning_rate": 1.910489718488645e-06, "loss": 0.5574, "step": 4653 }, { "epoch": 0.81, "learning_rate": 1.907192426198724e-06, "loss": 0.5648, "step": 4654 }, { "epoch": 0.81, "learning_rate": 1.9038976817831945e-06, "loss": 0.5498, "step": 4655 }, { "epoch": 0.81, "learning_rate": 1.9006054862793433e-06, "loss": 0.5744, "step": 4656 }, { "epoch": 0.81, "learning_rate": 1.897315840723658e-06, "loss": 0.5634, "step": 4657 }, { "epoch": 0.81, "learning_rate": 1.8940287461518259e-06, "loss": 0.5435, "step": 4658 }, { "epoch": 0.81, "learning_rate": 1.890744203598729e-06, "loss": 0.5471, "step": 4659 }, { "epoch": 0.81, "learning_rate": 1.8874622140984422e-06, "loss": 0.552, "step": 4660 }, { "epoch": 0.81, "learning_rate": 1.884182778684247e-06, "loss": 0.5543, "step": 4661 }, { "epoch": 0.81, "learning_rate": 1.880905898388612e-06, "loss": 0.5617, "step": 4662 }, { "epoch": 0.81, "learning_rate": 1.8776315742432026e-06, "loss": 0.543, "step": 4663 }, { "epoch": 0.81, "learning_rate": 1.8743598072788805e-06, "loss": 0.5716, "step": 4664 }, { "epoch": 0.81, "learning_rate": 1.8710905985257044e-06, "loss": 0.5418, "step": 4665 }, { "epoch": 0.81, "learning_rate": 1.8678239490129258e-06, "loss": 0.5664, "step": 4666 }, { "epoch": 0.81, "learning_rate": 1.8645598597689861e-06, "loss": 0.5694, "step": 4667 }, { "epoch": 0.81, "learning_rate": 1.8612983318215316e-06, "loss": 0.5808, "step": 4668 }, { "epoch": 0.81, "learning_rate": 1.8580393661973916e-06, "loss": 0.5502, "step": 4669 }, { "epoch": 0.81, "learning_rate": 1.8547829639225922e-06, "loss": 0.5485, "step": 4670 }, { "epoch": 0.81, "learning_rate": 1.8515291260223523e-06, "loss": 0.5623, "step": 4671 }, { "epoch": 0.81, "learning_rate": 1.84827785352109e-06, "loss": 0.5499, "step": 4672 }, { "epoch": 0.81, "learning_rate": 1.8450291474423999e-06, "loss": 0.5575, "step": 4673 }, { "epoch": 0.81, "learning_rate": 1.841783008809086e-06, "loss": 0.5425, "step": 4674 }, { "epoch": 0.81, "learning_rate": 1.8385394386431343e-06, "loss": 0.5512, "step": 4675 }, { "epoch": 0.81, "learning_rate": 1.8352984379657247e-06, "loss": 0.5708, "step": 4676 }, { "epoch": 0.81, "learning_rate": 1.832060007797225e-06, "loss": 0.5389, "step": 4677 }, { "epoch": 0.81, "learning_rate": 1.8288241491572e-06, "loss": 0.556, "step": 4678 }, { "epoch": 0.81, "learning_rate": 1.8255908630644015e-06, "loss": 0.5471, "step": 4679 }, { "epoch": 0.81, "learning_rate": 1.822360150536766e-06, "loss": 0.5569, "step": 4680 }, { "epoch": 0.81, "learning_rate": 1.8191320125914324e-06, "loss": 0.5528, "step": 4681 }, { "epoch": 0.81, "learning_rate": 1.8159064502447177e-06, "loss": 0.5635, "step": 4682 }, { "epoch": 0.81, "learning_rate": 1.8126834645121305e-06, "loss": 0.553, "step": 4683 }, { "epoch": 0.81, "learning_rate": 1.8094630564083737e-06, "loss": 0.5511, "step": 4684 }, { "epoch": 0.81, "learning_rate": 1.806245226947333e-06, "loss": 0.5644, "step": 4685 }, { "epoch": 0.81, "learning_rate": 1.8030299771420834e-06, "loss": 0.5652, "step": 4686 }, { "epoch": 0.81, "learning_rate": 1.7998173080048876e-06, "loss": 0.5547, "step": 4687 }, { "epoch": 0.81, "learning_rate": 1.7966072205471985e-06, "loss": 0.5517, "step": 4688 }, { "epoch": 0.81, "learning_rate": 1.7933997157796523e-06, "loss": 0.5606, "step": 4689 }, { "epoch": 0.81, "learning_rate": 1.7901947947120724e-06, "loss": 0.5444, "step": 4690 }, { "epoch": 0.81, "learning_rate": 1.7869924583534749e-06, "loss": 0.5771, "step": 4691 }, { "epoch": 0.81, "learning_rate": 1.7837927077120554e-06, "loss": 0.5345, "step": 4692 }, { "epoch": 0.81, "learning_rate": 1.780595543795195e-06, "loss": 0.5441, "step": 4693 }, { "epoch": 0.81, "learning_rate": 1.7774009676094683e-06, "loss": 0.5641, "step": 4694 }, { "epoch": 0.81, "learning_rate": 1.7742089801606278e-06, "loss": 0.5705, "step": 4695 }, { "epoch": 0.81, "learning_rate": 1.7710195824536092e-06, "loss": 0.551, "step": 4696 }, { "epoch": 0.81, "learning_rate": 1.767832775492543e-06, "loss": 0.5659, "step": 4697 }, { "epoch": 0.81, "learning_rate": 1.764648560280734e-06, "loss": 0.5693, "step": 4698 }, { "epoch": 0.81, "learning_rate": 1.7614669378206773e-06, "loss": 0.5458, "step": 4699 }, { "epoch": 0.81, "learning_rate": 1.758287909114047e-06, "loss": 0.5655, "step": 4700 }, { "epoch": 0.81, "learning_rate": 1.7551114751617049e-06, "loss": 0.538, "step": 4701 }, { "epoch": 0.81, "learning_rate": 1.751937636963692e-06, "loss": 0.546, "step": 4702 }, { "epoch": 0.81, "learning_rate": 1.7487663955192325e-06, "loss": 0.5471, "step": 4703 }, { "epoch": 0.81, "learning_rate": 1.7455977518267387e-06, "loss": 0.5454, "step": 4704 }, { "epoch": 0.82, "learning_rate": 1.7424317068837993e-06, "loss": 0.5642, "step": 4705 }, { "epoch": 0.82, "learning_rate": 1.7392682616871836e-06, "loss": 0.5551, "step": 4706 }, { "epoch": 0.82, "learning_rate": 1.7361074172328507e-06, "loss": 0.5526, "step": 4707 }, { "epoch": 0.82, "learning_rate": 1.732949174515932e-06, "loss": 0.552, "step": 4708 }, { "epoch": 0.82, "learning_rate": 1.7297935345307415e-06, "loss": 0.5581, "step": 4709 }, { "epoch": 0.82, "learning_rate": 1.7266404982707796e-06, "loss": 0.5522, "step": 4710 }, { "epoch": 0.82, "learning_rate": 1.7234900667287214e-06, "loss": 0.5607, "step": 4711 }, { "epoch": 0.82, "learning_rate": 1.7203422408964222e-06, "loss": 0.5575, "step": 4712 }, { "epoch": 0.82, "learning_rate": 1.717197021764917e-06, "loss": 0.5424, "step": 4713 }, { "epoch": 0.82, "learning_rate": 1.7140544103244272e-06, "loss": 0.5607, "step": 4714 }, { "epoch": 0.82, "learning_rate": 1.7109144075643392e-06, "loss": 0.5657, "step": 4715 }, { "epoch": 0.82, "learning_rate": 1.707777014473232e-06, "loss": 0.5648, "step": 4716 }, { "epoch": 0.82, "learning_rate": 1.7046422320388556e-06, "loss": 0.5591, "step": 4717 }, { "epoch": 0.82, "learning_rate": 1.7015100612481395e-06, "loss": 0.5552, "step": 4718 }, { "epoch": 0.82, "learning_rate": 1.698380503087188e-06, "loss": 0.5662, "step": 4719 }, { "epoch": 0.82, "learning_rate": 1.6952535585412921e-06, "loss": 0.5568, "step": 4720 }, { "epoch": 0.82, "learning_rate": 1.6921292285949108e-06, "loss": 0.5575, "step": 4721 }, { "epoch": 0.82, "learning_rate": 1.6890075142316809e-06, "loss": 0.5799, "step": 4722 }, { "epoch": 0.82, "learning_rate": 1.6858884164344224e-06, "loss": 0.563, "step": 4723 }, { "epoch": 0.82, "learning_rate": 1.6827719361851236e-06, "loss": 0.5493, "step": 4724 }, { "epoch": 0.82, "learning_rate": 1.6796580744649537e-06, "loss": 0.5605, "step": 4725 }, { "epoch": 0.82, "learning_rate": 1.6765468322542522e-06, "loss": 0.5431, "step": 4726 }, { "epoch": 0.82, "learning_rate": 1.673438210532543e-06, "loss": 0.5649, "step": 4727 }, { "epoch": 0.82, "learning_rate": 1.6703322102785168e-06, "loss": 0.5597, "step": 4728 }, { "epoch": 0.82, "learning_rate": 1.6672288324700413e-06, "loss": 0.5706, "step": 4729 }, { "epoch": 0.82, "learning_rate": 1.6641280780841606e-06, "loss": 0.5584, "step": 4730 }, { "epoch": 0.82, "learning_rate": 1.6610299480970893e-06, "loss": 0.5648, "step": 4731 }, { "epoch": 0.82, "learning_rate": 1.6579344434842171e-06, "loss": 0.5711, "step": 4732 }, { "epoch": 0.82, "learning_rate": 1.6548415652201112e-06, "loss": 0.5383, "step": 4733 }, { "epoch": 0.82, "learning_rate": 1.651751314278507e-06, "loss": 0.5642, "step": 4734 }, { "epoch": 0.82, "learning_rate": 1.6486636916323106e-06, "loss": 0.5532, "step": 4735 }, { "epoch": 0.82, "learning_rate": 1.6455786982536103e-06, "loss": 0.5594, "step": 4736 }, { "epoch": 0.82, "learning_rate": 1.642496335113658e-06, "loss": 0.557, "step": 4737 }, { "epoch": 0.82, "learning_rate": 1.6394166031828796e-06, "loss": 0.5652, "step": 4738 }, { "epoch": 0.82, "learning_rate": 1.6363395034308704e-06, "loss": 0.558, "step": 4739 }, { "epoch": 0.82, "learning_rate": 1.633265036826406e-06, "loss": 0.5627, "step": 4740 }, { "epoch": 0.82, "learning_rate": 1.6301932043374226e-06, "loss": 0.5524, "step": 4741 }, { "epoch": 0.82, "learning_rate": 1.6271240069310323e-06, "loss": 0.5743, "step": 4742 }, { "epoch": 0.82, "learning_rate": 1.6240574455735158e-06, "loss": 0.5647, "step": 4743 }, { "epoch": 0.82, "learning_rate": 1.6209935212303251e-06, "loss": 0.5617, "step": 4744 }, { "epoch": 0.82, "learning_rate": 1.6179322348660798e-06, "loss": 0.565, "step": 4745 }, { "epoch": 0.82, "learning_rate": 1.6148735874445741e-06, "loss": 0.5667, "step": 4746 }, { "epoch": 0.82, "learning_rate": 1.611817579928765e-06, "loss": 0.5493, "step": 4747 }, { "epoch": 0.82, "learning_rate": 1.6087642132807813e-06, "loss": 0.5576, "step": 4748 }, { "epoch": 0.82, "learning_rate": 1.605713488461923e-06, "loss": 0.5524, "step": 4749 }, { "epoch": 0.82, "learning_rate": 1.6026654064326553e-06, "loss": 0.5581, "step": 4750 }, { "epoch": 0.82, "learning_rate": 1.5996199681526092e-06, "loss": 0.5443, "step": 4751 }, { "epoch": 0.82, "learning_rate": 1.596577174580586e-06, "loss": 0.5378, "step": 4752 }, { "epoch": 0.82, "learning_rate": 1.5935370266745575e-06, "loss": 0.5788, "step": 4753 }, { "epoch": 0.82, "learning_rate": 1.5904995253916578e-06, "loss": 0.5546, "step": 4754 }, { "epoch": 0.82, "learning_rate": 1.587464671688187e-06, "loss": 0.5695, "step": 4755 }, { "epoch": 0.82, "learning_rate": 1.5844324665196209e-06, "loss": 0.5565, "step": 4756 }, { "epoch": 0.82, "learning_rate": 1.581402910840588e-06, "loss": 0.5606, "step": 4757 }, { "epoch": 0.82, "learning_rate": 1.578376005604888e-06, "loss": 0.5649, "step": 4758 }, { "epoch": 0.82, "learning_rate": 1.5753517517654937e-06, "loss": 0.5693, "step": 4759 }, { "epoch": 0.82, "learning_rate": 1.5723301502745325e-06, "loss": 0.5458, "step": 4760 }, { "epoch": 0.82, "learning_rate": 1.5693112020833012e-06, "loss": 0.5632, "step": 4761 }, { "epoch": 0.82, "learning_rate": 1.5662949081422629e-06, "loss": 0.5616, "step": 4762 }, { "epoch": 0.83, "learning_rate": 1.5632812694010435e-06, "loss": 0.5632, "step": 4763 }, { "epoch": 0.83, "learning_rate": 1.5602702868084319e-06, "loss": 0.5485, "step": 4764 }, { "epoch": 0.83, "learning_rate": 1.5572619613123797e-06, "loss": 0.5509, "step": 4765 }, { "epoch": 0.83, "learning_rate": 1.554256293860007e-06, "loss": 0.5569, "step": 4766 }, { "epoch": 0.83, "learning_rate": 1.5512532853975925e-06, "loss": 0.5675, "step": 4767 }, { "epoch": 0.83, "learning_rate": 1.548252936870578e-06, "loss": 0.5449, "step": 4768 }, { "epoch": 0.83, "learning_rate": 1.545255249223573e-06, "loss": 0.5475, "step": 4769 }, { "epoch": 0.83, "learning_rate": 1.5422602234003436e-06, "loss": 0.546, "step": 4770 }, { "epoch": 0.83, "learning_rate": 1.539267860343815e-06, "loss": 0.5533, "step": 4771 }, { "epoch": 0.83, "learning_rate": 1.5362781609960853e-06, "loss": 0.5522, "step": 4772 }, { "epoch": 0.83, "learning_rate": 1.533291126298404e-06, "loss": 0.5509, "step": 4773 }, { "epoch": 0.83, "learning_rate": 1.5303067571911834e-06, "loss": 0.5741, "step": 4774 }, { "epoch": 0.83, "learning_rate": 1.5273250546140028e-06, "loss": 0.5642, "step": 4775 }, { "epoch": 0.83, "learning_rate": 1.524346019505596e-06, "loss": 0.5472, "step": 4776 }, { "epoch": 0.83, "learning_rate": 1.5213696528038556e-06, "loss": 0.5588, "step": 4777 }, { "epoch": 0.83, "learning_rate": 1.5183959554458383e-06, "loss": 0.5555, "step": 4778 }, { "epoch": 0.83, "learning_rate": 1.5154249283677613e-06, "loss": 0.5691, "step": 4779 }, { "epoch": 0.83, "learning_rate": 1.5124565725049978e-06, "loss": 0.5707, "step": 4780 }, { "epoch": 0.83, "learning_rate": 1.5094908887920767e-06, "loss": 0.5832, "step": 4781 }, { "epoch": 0.83, "learning_rate": 1.5065278781626968e-06, "loss": 0.5606, "step": 4782 }, { "epoch": 0.83, "learning_rate": 1.5035675415497064e-06, "loss": 0.5651, "step": 4783 }, { "epoch": 0.83, "learning_rate": 1.5006098798851122e-06, "loss": 0.5589, "step": 4784 }, { "epoch": 0.83, "learning_rate": 1.4976548941000813e-06, "loss": 0.5568, "step": 4785 }, { "epoch": 0.83, "learning_rate": 1.4947025851249376e-06, "loss": 0.5598, "step": 4786 }, { "epoch": 0.83, "learning_rate": 1.4917529538891607e-06, "loss": 0.5678, "step": 4787 }, { "epoch": 0.83, "learning_rate": 1.4888060013213934e-06, "loss": 0.5552, "step": 4788 }, { "epoch": 0.83, "learning_rate": 1.485861728349427e-06, "loss": 0.5618, "step": 4789 }, { "epoch": 0.83, "learning_rate": 1.482920135900211e-06, "loss": 0.5519, "step": 4790 }, { "epoch": 0.83, "learning_rate": 1.4799812248998568e-06, "loss": 0.5454, "step": 4791 }, { "epoch": 0.83, "learning_rate": 1.4770449962736265e-06, "loss": 0.5544, "step": 4792 }, { "epoch": 0.83, "learning_rate": 1.4741114509459376e-06, "loss": 0.5623, "step": 4793 }, { "epoch": 0.83, "learning_rate": 1.471180589840363e-06, "loss": 0.5451, "step": 4794 }, { "epoch": 0.83, "learning_rate": 1.4682524138796338e-06, "loss": 0.5519, "step": 4795 }, { "epoch": 0.83, "learning_rate": 1.4653269239856326e-06, "loss": 0.572, "step": 4796 }, { "epoch": 0.83, "learning_rate": 1.4624041210793939e-06, "loss": 0.56, "step": 4797 }, { "epoch": 0.83, "learning_rate": 1.4594840060811178e-06, "loss": 0.564, "step": 4798 }, { "epoch": 0.83, "learning_rate": 1.4565665799101402e-06, "loss": 0.5495, "step": 4799 }, { "epoch": 0.83, "learning_rate": 1.4536518434849633e-06, "loss": 0.558, "step": 4800 }, { "epoch": 0.83, "learning_rate": 1.4507397977232418e-06, "loss": 0.5557, "step": 4801 }, { "epoch": 0.83, "learning_rate": 1.447830443541779e-06, "loss": 0.5625, "step": 4802 }, { "epoch": 0.83, "learning_rate": 1.4449237818565298e-06, "loss": 0.544, "step": 4803 }, { "epoch": 0.83, "learning_rate": 1.4420198135826103e-06, "loss": 0.55, "step": 4804 }, { "epoch": 0.83, "learning_rate": 1.4391185396342787e-06, "loss": 0.5404, "step": 4805 }, { "epoch": 0.83, "learning_rate": 1.4362199609249482e-06, "loss": 0.56, "step": 4806 }, { "epoch": 0.83, "learning_rate": 1.433324078367183e-06, "loss": 0.5718, "step": 4807 }, { "epoch": 0.83, "learning_rate": 1.430430892872704e-06, "loss": 0.5778, "step": 4808 }, { "epoch": 0.83, "learning_rate": 1.4275404053523757e-06, "loss": 0.5408, "step": 4809 }, { "epoch": 0.83, "learning_rate": 1.4246526167162144e-06, "loss": 0.5557, "step": 4810 }, { "epoch": 0.83, "learning_rate": 1.421767527873391e-06, "loss": 0.5722, "step": 4811 }, { "epoch": 0.83, "learning_rate": 1.4188851397322255e-06, "loss": 0.5614, "step": 4812 }, { "epoch": 0.83, "learning_rate": 1.4160054532001777e-06, "loss": 0.5544, "step": 4813 }, { "epoch": 0.83, "learning_rate": 1.4131284691838721e-06, "loss": 0.5658, "step": 4814 }, { "epoch": 0.83, "learning_rate": 1.410254188589074e-06, "loss": 0.5595, "step": 4815 }, { "epoch": 0.83, "learning_rate": 1.4073826123206946e-06, "loss": 0.5583, "step": 4816 }, { "epoch": 0.83, "learning_rate": 1.4045137412828036e-06, "loss": 0.5594, "step": 4817 }, { "epoch": 0.83, "learning_rate": 1.4016475763786109e-06, "loss": 0.556, "step": 4818 }, { "epoch": 0.83, "learning_rate": 1.3987841185104767e-06, "loss": 0.5512, "step": 4819 }, { "epoch": 0.83, "learning_rate": 1.3959233685799068e-06, "loss": 0.5414, "step": 4820 }, { "epoch": 0.84, "learning_rate": 1.3930653274875616e-06, "loss": 0.5652, "step": 4821 }, { "epoch": 0.84, "learning_rate": 1.3902099961332405e-06, "loss": 0.5613, "step": 4822 }, { "epoch": 0.84, "learning_rate": 1.387357375415891e-06, "loss": 0.5658, "step": 4823 }, { "epoch": 0.84, "learning_rate": 1.3845074662336134e-06, "loss": 0.5369, "step": 4824 }, { "epoch": 0.84, "learning_rate": 1.3816602694836501e-06, "loss": 0.5687, "step": 4825 }, { "epoch": 0.84, "learning_rate": 1.3788157860623862e-06, "loss": 0.54, "step": 4826 }, { "epoch": 0.84, "learning_rate": 1.375974016865359e-06, "loss": 0.5723, "step": 4827 }, { "epoch": 0.84, "learning_rate": 1.373134962787247e-06, "loss": 0.5467, "step": 4828 }, { "epoch": 0.84, "learning_rate": 1.370298624721873e-06, "loss": 0.557, "step": 4829 }, { "epoch": 0.84, "learning_rate": 1.3674650035622105e-06, "loss": 0.5401, "step": 4830 }, { "epoch": 0.84, "learning_rate": 1.3646341002003738e-06, "loss": 0.5577, "step": 4831 }, { "epoch": 0.84, "learning_rate": 1.3618059155276198e-06, "loss": 0.5597, "step": 4832 }, { "epoch": 0.84, "learning_rate": 1.3589804504343497e-06, "loss": 0.5495, "step": 4833 }, { "epoch": 0.84, "learning_rate": 1.3561577058101138e-06, "loss": 0.5675, "step": 4834 }, { "epoch": 0.84, "learning_rate": 1.3533376825436017e-06, "loss": 0.5591, "step": 4835 }, { "epoch": 0.84, "learning_rate": 1.3505203815226443e-06, "loss": 0.5607, "step": 4836 }, { "epoch": 0.84, "learning_rate": 1.3477058036342216e-06, "loss": 0.5476, "step": 4837 }, { "epoch": 0.84, "learning_rate": 1.3448939497644508e-06, "loss": 0.5643, "step": 4838 }, { "epoch": 0.84, "learning_rate": 1.3420848207985937e-06, "loss": 0.5572, "step": 4839 }, { "epoch": 0.84, "learning_rate": 1.3392784176210539e-06, "loss": 0.5576, "step": 4840 }, { "epoch": 0.84, "learning_rate": 1.3364747411153756e-06, "loss": 0.5397, "step": 4841 }, { "epoch": 0.84, "learning_rate": 1.333673792164244e-06, "loss": 0.5489, "step": 4842 }, { "epoch": 0.84, "learning_rate": 1.3308755716494926e-06, "loss": 0.5596, "step": 4843 }, { "epoch": 0.84, "learning_rate": 1.3280800804520888e-06, "loss": 0.5485, "step": 4844 }, { "epoch": 0.84, "learning_rate": 1.32528731945214e-06, "loss": 0.553, "step": 4845 }, { "epoch": 0.84, "learning_rate": 1.3224972895288967e-06, "loss": 0.5524, "step": 4846 }, { "epoch": 0.84, "learning_rate": 1.3197099915607536e-06, "loss": 0.553, "step": 4847 }, { "epoch": 0.84, "learning_rate": 1.3169254264252384e-06, "loss": 0.538, "step": 4848 }, { "epoch": 0.84, "learning_rate": 1.3141435949990188e-06, "loss": 0.5595, "step": 4849 }, { "epoch": 0.84, "learning_rate": 1.3113644981579088e-06, "loss": 0.5351, "step": 4850 }, { "epoch": 0.84, "learning_rate": 1.308588136776855e-06, "loss": 0.5567, "step": 4851 }, { "epoch": 0.84, "learning_rate": 1.3058145117299436e-06, "loss": 0.5635, "step": 4852 }, { "epoch": 0.84, "learning_rate": 1.3030436238903998e-06, "loss": 0.5518, "step": 4853 }, { "epoch": 0.84, "learning_rate": 1.3002754741305935e-06, "loss": 0.5587, "step": 4854 }, { "epoch": 0.84, "learning_rate": 1.2975100633220183e-06, "loss": 0.5532, "step": 4855 }, { "epoch": 0.84, "learning_rate": 1.2947473923353194e-06, "loss": 0.5494, "step": 4856 }, { "epoch": 0.84, "learning_rate": 1.2919874620402727e-06, "loss": 0.5633, "step": 4857 }, { "epoch": 0.84, "learning_rate": 1.2892302733057915e-06, "loss": 0.5679, "step": 4858 }, { "epoch": 0.84, "learning_rate": 1.2864758269999256e-06, "loss": 0.5745, "step": 4859 }, { "epoch": 0.84, "learning_rate": 1.2837241239898669e-06, "loss": 0.5675, "step": 4860 }, { "epoch": 0.84, "learning_rate": 1.2809751651419366e-06, "loss": 0.5636, "step": 4861 }, { "epoch": 0.84, "learning_rate": 1.278228951321594e-06, "loss": 0.5412, "step": 4862 }, { "epoch": 0.84, "learning_rate": 1.2754854833934371e-06, "loss": 0.5496, "step": 4863 }, { "epoch": 0.84, "learning_rate": 1.2727447622211974e-06, "loss": 0.5507, "step": 4864 }, { "epoch": 0.84, "learning_rate": 1.270006788667738e-06, "loss": 0.5522, "step": 4865 }, { "epoch": 0.84, "learning_rate": 1.267271563595065e-06, "loss": 0.5616, "step": 4866 }, { "epoch": 0.84, "learning_rate": 1.2645390878643128e-06, "loss": 0.5577, "step": 4867 }, { "epoch": 0.84, "learning_rate": 1.2618093623357507e-06, "loss": 0.5504, "step": 4868 }, { "epoch": 0.84, "learning_rate": 1.2590823878687853e-06, "loss": 0.5573, "step": 4869 }, { "epoch": 0.84, "learning_rate": 1.2563581653219548e-06, "loss": 0.5688, "step": 4870 }, { "epoch": 0.84, "learning_rate": 1.253636695552931e-06, "loss": 0.5897, "step": 4871 }, { "epoch": 0.84, "learning_rate": 1.2509179794185188e-06, "loss": 0.5649, "step": 4872 }, { "epoch": 0.84, "learning_rate": 1.248202017774659e-06, "loss": 0.5585, "step": 4873 }, { "epoch": 0.84, "learning_rate": 1.2454888114764219e-06, "loss": 0.5678, "step": 4874 }, { "epoch": 0.84, "learning_rate": 1.2427783613780098e-06, "loss": 0.5458, "step": 4875 }, { "epoch": 0.84, "learning_rate": 1.2400706683327636e-06, "loss": 0.5567, "step": 4876 }, { "epoch": 0.84, "learning_rate": 1.2373657331931478e-06, "loss": 0.5482, "step": 4877 }, { "epoch": 0.84, "learning_rate": 1.2346635568107623e-06, "loss": 0.5651, "step": 4878 }, { "epoch": 0.85, "learning_rate": 1.2319641400363413e-06, "loss": 0.5405, "step": 4879 }, { "epoch": 0.85, "learning_rate": 1.229267483719746e-06, "loss": 0.5662, "step": 4880 }, { "epoch": 0.85, "learning_rate": 1.2265735887099694e-06, "loss": 0.5599, "step": 4881 }, { "epoch": 0.85, "learning_rate": 1.2238824558551365e-06, "loss": 0.5759, "step": 4882 }, { "epoch": 0.85, "learning_rate": 1.2211940860025017e-06, "loss": 0.564, "step": 4883 }, { "epoch": 0.85, "learning_rate": 1.218508479998447e-06, "loss": 0.5669, "step": 4884 }, { "epoch": 0.85, "learning_rate": 1.2158256386884926e-06, "loss": 0.5448, "step": 4885 }, { "epoch": 0.85, "learning_rate": 1.2131455629172784e-06, "loss": 0.5603, "step": 4886 }, { "epoch": 0.85, "learning_rate": 1.2104682535285795e-06, "loss": 0.5397, "step": 4887 }, { "epoch": 0.85, "learning_rate": 1.207793711365296e-06, "loss": 0.5778, "step": 4888 }, { "epoch": 0.85, "learning_rate": 1.2051219372694634e-06, "loss": 0.5563, "step": 4889 }, { "epoch": 0.85, "learning_rate": 1.2024529320822398e-06, "loss": 0.5551, "step": 4890 }, { "epoch": 0.85, "learning_rate": 1.1997866966439097e-06, "loss": 0.5447, "step": 4891 }, { "epoch": 0.85, "learning_rate": 1.1971232317938952e-06, "loss": 0.5701, "step": 4892 }, { "epoch": 0.85, "learning_rate": 1.1944625383707377e-06, "loss": 0.553, "step": 4893 }, { "epoch": 0.85, "learning_rate": 1.1918046172121067e-06, "loss": 0.5682, "step": 4894 }, { "epoch": 0.85, "learning_rate": 1.1891494691548012e-06, "loss": 0.5642, "step": 4895 }, { "epoch": 0.85, "learning_rate": 1.186497095034752e-06, "loss": 0.5587, "step": 4896 }, { "epoch": 0.85, "learning_rate": 1.1838474956870016e-06, "loss": 0.5599, "step": 4897 }, { "epoch": 0.85, "learning_rate": 1.1812006719457369e-06, "loss": 0.5568, "step": 4898 }, { "epoch": 0.85, "learning_rate": 1.1785566246442582e-06, "loss": 0.5538, "step": 4899 }, { "epoch": 0.85, "learning_rate": 1.1759153546149983e-06, "loss": 0.549, "step": 4900 }, { "epoch": 0.85, "learning_rate": 1.1732768626895098e-06, "loss": 0.5716, "step": 4901 }, { "epoch": 0.85, "learning_rate": 1.1706411496984781e-06, "loss": 0.5574, "step": 4902 }, { "epoch": 0.85, "learning_rate": 1.16800821647171e-06, "loss": 0.5304, "step": 4903 }, { "epoch": 0.85, "learning_rate": 1.165378063838133e-06, "loss": 0.5594, "step": 4904 }, { "epoch": 0.85, "learning_rate": 1.162750692625807e-06, "loss": 0.5423, "step": 4905 }, { "epoch": 0.85, "learning_rate": 1.1601261036619116e-06, "loss": 0.5469, "step": 4906 }, { "epoch": 0.85, "learning_rate": 1.1575042977727513e-06, "loss": 0.5725, "step": 4907 }, { "epoch": 0.85, "learning_rate": 1.1548852757837513e-06, "loss": 0.5578, "step": 4908 }, { "epoch": 0.85, "learning_rate": 1.152269038519469e-06, "loss": 0.5583, "step": 4909 }, { "epoch": 0.85, "learning_rate": 1.1496555868035753e-06, "loss": 0.5694, "step": 4910 }, { "epoch": 0.85, "learning_rate": 1.1470449214588697e-06, "loss": 0.5476, "step": 4911 }, { "epoch": 0.85, "learning_rate": 1.1444370433072727e-06, "loss": 0.5618, "step": 4912 }, { "epoch": 0.85, "learning_rate": 1.1418319531698286e-06, "loss": 0.5554, "step": 4913 }, { "epoch": 0.85, "learning_rate": 1.139229651866699e-06, "loss": 0.5531, "step": 4914 }, { "epoch": 0.85, "learning_rate": 1.1366301402171775e-06, "loss": 0.5583, "step": 4915 }, { "epoch": 0.85, "learning_rate": 1.1340334190396706e-06, "loss": 0.5527, "step": 4916 }, { "epoch": 0.85, "learning_rate": 1.1314394891517066e-06, "loss": 0.5582, "step": 4917 }, { "epoch": 0.85, "learning_rate": 1.1288483513699421e-06, "loss": 0.568, "step": 4918 }, { "epoch": 0.85, "learning_rate": 1.1262600065101481e-06, "loss": 0.5216, "step": 4919 }, { "epoch": 0.85, "learning_rate": 1.123674455387218e-06, "loss": 0.579, "step": 4920 }, { "epoch": 0.85, "learning_rate": 1.1210916988151643e-06, "loss": 0.5689, "step": 4921 }, { "epoch": 0.85, "learning_rate": 1.118511737607124e-06, "loss": 0.5464, "step": 4922 }, { "epoch": 0.85, "learning_rate": 1.1159345725753512e-06, "loss": 0.5654, "step": 4923 }, { "epoch": 0.85, "learning_rate": 1.1133602045312174e-06, "loss": 0.5521, "step": 4924 }, { "epoch": 0.85, "learning_rate": 1.1107886342852204e-06, "loss": 0.5702, "step": 4925 }, { "epoch": 0.85, "learning_rate": 1.1082198626469687e-06, "loss": 0.5687, "step": 4926 }, { "epoch": 0.85, "learning_rate": 1.1056538904251911e-06, "loss": 0.559, "step": 4927 }, { "epoch": 0.85, "learning_rate": 1.1030907184277452e-06, "loss": 0.5576, "step": 4928 }, { "epoch": 0.85, "learning_rate": 1.1005303474615935e-06, "loss": 0.5618, "step": 4929 }, { "epoch": 0.85, "learning_rate": 1.0979727783328243e-06, "loss": 0.5579, "step": 4930 }, { "epoch": 0.85, "learning_rate": 1.0954180118466428e-06, "loss": 0.5658, "step": 4931 }, { "epoch": 0.85, "learning_rate": 1.0928660488073717e-06, "loss": 0.5691, "step": 4932 }, { "epoch": 0.85, "learning_rate": 1.090316890018449e-06, "loss": 0.5805, "step": 4933 }, { "epoch": 0.85, "learning_rate": 1.0877705362824298e-06, "loss": 0.5624, "step": 4934 }, { "epoch": 0.85, "learning_rate": 1.0852269884009914e-06, "loss": 0.55, "step": 4935 }, { "epoch": 0.86, "learning_rate": 1.0826862471749223e-06, "loss": 0.5509, "step": 4936 }, { "epoch": 0.86, "learning_rate": 1.080148313404127e-06, "loss": 0.5546, "step": 4937 }, { "epoch": 0.86, "learning_rate": 1.0776131878876316e-06, "loss": 0.5624, "step": 4938 }, { "epoch": 0.86, "learning_rate": 1.075080871423575e-06, "loss": 0.5487, "step": 4939 }, { "epoch": 0.86, "learning_rate": 1.0725513648092056e-06, "loss": 0.546, "step": 4940 }, { "epoch": 0.86, "learning_rate": 1.0700246688408977e-06, "loss": 0.5578, "step": 4941 }, { "epoch": 0.86, "learning_rate": 1.067500784314136e-06, "loss": 0.5544, "step": 4942 }, { "epoch": 0.86, "learning_rate": 1.064979712023516e-06, "loss": 0.559, "step": 4943 }, { "epoch": 0.86, "learning_rate": 1.0624614527627563e-06, "loss": 0.5515, "step": 4944 }, { "epoch": 0.86, "learning_rate": 1.0599460073246848e-06, "loss": 0.5644, "step": 4945 }, { "epoch": 0.86, "learning_rate": 1.0574333765012423e-06, "loss": 0.5588, "step": 4946 }, { "epoch": 0.86, "learning_rate": 1.054923561083484e-06, "loss": 0.5601, "step": 4947 }, { "epoch": 0.86, "learning_rate": 1.0524165618615845e-06, "loss": 0.5501, "step": 4948 }, { "epoch": 0.86, "learning_rate": 1.0499123796248234e-06, "loss": 0.5511, "step": 4949 }, { "epoch": 0.86, "learning_rate": 1.047411015161598e-06, "loss": 0.561, "step": 4950 }, { "epoch": 0.86, "learning_rate": 1.0449124692594193e-06, "loss": 0.5429, "step": 4951 }, { "epoch": 0.86, "learning_rate": 1.0424167427049082e-06, "loss": 0.5726, "step": 4952 }, { "epoch": 0.86, "learning_rate": 1.0399238362838004e-06, "loss": 0.5463, "step": 4953 }, { "epoch": 0.86, "learning_rate": 1.0374337507809407e-06, "loss": 0.5818, "step": 4954 }, { "epoch": 0.86, "learning_rate": 1.0349464869802883e-06, "loss": 0.5598, "step": 4955 }, { "epoch": 0.86, "learning_rate": 1.0324620456649115e-06, "loss": 0.5571, "step": 4956 }, { "epoch": 0.86, "learning_rate": 1.0299804276169944e-06, "loss": 0.5609, "step": 4957 }, { "epoch": 0.86, "learning_rate": 1.02750163361783e-06, "loss": 0.5715, "step": 4958 }, { "epoch": 0.86, "learning_rate": 1.0250256644478196e-06, "loss": 0.548, "step": 4959 }, { "epoch": 0.86, "learning_rate": 1.0225525208864796e-06, "loss": 0.5645, "step": 4960 }, { "epoch": 0.86, "learning_rate": 1.0200822037124325e-06, "loss": 0.56, "step": 4961 }, { "epoch": 0.86, "learning_rate": 1.0176147137034155e-06, "loss": 0.5488, "step": 4962 }, { "epoch": 0.86, "learning_rate": 1.015150051636269e-06, "loss": 0.5589, "step": 4963 }, { "epoch": 0.86, "learning_rate": 1.0126882182869524e-06, "loss": 0.5506, "step": 4964 }, { "epoch": 0.86, "learning_rate": 1.0102292144305282e-06, "loss": 0.5513, "step": 4965 }, { "epoch": 0.86, "learning_rate": 1.0077730408411657e-06, "loss": 0.5577, "step": 4966 }, { "epoch": 0.86, "learning_rate": 1.0053196982921542e-06, "loss": 0.5684, "step": 4967 }, { "epoch": 0.86, "learning_rate": 1.0028691875558771e-06, "loss": 0.5593, "step": 4968 }, { "epoch": 0.86, "learning_rate": 1.000421509403834e-06, "loss": 0.5511, "step": 4969 }, { "epoch": 0.86, "learning_rate": 9.979766646066368e-07, "loss": 0.5536, "step": 4970 }, { "epoch": 0.86, "learning_rate": 9.955346539339983e-07, "loss": 0.5497, "step": 4971 }, { "epoch": 0.86, "learning_rate": 9.930954781547398e-07, "loss": 0.5656, "step": 4972 }, { "epoch": 0.86, "learning_rate": 9.906591380367948e-07, "loss": 0.5609, "step": 4973 }, { "epoch": 0.86, "learning_rate": 9.882256343471996e-07, "loss": 0.5563, "step": 4974 }, { "epoch": 0.86, "learning_rate": 9.857949678520984e-07, "loss": 0.5639, "step": 4975 }, { "epoch": 0.86, "learning_rate": 9.833671393167421e-07, "loss": 0.5757, "step": 4976 }, { "epoch": 0.86, "learning_rate": 9.809421495054916e-07, "loss": 0.537, "step": 4977 }, { "epoch": 0.86, "learning_rate": 9.785199991818095e-07, "loss": 0.5695, "step": 4978 }, { "epoch": 0.86, "learning_rate": 9.761006891082636e-07, "loss": 0.5601, "step": 4979 }, { "epoch": 0.86, "learning_rate": 9.736842200465345e-07, "loss": 0.5499, "step": 4980 }, { "epoch": 0.86, "learning_rate": 9.71270592757404e-07, "loss": 0.5499, "step": 4981 }, { "epoch": 0.86, "learning_rate": 9.688598080007526e-07, "loss": 0.5671, "step": 4982 }, { "epoch": 0.86, "learning_rate": 9.664518665355782e-07, "loss": 0.5557, "step": 4983 }, { "epoch": 0.86, "learning_rate": 9.640467691199773e-07, "loss": 0.5563, "step": 4984 }, { "epoch": 0.86, "learning_rate": 9.616445165111477e-07, "loss": 0.5727, "step": 4985 }, { "epoch": 0.86, "learning_rate": 9.592451094653988e-07, "loss": 0.5525, "step": 4986 }, { "epoch": 0.86, "learning_rate": 9.568485487381407e-07, "loss": 0.5508, "step": 4987 }, { "epoch": 0.86, "learning_rate": 9.544548350838856e-07, "loss": 0.5501, "step": 4988 }, { "epoch": 0.86, "learning_rate": 9.520639692562506e-07, "loss": 0.5662, "step": 4989 }, { "epoch": 0.86, "learning_rate": 9.496759520079579e-07, "loss": 0.5646, "step": 4990 }, { "epoch": 0.86, "learning_rate": 9.47290784090833e-07, "loss": 0.5513, "step": 4991 }, { "epoch": 0.86, "learning_rate": 9.449084662557984e-07, "loss": 0.5499, "step": 4992 }, { "epoch": 0.86, "learning_rate": 9.425289992528897e-07, "loss": 0.5578, "step": 4993 }, { "epoch": 0.87, "learning_rate": 9.401523838312354e-07, "loss": 0.5628, "step": 4994 }, { "epoch": 0.87, "learning_rate": 9.377786207390716e-07, "loss": 0.5625, "step": 4995 }, { "epoch": 0.87, "learning_rate": 9.354077107237325e-07, "loss": 0.5736, "step": 4996 }, { "epoch": 0.87, "learning_rate": 9.330396545316589e-07, "loss": 0.5617, "step": 4997 }, { "epoch": 0.87, "learning_rate": 9.306744529083878e-07, "loss": 0.557, "step": 4998 }, { "epoch": 0.87, "learning_rate": 9.283121065985634e-07, "loss": 0.5398, "step": 4999 }, { "epoch": 0.87, "learning_rate": 9.259526163459265e-07, "loss": 0.5562, "step": 5000 }, { "epoch": 0.87, "learning_rate": 9.235959828933195e-07, "loss": 0.5575, "step": 5001 }, { "epoch": 0.87, "learning_rate": 9.212422069826843e-07, "loss": 0.543, "step": 5002 }, { "epoch": 0.87, "learning_rate": 9.188912893550694e-07, "loss": 0.5698, "step": 5003 }, { "epoch": 0.87, "learning_rate": 9.16543230750615e-07, "loss": 0.5474, "step": 5004 }, { "epoch": 0.87, "learning_rate": 9.141980319085642e-07, "loss": 0.5567, "step": 5005 }, { "epoch": 0.87, "learning_rate": 9.118556935672651e-07, "loss": 0.5555, "step": 5006 }, { "epoch": 0.87, "learning_rate": 9.095162164641569e-07, "loss": 0.5557, "step": 5007 }, { "epoch": 0.87, "learning_rate": 9.071796013357836e-07, "loss": 0.5486, "step": 5008 }, { "epoch": 0.87, "learning_rate": 9.048458489177847e-07, "loss": 0.5688, "step": 5009 }, { "epoch": 0.87, "learning_rate": 9.025149599449012e-07, "loss": 0.5612, "step": 5010 }, { "epoch": 0.87, "learning_rate": 9.001869351509684e-07, "loss": 0.5642, "step": 5011 }, { "epoch": 0.87, "learning_rate": 8.978617752689289e-07, "loss": 0.5678, "step": 5012 }, { "epoch": 0.87, "learning_rate": 8.955394810308126e-07, "loss": 0.5676, "step": 5013 }, { "epoch": 0.87, "learning_rate": 8.932200531677537e-07, "loss": 0.5658, "step": 5014 }, { "epoch": 0.87, "learning_rate": 8.909034924099791e-07, "loss": 0.5715, "step": 5015 }, { "epoch": 0.87, "learning_rate": 8.885897994868209e-07, "loss": 0.5546, "step": 5016 }, { "epoch": 0.87, "learning_rate": 8.86278975126702e-07, "loss": 0.5532, "step": 5017 }, { "epoch": 0.87, "learning_rate": 8.839710200571406e-07, "loss": 0.5484, "step": 5018 }, { "epoch": 0.87, "learning_rate": 8.816659350047596e-07, "loss": 0.543, "step": 5019 }, { "epoch": 0.87, "learning_rate": 8.79363720695271e-07, "loss": 0.5547, "step": 5020 }, { "epoch": 0.87, "learning_rate": 8.770643778534849e-07, "loss": 0.5625, "step": 5021 }, { "epoch": 0.87, "learning_rate": 8.747679072033067e-07, "loss": 0.5525, "step": 5022 }, { "epoch": 0.87, "learning_rate": 8.724743094677446e-07, "loss": 0.5607, "step": 5023 }, { "epoch": 0.87, "learning_rate": 8.701835853688889e-07, "loss": 0.552, "step": 5024 }, { "epoch": 0.87, "learning_rate": 8.678957356279372e-07, "loss": 0.5575, "step": 5025 }, { "epoch": 0.87, "learning_rate": 8.656107609651765e-07, "loss": 0.5534, "step": 5026 }, { "epoch": 0.87, "learning_rate": 8.633286620999915e-07, "loss": 0.566, "step": 5027 }, { "epoch": 0.87, "learning_rate": 8.610494397508551e-07, "loss": 0.5509, "step": 5028 }, { "epoch": 0.87, "learning_rate": 8.587730946353445e-07, "loss": 0.5539, "step": 5029 }, { "epoch": 0.87, "learning_rate": 8.564996274701242e-07, "loss": 0.5582, "step": 5030 }, { "epoch": 0.87, "learning_rate": 8.542290389709529e-07, "loss": 0.5528, "step": 5031 }, { "epoch": 0.87, "learning_rate": 8.519613298526863e-07, "loss": 0.543, "step": 5032 }, { "epoch": 0.87, "learning_rate": 8.496965008292712e-07, "loss": 0.5604, "step": 5033 }, { "epoch": 0.87, "learning_rate": 8.474345526137462e-07, "loss": 0.5536, "step": 5034 }, { "epoch": 0.87, "learning_rate": 8.451754859182482e-07, "loss": 0.5517, "step": 5035 }, { "epoch": 0.87, "learning_rate": 8.429193014540015e-07, "loss": 0.5666, "step": 5036 }, { "epoch": 0.87, "learning_rate": 8.40665999931326e-07, "loss": 0.5565, "step": 5037 }, { "epoch": 0.87, "learning_rate": 8.384155820596318e-07, "loss": 0.574, "step": 5038 }, { "epoch": 0.87, "learning_rate": 8.361680485474233e-07, "loss": 0.563, "step": 5039 }, { "epoch": 0.87, "learning_rate": 8.339234001022956e-07, "loss": 0.546, "step": 5040 }, { "epoch": 0.87, "learning_rate": 8.316816374309345e-07, "loss": 0.5685, "step": 5041 }, { "epoch": 0.87, "learning_rate": 8.29442761239121e-07, "loss": 0.5578, "step": 5042 }, { "epoch": 0.87, "learning_rate": 8.272067722317234e-07, "loss": 0.5596, "step": 5043 }, { "epoch": 0.87, "learning_rate": 8.249736711127009e-07, "loss": 0.5606, "step": 5044 }, { "epoch": 0.87, "learning_rate": 8.22743458585109e-07, "loss": 0.5621, "step": 5045 }, { "epoch": 0.87, "learning_rate": 8.205161353510871e-07, "loss": 0.5643, "step": 5046 }, { "epoch": 0.87, "learning_rate": 8.182917021118664e-07, "loss": 0.5516, "step": 5047 }, { "epoch": 0.87, "learning_rate": 8.160701595677745e-07, "loss": 0.5508, "step": 5048 }, { "epoch": 0.87, "learning_rate": 8.138515084182197e-07, "loss": 0.5454, "step": 5049 }, { "epoch": 0.87, "learning_rate": 8.116357493617055e-07, "loss": 0.5562, "step": 5050 }, { "epoch": 0.87, "learning_rate": 8.094228830958239e-07, "loss": 0.5653, "step": 5051 }, { "epoch": 0.88, "learning_rate": 8.072129103172555e-07, "loss": 0.5663, "step": 5052 }, { "epoch": 0.88, "learning_rate": 8.050058317217713e-07, "loss": 0.5671, "step": 5053 }, { "epoch": 0.88, "learning_rate": 8.028016480042267e-07, "loss": 0.5382, "step": 5054 }, { "epoch": 0.88, "learning_rate": 8.006003598585743e-07, "loss": 0.5511, "step": 5055 }, { "epoch": 0.88, "learning_rate": 7.984019679778465e-07, "loss": 0.57, "step": 5056 }, { "epoch": 0.88, "learning_rate": 7.962064730541674e-07, "loss": 0.5549, "step": 5057 }, { "epoch": 0.88, "learning_rate": 7.940138757787507e-07, "loss": 0.5394, "step": 5058 }, { "epoch": 0.88, "learning_rate": 7.918241768418955e-07, "loss": 0.54, "step": 5059 }, { "epoch": 0.88, "learning_rate": 7.896373769329868e-07, "loss": 0.5561, "step": 5060 }, { "epoch": 0.88, "learning_rate": 7.874534767405029e-07, "loss": 0.5497, "step": 5061 }, { "epoch": 0.88, "learning_rate": 7.852724769520026e-07, "loss": 0.5538, "step": 5062 }, { "epoch": 0.88, "learning_rate": 7.830943782541356e-07, "loss": 0.5513, "step": 5063 }, { "epoch": 0.88, "learning_rate": 7.809191813326356e-07, "loss": 0.5459, "step": 5064 }, { "epoch": 0.88, "learning_rate": 7.787468868723269e-07, "loss": 0.5488, "step": 5065 }, { "epoch": 0.88, "learning_rate": 7.765774955571126e-07, "loss": 0.5392, "step": 5066 }, { "epoch": 0.88, "learning_rate": 7.744110080699907e-07, "loss": 0.5565, "step": 5067 }, { "epoch": 0.88, "learning_rate": 7.722474250930389e-07, "loss": 0.5397, "step": 5068 }, { "epoch": 0.88, "learning_rate": 7.700867473074225e-07, "loss": 0.5643, "step": 5069 }, { "epoch": 0.88, "learning_rate": 7.679289753933905e-07, "loss": 0.5681, "step": 5070 }, { "epoch": 0.88, "learning_rate": 7.657741100302807e-07, "loss": 0.5533, "step": 5071 }, { "epoch": 0.88, "learning_rate": 7.636221518965126e-07, "loss": 0.5603, "step": 5072 }, { "epoch": 0.88, "learning_rate": 7.614731016695898e-07, "loss": 0.5815, "step": 5073 }, { "epoch": 0.88, "learning_rate": 7.593269600261055e-07, "loss": 0.55, "step": 5074 }, { "epoch": 0.88, "learning_rate": 7.571837276417305e-07, "loss": 0.5351, "step": 5075 }, { "epoch": 0.88, "learning_rate": 7.550434051912259e-07, "loss": 0.5426, "step": 5076 }, { "epoch": 0.88, "learning_rate": 7.529059933484284e-07, "loss": 0.5762, "step": 5077 }, { "epoch": 0.88, "learning_rate": 7.507714927862686e-07, "loss": 0.5728, "step": 5078 }, { "epoch": 0.88, "learning_rate": 7.486399041767544e-07, "loss": 0.5562, "step": 5079 }, { "epoch": 0.88, "learning_rate": 7.46511228190977e-07, "loss": 0.5673, "step": 5080 }, { "epoch": 0.88, "learning_rate": 7.443854654991123e-07, "loss": 0.5521, "step": 5081 }, { "epoch": 0.88, "learning_rate": 7.422626167704172e-07, "loss": 0.5661, "step": 5082 }, { "epoch": 0.88, "learning_rate": 7.401426826732327e-07, "loss": 0.5564, "step": 5083 }, { "epoch": 0.88, "learning_rate": 7.380256638749828e-07, "loss": 0.5587, "step": 5084 }, { "epoch": 0.88, "learning_rate": 7.359115610421719e-07, "loss": 0.5545, "step": 5085 }, { "epoch": 0.88, "learning_rate": 7.338003748403855e-07, "loss": 0.5479, "step": 5086 }, { "epoch": 0.88, "learning_rate": 7.316921059342963e-07, "loss": 0.5619, "step": 5087 }, { "epoch": 0.88, "learning_rate": 7.295867549876523e-07, "loss": 0.5609, "step": 5088 }, { "epoch": 0.88, "learning_rate": 7.274843226632855e-07, "loss": 0.5587, "step": 5089 }, { "epoch": 0.88, "learning_rate": 7.253848096231065e-07, "loss": 0.5735, "step": 5090 }, { "epoch": 0.88, "learning_rate": 7.232882165281141e-07, "loss": 0.548, "step": 5091 }, { "epoch": 0.88, "learning_rate": 7.211945440383794e-07, "loss": 0.5483, "step": 5092 }, { "epoch": 0.88, "learning_rate": 7.191037928130585e-07, "loss": 0.5566, "step": 5093 }, { "epoch": 0.88, "learning_rate": 7.170159635103857e-07, "loss": 0.5615, "step": 5094 }, { "epoch": 0.88, "learning_rate": 7.149310567876766e-07, "loss": 0.5623, "step": 5095 }, { "epoch": 0.88, "learning_rate": 7.12849073301326e-07, "loss": 0.5686, "step": 5096 }, { "epoch": 0.88, "learning_rate": 7.107700137068108e-07, "loss": 0.5485, "step": 5097 }, { "epoch": 0.88, "learning_rate": 7.086938786586839e-07, "loss": 0.5699, "step": 5098 }, { "epoch": 0.88, "learning_rate": 7.066206688105781e-07, "loss": 0.5464, "step": 5099 }, { "epoch": 0.88, "learning_rate": 7.04550384815208e-07, "loss": 0.5647, "step": 5100 }, { "epoch": 0.88, "learning_rate": 7.024830273243654e-07, "loss": 0.5576, "step": 5101 }, { "epoch": 0.88, "learning_rate": 7.004185969889188e-07, "loss": 0.5676, "step": 5102 }, { "epoch": 0.88, "learning_rate": 6.983570944588158e-07, "loss": 0.5406, "step": 5103 }, { "epoch": 0.88, "learning_rate": 6.962985203830874e-07, "loss": 0.5658, "step": 5104 }, { "epoch": 0.88, "learning_rate": 6.942428754098351e-07, "loss": 0.568, "step": 5105 }, { "epoch": 0.88, "learning_rate": 6.921901601862413e-07, "loss": 0.561, "step": 5106 }, { "epoch": 0.88, "learning_rate": 6.901403753585712e-07, "loss": 0.5543, "step": 5107 }, { "epoch": 0.88, "learning_rate": 6.880935215721563e-07, "loss": 0.5804, "step": 5108 }, { "epoch": 0.88, "learning_rate": 6.860495994714133e-07, "loss": 0.5588, "step": 5109 }, { "epoch": 0.89, "learning_rate": 6.840086096998367e-07, "loss": 0.5473, "step": 5110 }, { "epoch": 0.89, "learning_rate": 6.819705528999932e-07, "loss": 0.5609, "step": 5111 }, { "epoch": 0.89, "learning_rate": 6.799354297135275e-07, "loss": 0.5583, "step": 5112 }, { "epoch": 0.89, "learning_rate": 6.779032407811636e-07, "loss": 0.5509, "step": 5113 }, { "epoch": 0.89, "learning_rate": 6.758739867426989e-07, "loss": 0.5484, "step": 5114 }, { "epoch": 0.89, "learning_rate": 6.738476682370065e-07, "loss": 0.5437, "step": 5115 }, { "epoch": 0.89, "learning_rate": 6.718242859020352e-07, "loss": 0.5668, "step": 5116 }, { "epoch": 0.89, "learning_rate": 6.698038403748142e-07, "loss": 0.5556, "step": 5117 }, { "epoch": 0.89, "learning_rate": 6.677863322914413e-07, "loss": 0.5525, "step": 5118 }, { "epoch": 0.89, "learning_rate": 6.657717622870919e-07, "loss": 0.5532, "step": 5119 }, { "epoch": 0.89, "learning_rate": 6.637601309960196e-07, "loss": 0.5534, "step": 5120 }, { "epoch": 0.89, "learning_rate": 6.6175143905155e-07, "loss": 0.5409, "step": 5121 }, { "epoch": 0.89, "learning_rate": 6.597456870860808e-07, "loss": 0.5522, "step": 5122 }, { "epoch": 0.89, "learning_rate": 6.577428757310888e-07, "loss": 0.5495, "step": 5123 }, { "epoch": 0.89, "learning_rate": 6.557430056171221e-07, "loss": 0.5777, "step": 5124 }, { "epoch": 0.89, "learning_rate": 6.537460773738035e-07, "loss": 0.547, "step": 5125 }, { "epoch": 0.89, "learning_rate": 6.517520916298325e-07, "loss": 0.5569, "step": 5126 }, { "epoch": 0.89, "learning_rate": 6.497610490129769e-07, "loss": 0.5434, "step": 5127 }, { "epoch": 0.89, "learning_rate": 6.477729501500796e-07, "loss": 0.5632, "step": 5128 }, { "epoch": 0.89, "learning_rate": 6.457877956670611e-07, "loss": 0.5477, "step": 5129 }, { "epoch": 0.89, "learning_rate": 6.438055861889092e-07, "loss": 0.5792, "step": 5130 }, { "epoch": 0.89, "learning_rate": 6.41826322339687e-07, "loss": 0.5592, "step": 5131 }, { "epoch": 0.89, "learning_rate": 6.398500047425283e-07, "loss": 0.5571, "step": 5132 }, { "epoch": 0.89, "learning_rate": 6.378766340196451e-07, "loss": 0.5607, "step": 5133 }, { "epoch": 0.89, "learning_rate": 6.359062107923153e-07, "loss": 0.5601, "step": 5134 }, { "epoch": 0.89, "learning_rate": 6.339387356808912e-07, "loss": 0.5585, "step": 5135 }, { "epoch": 0.89, "learning_rate": 6.319742093047964e-07, "loss": 0.5639, "step": 5136 }, { "epoch": 0.89, "learning_rate": 6.300126322825273e-07, "loss": 0.5416, "step": 5137 }, { "epoch": 0.89, "learning_rate": 6.280540052316486e-07, "loss": 0.5575, "step": 5138 }, { "epoch": 0.89, "learning_rate": 6.260983287688016e-07, "loss": 0.5626, "step": 5139 }, { "epoch": 0.89, "learning_rate": 6.24145603509696e-07, "loss": 0.5549, "step": 5140 }, { "epoch": 0.89, "learning_rate": 6.221958300691088e-07, "loss": 0.5541, "step": 5141 }, { "epoch": 0.89, "learning_rate": 6.202490090608948e-07, "loss": 0.5631, "step": 5142 }, { "epoch": 0.89, "learning_rate": 6.183051410979734e-07, "loss": 0.5517, "step": 5143 }, { "epoch": 0.89, "learning_rate": 6.163642267923376e-07, "loss": 0.5524, "step": 5144 }, { "epoch": 0.89, "learning_rate": 6.14426266755046e-07, "loss": 0.5575, "step": 5145 }, { "epoch": 0.89, "learning_rate": 6.124912615962341e-07, "loss": 0.5401, "step": 5146 }, { "epoch": 0.89, "learning_rate": 6.105592119251025e-07, "loss": 0.5793, "step": 5147 }, { "epoch": 0.89, "learning_rate": 6.086301183499188e-07, "loss": 0.5535, "step": 5148 }, { "epoch": 0.89, "learning_rate": 6.067039814780307e-07, "loss": 0.5548, "step": 5149 }, { "epoch": 0.89, "learning_rate": 6.047808019158408e-07, "loss": 0.5628, "step": 5150 }, { "epoch": 0.89, "learning_rate": 6.028605802688281e-07, "loss": 0.5511, "step": 5151 }, { "epoch": 0.89, "learning_rate": 6.009433171415424e-07, "loss": 0.5651, "step": 5152 }, { "epoch": 0.89, "learning_rate": 5.990290131375986e-07, "loss": 0.5475, "step": 5153 }, { "epoch": 0.89, "learning_rate": 5.971176688596791e-07, "loss": 0.5667, "step": 5154 }, { "epoch": 0.89, "learning_rate": 5.952092849095393e-07, "loss": 0.5386, "step": 5155 }, { "epoch": 0.89, "learning_rate": 5.933038618879972e-07, "loss": 0.5567, "step": 5156 }, { "epoch": 0.89, "learning_rate": 5.914014003949408e-07, "loss": 0.5584, "step": 5157 }, { "epoch": 0.89, "learning_rate": 5.895019010293257e-07, "loss": 0.5622, "step": 5158 }, { "epoch": 0.89, "learning_rate": 5.876053643891766e-07, "loss": 0.5496, "step": 5159 }, { "epoch": 0.89, "learning_rate": 5.857117910715826e-07, "loss": 0.5675, "step": 5160 }, { "epoch": 0.89, "learning_rate": 5.838211816727002e-07, "loss": 0.5591, "step": 5161 }, { "epoch": 0.89, "learning_rate": 5.819335367877554e-07, "loss": 0.5619, "step": 5162 }, { "epoch": 0.89, "learning_rate": 5.800488570110396e-07, "loss": 0.5439, "step": 5163 }, { "epoch": 0.89, "learning_rate": 5.78167142935907e-07, "loss": 0.5595, "step": 5164 }, { "epoch": 0.89, "learning_rate": 5.762883951547848e-07, "loss": 0.5432, "step": 5165 }, { "epoch": 0.89, "learning_rate": 5.74412614259161e-07, "loss": 0.5629, "step": 5166 }, { "epoch": 0.9, "learning_rate": 5.725398008395899e-07, "loss": 0.5469, "step": 5167 }, { "epoch": 0.9, "learning_rate": 5.706699554856965e-07, "loss": 0.5575, "step": 5168 }, { "epoch": 0.9, "learning_rate": 5.688030787861664e-07, "loss": 0.5432, "step": 5169 }, { "epoch": 0.9, "learning_rate": 5.669391713287519e-07, "loss": 0.5689, "step": 5170 }, { "epoch": 0.9, "learning_rate": 5.65078233700268e-07, "loss": 0.5555, "step": 5171 }, { "epoch": 0.9, "learning_rate": 5.632202664866026e-07, "loss": 0.5443, "step": 5172 }, { "epoch": 0.9, "learning_rate": 5.613652702727002e-07, "loss": 0.5519, "step": 5173 }, { "epoch": 0.9, "learning_rate": 5.595132456425711e-07, "loss": 0.5764, "step": 5174 }, { "epoch": 0.9, "learning_rate": 5.576641931792947e-07, "loss": 0.5646, "step": 5175 }, { "epoch": 0.9, "learning_rate": 5.55818113465012e-07, "loss": 0.568, "step": 5176 }, { "epoch": 0.9, "learning_rate": 5.539750070809258e-07, "loss": 0.5545, "step": 5177 }, { "epoch": 0.9, "learning_rate": 5.521348746073063e-07, "loss": 0.5381, "step": 5178 }, { "epoch": 0.9, "learning_rate": 5.502977166234857e-07, "loss": 0.5628, "step": 5179 }, { "epoch": 0.9, "learning_rate": 5.484635337078581e-07, "loss": 0.5457, "step": 5180 }, { "epoch": 0.9, "learning_rate": 5.466323264378859e-07, "loss": 0.5593, "step": 5181 }, { "epoch": 0.9, "learning_rate": 5.448040953900902e-07, "loss": 0.5576, "step": 5182 }, { "epoch": 0.9, "learning_rate": 5.429788411400571e-07, "loss": 0.5607, "step": 5183 }, { "epoch": 0.9, "learning_rate": 5.411565642624328e-07, "loss": 0.5651, "step": 5184 }, { "epoch": 0.9, "learning_rate": 5.393372653309315e-07, "loss": 0.5502, "step": 5185 }, { "epoch": 0.9, "learning_rate": 5.375209449183261e-07, "loss": 0.5475, "step": 5186 }, { "epoch": 0.9, "learning_rate": 5.357076035964492e-07, "loss": 0.5677, "step": 5187 }, { "epoch": 0.9, "learning_rate": 5.33897241936202e-07, "loss": 0.5622, "step": 5188 }, { "epoch": 0.9, "learning_rate": 5.320898605075431e-07, "loss": 0.5513, "step": 5189 }, { "epoch": 0.9, "learning_rate": 5.302854598794938e-07, "loss": 0.5608, "step": 5190 }, { "epoch": 0.9, "learning_rate": 5.284840406201375e-07, "loss": 0.5704, "step": 5191 }, { "epoch": 0.9, "learning_rate": 5.266856032966172e-07, "loss": 0.5871, "step": 5192 }, { "epoch": 0.9, "learning_rate": 5.24890148475139e-07, "loss": 0.5462, "step": 5193 }, { "epoch": 0.9, "learning_rate": 5.230976767209706e-07, "loss": 0.5561, "step": 5194 }, { "epoch": 0.9, "learning_rate": 5.213081885984384e-07, "loss": 0.5488, "step": 5195 }, { "epoch": 0.9, "learning_rate": 5.195216846709305e-07, "loss": 0.5555, "step": 5196 }, { "epoch": 0.9, "learning_rate": 5.17738165500894e-07, "loss": 0.5435, "step": 5197 }, { "epoch": 0.9, "learning_rate": 5.159576316498416e-07, "loss": 0.5645, "step": 5198 }, { "epoch": 0.9, "learning_rate": 5.141800836783383e-07, "loss": 0.5588, "step": 5199 }, { "epoch": 0.9, "learning_rate": 5.124055221460145e-07, "loss": 0.5583, "step": 5200 }, { "epoch": 0.9, "learning_rate": 5.106339476115596e-07, "loss": 0.5476, "step": 5201 }, { "epoch": 0.9, "learning_rate": 5.088653606327209e-07, "loss": 0.5545, "step": 5202 }, { "epoch": 0.9, "learning_rate": 5.070997617663054e-07, "loss": 0.5446, "step": 5203 }, { "epoch": 0.9, "learning_rate": 5.053371515681826e-07, "loss": 0.5756, "step": 5204 }, { "epoch": 0.9, "learning_rate": 5.035775305932777e-07, "loss": 0.5556, "step": 5205 }, { "epoch": 0.9, "learning_rate": 5.018208993955731e-07, "loss": 0.5621, "step": 5206 }, { "epoch": 0.9, "learning_rate": 5.000672585281141e-07, "loss": 0.557, "step": 5207 }, { "epoch": 0.9, "learning_rate": 4.983166085430036e-07, "loss": 0.554, "step": 5208 }, { "epoch": 0.9, "learning_rate": 4.965689499914017e-07, "loss": 0.5704, "step": 5209 }, { "epoch": 0.9, "learning_rate": 4.948242834235251e-07, "loss": 0.5583, "step": 5210 }, { "epoch": 0.9, "learning_rate": 4.930826093886543e-07, "loss": 0.5488, "step": 5211 }, { "epoch": 0.9, "learning_rate": 4.913439284351207e-07, "loss": 0.557, "step": 5212 }, { "epoch": 0.9, "learning_rate": 4.896082411103175e-07, "loss": 0.562, "step": 5213 }, { "epoch": 0.9, "learning_rate": 4.878755479606967e-07, "loss": 0.5357, "step": 5214 }, { "epoch": 0.9, "learning_rate": 4.861458495317628e-07, "loss": 0.56, "step": 5215 }, { "epoch": 0.9, "learning_rate": 4.844191463680803e-07, "loss": 0.5557, "step": 5216 }, { "epoch": 0.9, "learning_rate": 4.826954390132721e-07, "loss": 0.5745, "step": 5217 }, { "epoch": 0.9, "learning_rate": 4.809747280100163e-07, "loss": 0.5576, "step": 5218 }, { "epoch": 0.9, "learning_rate": 4.792570139000463e-07, "loss": 0.5474, "step": 5219 }, { "epoch": 0.9, "learning_rate": 4.775422972241539e-07, "loss": 0.5557, "step": 5220 }, { "epoch": 0.9, "learning_rate": 4.7583057852218615e-07, "loss": 0.5369, "step": 5221 }, { "epoch": 0.9, "learning_rate": 4.7412185833304867e-07, "loss": 0.5623, "step": 5222 }, { "epoch": 0.9, "learning_rate": 4.7241613719469784e-07, "loss": 0.5664, "step": 5223 }, { "epoch": 0.9, "learning_rate": 4.7071341564415284e-07, "loss": 0.5523, "step": 5224 }, { "epoch": 0.91, "learning_rate": 4.6901369421748386e-07, "loss": 0.5722, "step": 5225 }, { "epoch": 0.91, "learning_rate": 4.6731697344981506e-07, "loss": 0.5702, "step": 5226 }, { "epoch": 0.91, "learning_rate": 4.6562325387533137e-07, "loss": 0.5635, "step": 5227 }, { "epoch": 0.91, "learning_rate": 4.639325360272684e-07, "loss": 0.5469, "step": 5228 }, { "epoch": 0.91, "learning_rate": 4.622448204379171e-07, "loss": 0.5461, "step": 5229 }, { "epoch": 0.91, "learning_rate": 4.605601076386268e-07, "loss": 0.5495, "step": 5230 }, { "epoch": 0.91, "learning_rate": 4.5887839815979774e-07, "loss": 0.5631, "step": 5231 }, { "epoch": 0.91, "learning_rate": 4.5719969253088524e-07, "loss": 0.5584, "step": 5232 }, { "epoch": 0.91, "learning_rate": 4.55523991280401e-07, "loss": 0.5626, "step": 5233 }, { "epoch": 0.91, "learning_rate": 4.538512949359075e-07, "loss": 0.557, "step": 5234 }, { "epoch": 0.91, "learning_rate": 4.5218160402402234e-07, "loss": 0.5714, "step": 5235 }, { "epoch": 0.91, "learning_rate": 4.5051491907042055e-07, "loss": 0.5381, "step": 5236 }, { "epoch": 0.91, "learning_rate": 4.4885124059982686e-07, "loss": 0.5581, "step": 5237 }, { "epoch": 0.91, "learning_rate": 4.47190569136019e-07, "loss": 0.548, "step": 5238 }, { "epoch": 0.91, "learning_rate": 4.455329052018287e-07, "loss": 0.555, "step": 5239 }, { "epoch": 0.91, "learning_rate": 4.4387824931914404e-07, "loss": 0.5548, "step": 5240 }, { "epoch": 0.91, "learning_rate": 4.4222660200890276e-07, "loss": 0.5373, "step": 5241 }, { "epoch": 0.91, "learning_rate": 4.4057796379109565e-07, "loss": 0.5515, "step": 5242 }, { "epoch": 0.91, "learning_rate": 4.389323351847674e-07, "loss": 0.5517, "step": 5243 }, { "epoch": 0.91, "learning_rate": 4.3728971670801366e-07, "loss": 0.5566, "step": 5244 }, { "epoch": 0.91, "learning_rate": 4.356501088779841e-07, "loss": 0.5628, "step": 5245 }, { "epoch": 0.91, "learning_rate": 4.3401351221087795e-07, "loss": 0.5479, "step": 5246 }, { "epoch": 0.91, "learning_rate": 4.3237992722195197e-07, "loss": 0.5542, "step": 5247 }, { "epoch": 0.91, "learning_rate": 4.3074935442550483e-07, "loss": 0.5473, "step": 5248 }, { "epoch": 0.91, "learning_rate": 4.291217943348991e-07, "loss": 0.5478, "step": 5249 }, { "epoch": 0.91, "learning_rate": 4.274972474625383e-07, "loss": 0.5561, "step": 5250 }, { "epoch": 0.91, "learning_rate": 4.258757143198844e-07, "loss": 0.5532, "step": 5251 }, { "epoch": 0.91, "learning_rate": 4.242571954174457e-07, "loss": 0.5481, "step": 5252 }, { "epoch": 0.91, "learning_rate": 4.226416912647857e-07, "loss": 0.551, "step": 5253 }, { "epoch": 0.91, "learning_rate": 4.210292023705165e-07, "loss": 0.5491, "step": 5254 }, { "epoch": 0.91, "learning_rate": 4.1941972924229854e-07, "loss": 0.555, "step": 5255 }, { "epoch": 0.91, "learning_rate": 4.1781327238684775e-07, "loss": 0.5472, "step": 5256 }, { "epoch": 0.91, "learning_rate": 4.162098323099284e-07, "loss": 0.569, "step": 5257 }, { "epoch": 0.91, "learning_rate": 4.146094095163544e-07, "loss": 0.5454, "step": 5258 }, { "epoch": 0.91, "learning_rate": 4.1301200450998724e-07, "loss": 0.5367, "step": 5259 }, { "epoch": 0.91, "learning_rate": 4.1141761779374457e-07, "loss": 0.555, "step": 5260 }, { "epoch": 0.91, "learning_rate": 4.098262498695893e-07, "loss": 0.564, "step": 5261 }, { "epoch": 0.91, "learning_rate": 4.0823790123853514e-07, "loss": 0.5714, "step": 5262 }, { "epoch": 0.91, "learning_rate": 4.0665257240064316e-07, "loss": 0.5577, "step": 5263 }, { "epoch": 0.91, "learning_rate": 4.0507026385502747e-07, "loss": 0.5539, "step": 5264 }, { "epoch": 0.91, "learning_rate": 4.034909760998473e-07, "loss": 0.5527, "step": 5265 }, { "epoch": 0.91, "learning_rate": 4.0191470963231503e-07, "loss": 0.559, "step": 5266 }, { "epoch": 0.91, "learning_rate": 4.003414649486892e-07, "loss": 0.5637, "step": 5267 }, { "epoch": 0.91, "learning_rate": 3.987712425442758e-07, "loss": 0.551, "step": 5268 }, { "epoch": 0.91, "learning_rate": 3.9720404291343385e-07, "loss": 0.5603, "step": 5269 }, { "epoch": 0.91, "learning_rate": 3.956398665495664e-07, "loss": 0.5506, "step": 5270 }, { "epoch": 0.91, "learning_rate": 3.9407871394512633e-07, "loss": 0.5689, "step": 5271 }, { "epoch": 0.91, "learning_rate": 3.9252058559161257e-07, "loss": 0.5648, "step": 5272 }, { "epoch": 0.91, "learning_rate": 3.9096548197957717e-07, "loss": 0.5649, "step": 5273 }, { "epoch": 0.91, "learning_rate": 3.894134035986141e-07, "loss": 0.54, "step": 5274 }, { "epoch": 0.91, "learning_rate": 3.87864350937367e-07, "loss": 0.5492, "step": 5275 }, { "epoch": 0.91, "learning_rate": 3.863183244835289e-07, "loss": 0.5704, "step": 5276 }, { "epoch": 0.91, "learning_rate": 3.8477532472383726e-07, "loss": 0.5586, "step": 5277 }, { "epoch": 0.91, "learning_rate": 3.832353521440768e-07, "loss": 0.5481, "step": 5278 }, { "epoch": 0.91, "learning_rate": 3.816984072290808e-07, "loss": 0.5469, "step": 5279 }, { "epoch": 0.91, "learning_rate": 3.8016449046273e-07, "loss": 0.5643, "step": 5280 }, { "epoch": 0.91, "learning_rate": 3.78633602327948e-07, "loss": 0.5435, "step": 5281 }, { "epoch": 0.91, "learning_rate": 3.7710574330670934e-07, "loss": 0.5449, "step": 5282 }, { "epoch": 0.92, "learning_rate": 3.755809138800326e-07, "loss": 0.5565, "step": 5283 }, { "epoch": 0.92, "learning_rate": 3.740591145279815e-07, "loss": 0.5501, "step": 5284 }, { "epoch": 0.92, "learning_rate": 3.725403457296672e-07, "loss": 0.5406, "step": 5285 }, { "epoch": 0.92, "learning_rate": 3.7102460796324844e-07, "loss": 0.5627, "step": 5286 }, { "epoch": 0.92, "learning_rate": 3.695119017059268e-07, "loss": 0.5482, "step": 5287 }, { "epoch": 0.92, "learning_rate": 3.680022274339501e-07, "loss": 0.5605, "step": 5288 }, { "epoch": 0.92, "learning_rate": 3.6649558562261377e-07, "loss": 0.5597, "step": 5289 }, { "epoch": 0.92, "learning_rate": 3.64991976746254e-07, "loss": 0.5394, "step": 5290 }, { "epoch": 0.92, "learning_rate": 3.6349140127825533e-07, "loss": 0.5541, "step": 5291 }, { "epoch": 0.92, "learning_rate": 3.619938596910499e-07, "loss": 0.5338, "step": 5292 }, { "epoch": 0.92, "learning_rate": 3.6049935245610846e-07, "loss": 0.5584, "step": 5293 }, { "epoch": 0.92, "learning_rate": 3.5900788004395113e-07, "loss": 0.563, "step": 5294 }, { "epoch": 0.92, "learning_rate": 3.575194429241402e-07, "loss": 0.5645, "step": 5295 }, { "epoch": 0.92, "learning_rate": 3.560340415652852e-07, "loss": 0.5514, "step": 5296 }, { "epoch": 0.92, "learning_rate": 3.545516764350343e-07, "loss": 0.5456, "step": 5297 }, { "epoch": 0.92, "learning_rate": 3.530723480000875e-07, "loss": 0.5435, "step": 5298 }, { "epoch": 0.92, "learning_rate": 3.5159605672618224e-07, "loss": 0.5586, "step": 5299 }, { "epoch": 0.92, "learning_rate": 3.5012280307810344e-07, "loss": 0.538, "step": 5300 }, { "epoch": 0.92, "learning_rate": 3.486525875196756e-07, "loss": 0.5523, "step": 5301 }, { "epoch": 0.92, "learning_rate": 3.47185410513774e-07, "loss": 0.5448, "step": 5302 }, { "epoch": 0.92, "learning_rate": 3.4572127252231134e-07, "loss": 0.5605, "step": 5303 }, { "epoch": 0.92, "learning_rate": 3.4426017400624124e-07, "loss": 0.5571, "step": 5304 }, { "epoch": 0.92, "learning_rate": 3.4280211542556896e-07, "loss": 0.5824, "step": 5305 }, { "epoch": 0.92, "learning_rate": 3.413470972393351e-07, "loss": 0.5458, "step": 5306 }, { "epoch": 0.92, "learning_rate": 3.398951199056266e-07, "loss": 0.564, "step": 5307 }, { "epoch": 0.92, "learning_rate": 3.3844618388157335e-07, "loss": 0.5491, "step": 5308 }, { "epoch": 0.92, "learning_rate": 3.3700028962334595e-07, "loss": 0.5746, "step": 5309 }, { "epoch": 0.92, "learning_rate": 3.35557437586157e-07, "loss": 0.5565, "step": 5310 }, { "epoch": 0.92, "learning_rate": 3.341176282242653e-07, "loss": 0.5477, "step": 5311 }, { "epoch": 0.92, "learning_rate": 3.3268086199096606e-07, "loss": 0.5658, "step": 5312 }, { "epoch": 0.92, "learning_rate": 3.3124713933860074e-07, "loss": 0.5584, "step": 5313 }, { "epoch": 0.92, "learning_rate": 3.298164607185494e-07, "loss": 0.5552, "step": 5314 }, { "epoch": 0.92, "learning_rate": 3.2838882658123736e-07, "loss": 0.5579, "step": 5315 }, { "epoch": 0.92, "learning_rate": 3.2696423737612946e-07, "loss": 0.5415, "step": 5316 }, { "epoch": 0.92, "learning_rate": 3.255426935517303e-07, "loss": 0.5518, "step": 5317 }, { "epoch": 0.92, "learning_rate": 3.241241955555874e-07, "loss": 0.5484, "step": 5318 }, { "epoch": 0.92, "learning_rate": 3.227087438342913e-07, "loss": 0.5543, "step": 5319 }, { "epoch": 0.92, "learning_rate": 3.2129633883346777e-07, "loss": 0.5718, "step": 5320 }, { "epoch": 0.92, "learning_rate": 3.1988698099779114e-07, "loss": 0.556, "step": 5321 }, { "epoch": 0.92, "learning_rate": 3.184806707709698e-07, "loss": 0.5685, "step": 5322 }, { "epoch": 0.92, "learning_rate": 3.1707740859575395e-07, "loss": 0.5455, "step": 5323 }, { "epoch": 0.92, "learning_rate": 3.156771949139392e-07, "loss": 0.5473, "step": 5324 }, { "epoch": 0.92, "learning_rate": 3.1428003016635514e-07, "loss": 0.5651, "step": 5325 }, { "epoch": 0.92, "learning_rate": 3.1288591479287424e-07, "loss": 0.5705, "step": 5326 }, { "epoch": 0.92, "learning_rate": 3.114948492324077e-07, "loss": 0.5497, "step": 5327 }, { "epoch": 0.92, "learning_rate": 3.1010683392290964e-07, "loss": 0.5599, "step": 5328 }, { "epoch": 0.92, "learning_rate": 3.087218693013694e-07, "loss": 0.5387, "step": 5329 }, { "epoch": 0.92, "learning_rate": 3.0733995580381817e-07, "loss": 0.5579, "step": 5330 }, { "epoch": 0.92, "learning_rate": 3.0596109386533015e-07, "loss": 0.5644, "step": 5331 }, { "epoch": 0.92, "learning_rate": 3.045852839200103e-07, "loss": 0.5496, "step": 5332 }, { "epoch": 0.92, "learning_rate": 3.0321252640100883e-07, "loss": 0.5641, "step": 5333 }, { "epoch": 0.92, "learning_rate": 3.018428217405145e-07, "loss": 0.557, "step": 5334 }, { "epoch": 0.92, "learning_rate": 3.0047617036975453e-07, "loss": 0.5665, "step": 5335 }, { "epoch": 0.92, "learning_rate": 2.9911257271899253e-07, "loss": 0.5486, "step": 5336 }, { "epoch": 0.92, "learning_rate": 2.977520292175351e-07, "loss": 0.5484, "step": 5337 }, { "epoch": 0.92, "learning_rate": 2.9639454029372405e-07, "loss": 0.556, "step": 5338 }, { "epoch": 0.92, "learning_rate": 2.9504010637493976e-07, "loss": 0.5257, "step": 5339 }, { "epoch": 0.92, "learning_rate": 2.9368872788759886e-07, "loss": 0.5548, "step": 5340 }, { "epoch": 0.93, "learning_rate": 2.9234040525716325e-07, "loss": 0.5605, "step": 5341 }, { "epoch": 0.93, "learning_rate": 2.9099513890812556e-07, "loss": 0.5549, "step": 5342 }, { "epoch": 0.93, "learning_rate": 2.8965292926401714e-07, "loss": 0.5456, "step": 5343 }, { "epoch": 0.93, "learning_rate": 2.8831377674741203e-07, "loss": 0.5493, "step": 5344 }, { "epoch": 0.93, "learning_rate": 2.8697768177991527e-07, "loss": 0.5528, "step": 5345 }, { "epoch": 0.93, "learning_rate": 2.856446447821737e-07, "loss": 0.5618, "step": 5346 }, { "epoch": 0.93, "learning_rate": 2.8431466617386936e-07, "loss": 0.5295, "step": 5347 }, { "epoch": 0.93, "learning_rate": 2.8298774637372295e-07, "loss": 0.5529, "step": 5348 }, { "epoch": 0.93, "learning_rate": 2.81663885799488e-07, "loss": 0.5645, "step": 5349 }, { "epoch": 0.93, "learning_rate": 2.8034308486796223e-07, "loss": 0.5617, "step": 5350 }, { "epoch": 0.93, "learning_rate": 2.7902534399497416e-07, "loss": 0.5496, "step": 5351 }, { "epoch": 0.93, "learning_rate": 2.777106635953908e-07, "loss": 0.5575, "step": 5352 }, { "epoch": 0.93, "learning_rate": 2.7639904408311435e-07, "loss": 0.5756, "step": 5353 }, { "epoch": 0.93, "learning_rate": 2.750904858710879e-07, "loss": 0.5645, "step": 5354 }, { "epoch": 0.93, "learning_rate": 2.737849893712841e-07, "loss": 0.5448, "step": 5355 }, { "epoch": 0.93, "learning_rate": 2.724825549947152e-07, "loss": 0.5547, "step": 5356 }, { "epoch": 0.93, "learning_rate": 2.711831831514311e-07, "loss": 0.5517, "step": 5357 }, { "epoch": 0.93, "learning_rate": 2.6988687425051453e-07, "loss": 0.56, "step": 5358 }, { "epoch": 0.93, "learning_rate": 2.685936287000856e-07, "loss": 0.5463, "step": 5359 }, { "epoch": 0.93, "learning_rate": 2.673034469072977e-07, "loss": 0.5588, "step": 5360 }, { "epoch": 0.93, "learning_rate": 2.660163292783424e-07, "loss": 0.5628, "step": 5361 }, { "epoch": 0.93, "learning_rate": 2.6473227621844457e-07, "loss": 0.5511, "step": 5362 }, { "epoch": 0.93, "learning_rate": 2.6345128813186737e-07, "loss": 0.5438, "step": 5363 }, { "epoch": 0.93, "learning_rate": 2.6217336542190494e-07, "loss": 0.5592, "step": 5364 }, { "epoch": 0.93, "learning_rate": 2.6089850849088773e-07, "loss": 0.559, "step": 5365 }, { "epoch": 0.93, "learning_rate": 2.5962671774018234e-07, "loss": 0.5545, "step": 5366 }, { "epoch": 0.93, "learning_rate": 2.5835799357018856e-07, "loss": 0.5489, "step": 5367 }, { "epoch": 0.93, "learning_rate": 2.5709233638034345e-07, "loss": 0.5447, "step": 5368 }, { "epoch": 0.93, "learning_rate": 2.558297465691129e-07, "loss": 0.5566, "step": 5369 }, { "epoch": 0.93, "learning_rate": 2.5457022453400313e-07, "loss": 0.5492, "step": 5370 }, { "epoch": 0.93, "learning_rate": 2.533137706715505e-07, "loss": 0.5566, "step": 5371 }, { "epoch": 0.93, "learning_rate": 2.5206038537732736e-07, "loss": 0.5523, "step": 5372 }, { "epoch": 0.93, "learning_rate": 2.508100690459392e-07, "loss": 0.5448, "step": 5373 }, { "epoch": 0.93, "learning_rate": 2.495628220710278e-07, "loss": 0.5497, "step": 5374 }, { "epoch": 0.93, "learning_rate": 2.483186448452624e-07, "loss": 0.5476, "step": 5375 }, { "epoch": 0.93, "learning_rate": 2.470775377603529e-07, "loss": 0.5736, "step": 5376 }, { "epoch": 0.93, "learning_rate": 2.458395012070369e-07, "loss": 0.554, "step": 5377 }, { "epoch": 0.93, "learning_rate": 2.4460453557509037e-07, "loss": 0.5481, "step": 5378 }, { "epoch": 0.93, "learning_rate": 2.4337264125331774e-07, "loss": 0.5443, "step": 5379 }, { "epoch": 0.93, "learning_rate": 2.4214381862956105e-07, "loss": 0.5591, "step": 5380 }, { "epoch": 0.93, "learning_rate": 2.4091806809069086e-07, "loss": 0.5583, "step": 5381 }, { "epoch": 0.93, "learning_rate": 2.396953900226129e-07, "loss": 0.5646, "step": 5382 }, { "epoch": 0.93, "learning_rate": 2.384757848102659e-07, "loss": 0.5426, "step": 5383 }, { "epoch": 0.93, "learning_rate": 2.3725925283762163e-07, "loss": 0.5543, "step": 5384 }, { "epoch": 0.93, "learning_rate": 2.360457944876804e-07, "loss": 0.5267, "step": 5385 }, { "epoch": 0.93, "learning_rate": 2.348354101424799e-07, "loss": 0.5523, "step": 5386 }, { "epoch": 0.93, "learning_rate": 2.3362810018308757e-07, "loss": 0.5628, "step": 5387 }, { "epoch": 0.93, "learning_rate": 2.3242386498960267e-07, "loss": 0.5501, "step": 5388 }, { "epoch": 0.93, "learning_rate": 2.3122270494115752e-07, "loss": 0.5604, "step": 5389 }, { "epoch": 0.93, "learning_rate": 2.30024620415914e-07, "loss": 0.5481, "step": 5390 }, { "epoch": 0.93, "learning_rate": 2.2882961179106933e-07, "loss": 0.5376, "step": 5391 }, { "epoch": 0.93, "learning_rate": 2.2763767944284809e-07, "loss": 0.5727, "step": 5392 }, { "epoch": 0.93, "learning_rate": 2.2644882374651233e-07, "loss": 0.5528, "step": 5393 }, { "epoch": 0.93, "learning_rate": 2.2526304507634933e-07, "loss": 0.545, "step": 5394 }, { "epoch": 0.93, "learning_rate": 2.2408034380567824e-07, "loss": 0.5471, "step": 5395 }, { "epoch": 0.93, "learning_rate": 2.2290072030685561e-07, "loss": 0.554, "step": 5396 }, { "epoch": 0.93, "learning_rate": 2.217241749512622e-07, "loss": 0.5509, "step": 5397 }, { "epoch": 0.94, "learning_rate": 2.2055070810931167e-07, "loss": 0.552, "step": 5398 }, { "epoch": 0.94, "learning_rate": 2.1938032015044964e-07, "loss": 0.5489, "step": 5399 }, { "epoch": 0.94, "learning_rate": 2.1821301144315132e-07, "loss": 0.5664, "step": 5400 }, { "epoch": 0.94, "learning_rate": 2.1704878235492388e-07, "loss": 0.5508, "step": 5401 }, { "epoch": 0.94, "learning_rate": 2.15887633252303e-07, "loss": 0.5548, "step": 5402 }, { "epoch": 0.94, "learning_rate": 2.1472956450085513e-07, "loss": 0.5484, "step": 5403 }, { "epoch": 0.94, "learning_rate": 2.1357457646517644e-07, "loss": 0.5563, "step": 5404 }, { "epoch": 0.94, "learning_rate": 2.12422669508896e-07, "loss": 0.55, "step": 5405 }, { "epoch": 0.94, "learning_rate": 2.1127384399467045e-07, "loss": 0.5469, "step": 5406 }, { "epoch": 0.94, "learning_rate": 2.1012810028418597e-07, "loss": 0.5592, "step": 5407 }, { "epoch": 0.94, "learning_rate": 2.0898543873815958e-07, "loss": 0.5563, "step": 5408 }, { "epoch": 0.94, "learning_rate": 2.0784585971633797e-07, "loss": 0.5568, "step": 5409 }, { "epoch": 0.94, "learning_rate": 2.067093635774975e-07, "loss": 0.5719, "step": 5410 }, { "epoch": 0.94, "learning_rate": 2.055759506794419e-07, "loss": 0.5668, "step": 5411 }, { "epoch": 0.94, "learning_rate": 2.0444562137900802e-07, "loss": 0.5704, "step": 5412 }, { "epoch": 0.94, "learning_rate": 2.0331837603205785e-07, "loss": 0.5562, "step": 5413 }, { "epoch": 0.94, "learning_rate": 2.0219421499348523e-07, "loss": 0.551, "step": 5414 }, { "epoch": 0.94, "learning_rate": 2.0107313861721045e-07, "loss": 0.5352, "step": 5415 }, { "epoch": 0.94, "learning_rate": 1.999551472561867e-07, "loss": 0.5561, "step": 5416 }, { "epoch": 0.94, "learning_rate": 1.9884024126239021e-07, "loss": 0.5336, "step": 5417 }, { "epoch": 0.94, "learning_rate": 1.977284209868313e-07, "loss": 0.5404, "step": 5418 }, { "epoch": 0.94, "learning_rate": 1.9661968677954668e-07, "loss": 0.5559, "step": 5419 }, { "epoch": 0.94, "learning_rate": 1.955140389895993e-07, "loss": 0.544, "step": 5420 }, { "epoch": 0.94, "learning_rate": 1.9441147796508408e-07, "loss": 0.5596, "step": 5421 }, { "epoch": 0.94, "learning_rate": 1.9331200405312222e-07, "loss": 0.5615, "step": 5422 }, { "epoch": 0.94, "learning_rate": 1.922156175998624e-07, "loss": 0.5557, "step": 5423 }, { "epoch": 0.94, "learning_rate": 1.9112231895048293e-07, "loss": 0.572, "step": 5424 }, { "epoch": 0.94, "learning_rate": 1.9003210844918963e-07, "loss": 0.5365, "step": 5425 }, { "epoch": 0.94, "learning_rate": 1.8894498643921454e-07, "loss": 0.5628, "step": 5426 }, { "epoch": 0.94, "learning_rate": 1.878609532628195e-07, "loss": 0.5676, "step": 5427 }, { "epoch": 0.94, "learning_rate": 1.8678000926129036e-07, "loss": 0.5465, "step": 5428 }, { "epoch": 0.94, "learning_rate": 1.8570215477494603e-07, "loss": 0.5541, "step": 5429 }, { "epoch": 0.94, "learning_rate": 1.8462739014312725e-07, "loss": 0.5625, "step": 5430 }, { "epoch": 0.94, "learning_rate": 1.835557157042056e-07, "loss": 0.5483, "step": 5431 }, { "epoch": 0.94, "learning_rate": 1.8248713179557788e-07, "loss": 0.5693, "step": 5432 }, { "epoch": 0.94, "learning_rate": 1.8142163875366824e-07, "loss": 0.5555, "step": 5433 }, { "epoch": 0.94, "learning_rate": 1.803592369139273e-07, "loss": 0.542, "step": 5434 }, { "epoch": 0.94, "learning_rate": 1.792999266108353e-07, "loss": 0.5615, "step": 5435 }, { "epoch": 0.94, "learning_rate": 1.7824370817789539e-07, "loss": 0.5598, "step": 5436 }, { "epoch": 0.94, "learning_rate": 1.7719058194763828e-07, "loss": 0.5806, "step": 5437 }, { "epoch": 0.94, "learning_rate": 1.761405482516232e-07, "loss": 0.5409, "step": 5438 }, { "epoch": 0.94, "learning_rate": 1.7509360742043346e-07, "loss": 0.5653, "step": 5439 }, { "epoch": 0.94, "learning_rate": 1.74049759783681e-07, "loss": 0.5528, "step": 5440 }, { "epoch": 0.94, "learning_rate": 1.7300900566999957e-07, "loss": 0.5581, "step": 5441 }, { "epoch": 0.94, "learning_rate": 1.7197134540705373e-07, "loss": 0.568, "step": 5442 }, { "epoch": 0.94, "learning_rate": 1.7093677932153218e-07, "loss": 0.558, "step": 5443 }, { "epoch": 0.94, "learning_rate": 1.699053077391488e-07, "loss": 0.5494, "step": 5444 }, { "epoch": 0.94, "learning_rate": 1.6887693098464387e-07, "loss": 0.565, "step": 5445 }, { "epoch": 0.94, "learning_rate": 1.6785164938178277e-07, "loss": 0.5551, "step": 5446 }, { "epoch": 0.94, "learning_rate": 1.6682946325335735e-07, "loss": 0.5554, "step": 5447 }, { "epoch": 0.94, "learning_rate": 1.6581037292118573e-07, "loss": 0.5692, "step": 5448 }, { "epoch": 0.94, "learning_rate": 1.6479437870610793e-07, "loss": 0.5551, "step": 5449 }, { "epoch": 0.94, "learning_rate": 1.637814809279925e-07, "loss": 0.5666, "step": 5450 }, { "epoch": 0.94, "learning_rate": 1.6277167990573217e-07, "loss": 0.5525, "step": 5451 }, { "epoch": 0.94, "learning_rate": 1.6176497595724372e-07, "loss": 0.5634, "step": 5452 }, { "epoch": 0.94, "learning_rate": 1.607613693994714e-07, "loss": 0.5554, "step": 5453 }, { "epoch": 0.94, "learning_rate": 1.5976086054838024e-07, "loss": 0.5613, "step": 5454 }, { "epoch": 0.94, "learning_rate": 1.5876344971896494e-07, "loss": 0.5646, "step": 5455 }, { "epoch": 0.95, "learning_rate": 1.5776913722523989e-07, "loss": 0.555, "step": 5456 }, { "epoch": 0.95, "learning_rate": 1.56777923380248e-07, "loss": 0.5562, "step": 5457 }, { "epoch": 0.95, "learning_rate": 1.5578980849605628e-07, "loss": 0.5715, "step": 5458 }, { "epoch": 0.95, "learning_rate": 1.548047928837515e-07, "loss": 0.5553, "step": 5459 }, { "epoch": 0.95, "learning_rate": 1.5382287685344887e-07, "loss": 0.5604, "step": 5460 }, { "epoch": 0.95, "learning_rate": 1.5284406071428893e-07, "loss": 0.5693, "step": 5461 }, { "epoch": 0.95, "learning_rate": 1.5186834477443402e-07, "loss": 0.5476, "step": 5462 }, { "epoch": 0.95, "learning_rate": 1.5089572934106843e-07, "loss": 0.5501, "step": 5463 }, { "epoch": 0.95, "learning_rate": 1.4992621472040393e-07, "loss": 0.5439, "step": 5464 }, { "epoch": 0.95, "learning_rate": 1.4895980121767627e-07, "loss": 0.554, "step": 5465 }, { "epoch": 0.95, "learning_rate": 1.4799648913714104e-07, "loss": 0.548, "step": 5466 }, { "epoch": 0.95, "learning_rate": 1.4703627878207894e-07, "loss": 0.5654, "step": 5467 }, { "epoch": 0.95, "learning_rate": 1.4607917045479814e-07, "loss": 0.5572, "step": 5468 }, { "epoch": 0.95, "learning_rate": 1.4512516445662428e-07, "loss": 0.5671, "step": 5469 }, { "epoch": 0.95, "learning_rate": 1.4417426108790934e-07, "loss": 0.5513, "step": 5470 }, { "epoch": 0.95, "learning_rate": 1.4322646064802937e-07, "loss": 0.5442, "step": 5471 }, { "epoch": 0.95, "learning_rate": 1.4228176343538236e-07, "loss": 0.5527, "step": 5472 }, { "epoch": 0.95, "learning_rate": 1.4134016974738595e-07, "loss": 0.5667, "step": 5473 }, { "epoch": 0.95, "learning_rate": 1.4040167988048748e-07, "loss": 0.549, "step": 5474 }, { "epoch": 0.95, "learning_rate": 1.3946629413015166e-07, "loss": 0.544, "step": 5475 }, { "epoch": 0.95, "learning_rate": 1.3853401279086853e-07, "loss": 0.5593, "step": 5476 }, { "epoch": 0.95, "learning_rate": 1.3760483615614995e-07, "loss": 0.5636, "step": 5477 }, { "epoch": 0.95, "learning_rate": 1.366787645185297e-07, "loss": 0.5605, "step": 5478 }, { "epoch": 0.95, "learning_rate": 1.3575579816956564e-07, "loss": 0.5561, "step": 5479 }, { "epoch": 0.95, "learning_rate": 1.3483593739983646e-07, "loss": 0.55, "step": 5480 }, { "epoch": 0.95, "learning_rate": 1.3391918249894496e-07, "loss": 0.5483, "step": 5481 }, { "epoch": 0.95, "learning_rate": 1.3300553375551362e-07, "loss": 0.5659, "step": 5482 }, { "epoch": 0.95, "learning_rate": 1.3209499145718785e-07, "loss": 0.5654, "step": 5483 }, { "epoch": 0.95, "learning_rate": 1.3118755589063725e-07, "loss": 0.5752, "step": 5484 }, { "epoch": 0.95, "learning_rate": 1.3028322734154997e-07, "loss": 0.5392, "step": 5485 }, { "epoch": 0.95, "learning_rate": 1.2938200609463826e-07, "loss": 0.5595, "step": 5486 }, { "epoch": 0.95, "learning_rate": 1.2848389243363514e-07, "loss": 0.5545, "step": 5487 }, { "epoch": 0.95, "learning_rate": 1.2758888664129553e-07, "loss": 0.5485, "step": 5488 }, { "epoch": 0.95, "learning_rate": 1.266969889993952e-07, "loss": 0.5705, "step": 5489 }, { "epoch": 0.95, "learning_rate": 1.2580819978873283e-07, "loss": 0.5621, "step": 5490 }, { "epoch": 0.95, "learning_rate": 1.249225192891279e-07, "loss": 0.5536, "step": 5491 }, { "epoch": 0.95, "learning_rate": 1.2403994777941964e-07, "loss": 0.5393, "step": 5492 }, { "epoch": 0.95, "learning_rate": 1.231604855374713e-07, "loss": 0.5365, "step": 5493 }, { "epoch": 0.95, "learning_rate": 1.2228413284016472e-07, "loss": 0.5588, "step": 5494 }, { "epoch": 0.95, "learning_rate": 1.2141088996340368e-07, "loss": 0.5483, "step": 5495 }, { "epoch": 0.95, "learning_rate": 1.2054075718211268e-07, "loss": 0.5516, "step": 5496 }, { "epoch": 0.95, "learning_rate": 1.1967373477023925e-07, "loss": 0.5444, "step": 5497 }, { "epoch": 0.95, "learning_rate": 1.1880982300074839e-07, "loss": 0.5426, "step": 5498 }, { "epoch": 0.95, "learning_rate": 1.1794902214562587e-07, "loss": 0.5521, "step": 5499 }, { "epoch": 0.95, "learning_rate": 1.1709133247588156e-07, "loss": 0.5494, "step": 5500 }, { "epoch": 0.95, "learning_rate": 1.1623675426154279e-07, "loss": 0.5598, "step": 5501 }, { "epoch": 0.95, "learning_rate": 1.1538528777165659e-07, "loss": 0.5404, "step": 5502 }, { "epoch": 0.95, "learning_rate": 1.1453693327429405e-07, "loss": 0.55, "step": 5503 }, { "epoch": 0.95, "learning_rate": 1.1369169103654375e-07, "loss": 0.5581, "step": 5504 }, { "epoch": 0.95, "learning_rate": 1.1284956132451286e-07, "loss": 0.5581, "step": 5505 }, { "epoch": 0.95, "learning_rate": 1.120105444033337e-07, "loss": 0.5659, "step": 5506 }, { "epoch": 0.95, "learning_rate": 1.1117464053715388e-07, "loss": 0.5647, "step": 5507 }, { "epoch": 0.95, "learning_rate": 1.1034184998914398e-07, "loss": 0.5483, "step": 5508 }, { "epoch": 0.95, "learning_rate": 1.0951217302148986e-07, "loss": 0.561, "step": 5509 }, { "epoch": 0.95, "learning_rate": 1.0868560989540477e-07, "loss": 0.5652, "step": 5510 }, { "epoch": 0.95, "learning_rate": 1.078621608711139e-07, "loss": 0.5615, "step": 5511 }, { "epoch": 0.95, "learning_rate": 1.0704182620786652e-07, "loss": 0.5406, "step": 5512 }, { "epoch": 0.95, "learning_rate": 1.0622460616393048e-07, "loss": 0.5573, "step": 5513 }, { "epoch": 0.96, "learning_rate": 1.0541050099659333e-07, "loss": 0.5588, "step": 5514 }, { "epoch": 0.96, "learning_rate": 1.0459951096215892e-07, "loss": 0.553, "step": 5515 }, { "epoch": 0.96, "learning_rate": 1.0379163631595413e-07, "loss": 0.5461, "step": 5516 }, { "epoch": 0.96, "learning_rate": 1.0298687731232548e-07, "loss": 0.5585, "step": 5517 }, { "epoch": 0.96, "learning_rate": 1.0218523420463367e-07, "loss": 0.5368, "step": 5518 }, { "epoch": 0.96, "learning_rate": 1.0138670724526345e-07, "loss": 0.5419, "step": 5519 }, { "epoch": 0.96, "learning_rate": 1.0059129668561707e-07, "loss": 0.5612, "step": 5520 }, { "epoch": 0.96, "learning_rate": 9.97990027761142e-08, "loss": 0.5793, "step": 5521 }, { "epoch": 0.96, "learning_rate": 9.900982576619423e-08, "loss": 0.539, "step": 5522 }, { "epoch": 0.96, "learning_rate": 9.822376590431503e-08, "loss": 0.5697, "step": 5523 }, { "epoch": 0.96, "learning_rate": 9.744082343795535e-08, "loss": 0.5438, "step": 5524 }, { "epoch": 0.96, "learning_rate": 9.6660998613608e-08, "loss": 0.5702, "step": 5525 }, { "epoch": 0.96, "learning_rate": 9.58842916767877e-08, "loss": 0.534, "step": 5526 }, { "epoch": 0.96, "learning_rate": 9.511070287202773e-08, "loss": 0.5563, "step": 5527 }, { "epoch": 0.96, "learning_rate": 9.43402324428766e-08, "loss": 0.5604, "step": 5528 }, { "epoch": 0.96, "learning_rate": 9.357288063190473e-08, "loss": 0.5497, "step": 5529 }, { "epoch": 0.96, "learning_rate": 9.280864768069775e-08, "loss": 0.5544, "step": 5530 }, { "epoch": 0.96, "learning_rate": 9.204753382986097e-08, "loss": 0.5468, "step": 5531 }, { "epoch": 0.96, "learning_rate": 9.12895393190183e-08, "loss": 0.5466, "step": 5532 }, { "epoch": 0.96, "learning_rate": 9.05346643868088e-08, "loss": 0.5762, "step": 5533 }, { "epoch": 0.96, "learning_rate": 8.978290927089239e-08, "loss": 0.5653, "step": 5534 }, { "epoch": 0.96, "learning_rate": 8.903427420794531e-08, "loss": 0.5615, "step": 5535 }, { "epoch": 0.96, "learning_rate": 8.828875943366233e-08, "loss": 0.5579, "step": 5536 }, { "epoch": 0.96, "learning_rate": 8.754636518275461e-08, "loss": 0.5598, "step": 5537 }, { "epoch": 0.96, "learning_rate": 8.680709168895185e-08, "loss": 0.5574, "step": 5538 }, { "epoch": 0.96, "learning_rate": 8.60709391850012e-08, "loss": 0.5603, "step": 5539 }, { "epoch": 0.96, "learning_rate": 8.533790790266728e-08, "loss": 0.5677, "step": 5540 }, { "epoch": 0.96, "learning_rate": 8.460799807272991e-08, "loss": 0.5663, "step": 5541 }, { "epoch": 0.96, "learning_rate": 8.388120992499083e-08, "loss": 0.5665, "step": 5542 }, { "epoch": 0.96, "learning_rate": 8.315754368826368e-08, "loss": 0.5449, "step": 5543 }, { "epoch": 0.96, "learning_rate": 8.243699959038287e-08, "loss": 0.5415, "step": 5544 }, { "epoch": 0.96, "learning_rate": 8.171957785819918e-08, "loss": 0.5626, "step": 5545 }, { "epoch": 0.96, "learning_rate": 8.100527871757858e-08, "loss": 0.5579, "step": 5546 }, { "epoch": 0.96, "learning_rate": 8.029410239340562e-08, "loss": 0.5622, "step": 5547 }, { "epoch": 0.96, "learning_rate": 7.958604910958235e-08, "loss": 0.5555, "step": 5548 }, { "epoch": 0.96, "learning_rate": 7.888111908902595e-08, "loss": 0.5453, "step": 5549 }, { "epoch": 0.96, "learning_rate": 7.817931255367006e-08, "loss": 0.565, "step": 5550 }, { "epoch": 0.96, "learning_rate": 7.74806297244668e-08, "loss": 0.5666, "step": 5551 }, { "epoch": 0.96, "learning_rate": 7.678507082138464e-08, "loss": 0.5532, "step": 5552 }, { "epoch": 0.96, "learning_rate": 7.609263606340622e-08, "loss": 0.5634, "step": 5553 }, { "epoch": 0.96, "learning_rate": 7.540332566853158e-08, "loss": 0.5734, "step": 5554 }, { "epoch": 0.96, "learning_rate": 7.471713985378049e-08, "loss": 0.5594, "step": 5555 }, { "epoch": 0.96, "learning_rate": 7.403407883518344e-08, "loss": 0.5637, "step": 5556 }, { "epoch": 0.96, "learning_rate": 7.335414282779063e-08, "loss": 0.5535, "step": 5557 }, { "epoch": 0.96, "learning_rate": 7.267733204566862e-08, "loss": 0.5485, "step": 5558 }, { "epoch": 0.96, "learning_rate": 7.200364670189807e-08, "loss": 0.5505, "step": 5559 }, { "epoch": 0.96, "learning_rate": 7.1333087008576e-08, "loss": 0.5595, "step": 5560 }, { "epoch": 0.96, "learning_rate": 7.066565317681795e-08, "loss": 0.5554, "step": 5561 }, { "epoch": 0.96, "learning_rate": 7.000134541675141e-08, "loss": 0.5446, "step": 5562 }, { "epoch": 0.96, "learning_rate": 6.934016393752352e-08, "loss": 0.5351, "step": 5563 }, { "epoch": 0.96, "learning_rate": 6.868210894729333e-08, "loss": 0.5603, "step": 5564 }, { "epoch": 0.96, "learning_rate": 6.802718065323955e-08, "loss": 0.5458, "step": 5565 }, { "epoch": 0.96, "learning_rate": 6.73753792615528e-08, "loss": 0.5482, "step": 5566 }, { "epoch": 0.96, "learning_rate": 6.672670497744227e-08, "loss": 0.5477, "step": 5567 }, { "epoch": 0.96, "learning_rate": 6.608115800513126e-08, "loss": 0.5524, "step": 5568 }, { "epoch": 0.96, "learning_rate": 6.543873854785831e-08, "loss": 0.5573, "step": 5569 }, { "epoch": 0.96, "learning_rate": 6.47994468078772e-08, "loss": 0.5443, "step": 5570 }, { "epoch": 0.97, "learning_rate": 6.416328298645802e-08, "loss": 0.5531, "step": 5571 }, { "epoch": 0.97, "learning_rate": 6.353024728388502e-08, "loss": 0.5558, "step": 5572 }, { "epoch": 0.97, "learning_rate": 6.290033989945877e-08, "loss": 0.5415, "step": 5573 }, { "epoch": 0.97, "learning_rate": 6.227356103149285e-08, "loss": 0.5613, "step": 5574 }, { "epoch": 0.97, "learning_rate": 6.164991087731831e-08, "loss": 0.5466, "step": 5575 }, { "epoch": 0.97, "learning_rate": 6.102938963328031e-08, "loss": 0.5419, "step": 5576 }, { "epoch": 0.97, "learning_rate": 6.041199749473814e-08, "loss": 0.559, "step": 5577 }, { "epoch": 0.97, "learning_rate": 5.97977346560663e-08, "loss": 0.5504, "step": 5578 }, { "epoch": 0.97, "learning_rate": 5.918660131065568e-08, "loss": 0.5565, "step": 5579 }, { "epoch": 0.97, "learning_rate": 5.8578597650909005e-08, "loss": 0.5664, "step": 5580 }, { "epoch": 0.97, "learning_rate": 5.797372386824651e-08, "loss": 0.5551, "step": 5581 }, { "epoch": 0.97, "learning_rate": 5.7371980153101416e-08, "loss": 0.5562, "step": 5582 }, { "epoch": 0.97, "learning_rate": 5.6773366694922174e-08, "loss": 0.5414, "step": 5583 }, { "epoch": 0.97, "learning_rate": 5.6177883682170254e-08, "loss": 0.5684, "step": 5584 }, { "epoch": 0.97, "learning_rate": 5.558553130232347e-08, "loss": 0.562, "step": 5585 }, { "epoch": 0.97, "learning_rate": 5.4996309741873755e-08, "loss": 0.5694, "step": 5586 }, { "epoch": 0.97, "learning_rate": 5.4410219186326055e-08, "loss": 0.5434, "step": 5587 }, { "epoch": 0.97, "learning_rate": 5.382725982020165e-08, "loss": 0.5771, "step": 5588 }, { "epoch": 0.97, "learning_rate": 5.324743182703262e-08, "loss": 0.5526, "step": 5589 }, { "epoch": 0.97, "learning_rate": 5.267073538936962e-08, "loss": 0.5584, "step": 5590 }, { "epoch": 0.97, "learning_rate": 5.2097170688772955e-08, "loss": 0.5635, "step": 5591 }, { "epoch": 0.97, "learning_rate": 5.152673790582152e-08, "loss": 0.5502, "step": 5592 }, { "epoch": 0.97, "learning_rate": 5.095943722010388e-08, "loss": 0.5508, "step": 5593 }, { "epoch": 0.97, "learning_rate": 5.039526881022494e-08, "loss": 0.5581, "step": 5594 }, { "epoch": 0.97, "learning_rate": 4.9834232853803734e-08, "loss": 0.5455, "step": 5595 }, { "epoch": 0.97, "learning_rate": 4.927632952747119e-08, "loss": 0.5613, "step": 5596 }, { "epoch": 0.97, "learning_rate": 4.8721559006873473e-08, "loss": 0.5537, "step": 5597 }, { "epoch": 0.97, "learning_rate": 4.8169921466670875e-08, "loss": 0.5544, "step": 5598 }, { "epoch": 0.97, "learning_rate": 4.762141708053558e-08, "loss": 0.5577, "step": 5599 }, { "epoch": 0.97, "learning_rate": 4.70760460211539e-08, "loss": 0.5539, "step": 5600 }, { "epoch": 0.97, "learning_rate": 4.653380846022737e-08, "loss": 0.5606, "step": 5601 }, { "epoch": 0.97, "learning_rate": 4.599470456846833e-08, "loss": 0.5545, "step": 5602 }, { "epoch": 0.97, "learning_rate": 4.545873451560545e-08, "loss": 0.5583, "step": 5603 }, { "epoch": 0.97, "learning_rate": 4.492589847037709e-08, "loss": 0.549, "step": 5604 }, { "epoch": 0.97, "learning_rate": 4.4396196600539066e-08, "loss": 0.5667, "step": 5605 }, { "epoch": 0.97, "learning_rate": 4.386962907285797e-08, "loss": 0.5808, "step": 5606 }, { "epoch": 0.97, "learning_rate": 4.334619605311341e-08, "loss": 0.5632, "step": 5607 }, { "epoch": 0.97, "learning_rate": 4.2825897706100237e-08, "loss": 0.5621, "step": 5608 }, { "epoch": 0.97, "learning_rate": 4.230873419562298e-08, "loss": 0.5509, "step": 5609 }, { "epoch": 0.97, "learning_rate": 4.1794705684501393e-08, "loss": 0.5693, "step": 5610 }, { "epoch": 0.97, "learning_rate": 4.1283812334570464e-08, "loss": 0.5431, "step": 5611 }, { "epoch": 0.97, "learning_rate": 4.077605430667375e-08, "loss": 0.5718, "step": 5612 }, { "epoch": 0.97, "learning_rate": 4.027143176067117e-08, "loss": 0.5431, "step": 5613 }, { "epoch": 0.97, "learning_rate": 3.976994485543117e-08, "loss": 0.5642, "step": 5614 }, { "epoch": 0.97, "learning_rate": 3.927159374884082e-08, "loss": 0.5466, "step": 5615 }, { "epoch": 0.97, "learning_rate": 3.8776378597795703e-08, "loss": 0.5571, "step": 5616 }, { "epoch": 0.97, "learning_rate": 3.8284299558205565e-08, "loss": 0.537, "step": 5617 }, { "epoch": 0.97, "learning_rate": 3.779535678499202e-08, "loss": 0.5473, "step": 5618 }, { "epoch": 0.97, "learning_rate": 3.730955043209083e-08, "loss": 0.5449, "step": 5619 }, { "epoch": 0.97, "learning_rate": 3.682688065244966e-08, "loss": 0.562, "step": 5620 }, { "epoch": 0.97, "learning_rate": 3.6347347598026935e-08, "loss": 0.5565, "step": 5621 }, { "epoch": 0.97, "learning_rate": 3.5870951419795244e-08, "loss": 0.5732, "step": 5622 }, { "epoch": 0.97, "learning_rate": 3.5397692267740145e-08, "loss": 0.5537, "step": 5623 }, { "epoch": 0.97, "learning_rate": 3.492757029085914e-08, "loss": 0.5567, "step": 5624 }, { "epoch": 0.97, "learning_rate": 3.446058563716048e-08, "loss": 0.5574, "step": 5625 }, { "epoch": 0.97, "learning_rate": 3.3996738453665466e-08, "loss": 0.5534, "step": 5626 }, { "epoch": 0.97, "learning_rate": 3.35360288864095e-08, "loss": 0.5538, "step": 5627 }, { "epoch": 0.97, "learning_rate": 3.307845708043877e-08, "loss": 0.5525, "step": 5628 }, { "epoch": 0.98, "learning_rate": 3.262402317980917e-08, "loss": 0.5513, "step": 5629 }, { "epoch": 0.98, "learning_rate": 3.2172727327594024e-08, "loss": 0.5668, "step": 5630 }, { "epoch": 0.98, "learning_rate": 3.172456966587301e-08, "loss": 0.5587, "step": 5631 }, { "epoch": 0.98, "learning_rate": 3.127955033574215e-08, "loss": 0.56, "step": 5632 }, { "epoch": 0.98, "learning_rate": 3.0837669477307154e-08, "loss": 0.5435, "step": 5633 }, { "epoch": 0.98, "learning_rate": 3.0398927229686734e-08, "loss": 0.5688, "step": 5634 }, { "epoch": 0.98, "learning_rate": 2.9963323731010406e-08, "loss": 0.5549, "step": 5635 }, { "epoch": 0.98, "learning_rate": 2.9530859118419573e-08, "loss": 0.5582, "step": 5636 }, { "epoch": 0.98, "learning_rate": 2.910153352806977e-08, "loss": 0.5489, "step": 5637 }, { "epoch": 0.98, "learning_rate": 2.8675347095125094e-08, "loss": 0.5447, "step": 5638 }, { "epoch": 0.98, "learning_rate": 2.8252299953761554e-08, "loss": 0.5582, "step": 5639 }, { "epoch": 0.98, "learning_rate": 2.783239223717038e-08, "loss": 0.5391, "step": 5640 }, { "epoch": 0.98, "learning_rate": 2.7415624077551383e-08, "loss": 0.5654, "step": 5641 }, { "epoch": 0.98, "learning_rate": 2.700199560611405e-08, "loss": 0.547, "step": 5642 }, { "epoch": 0.98, "learning_rate": 2.659150695308421e-08, "loss": 0.5578, "step": 5643 }, { "epoch": 0.98, "learning_rate": 2.6184158247697377e-08, "loss": 0.5637, "step": 5644 }, { "epoch": 0.98, "learning_rate": 2.577994961819652e-08, "loss": 0.5561, "step": 5645 }, { "epoch": 0.98, "learning_rate": 2.5378881191843176e-08, "loss": 0.5472, "step": 5646 }, { "epoch": 0.98, "learning_rate": 2.4980953094905226e-08, "loss": 0.562, "step": 5647 }, { "epoch": 0.98, "learning_rate": 2.458616545266246e-08, "loss": 0.5511, "step": 5648 }, { "epoch": 0.98, "learning_rate": 2.4194518389406564e-08, "loss": 0.5586, "step": 5649 }, { "epoch": 0.98, "learning_rate": 2.3806012028442237e-08, "loss": 0.5708, "step": 5650 }, { "epoch": 0.98, "learning_rate": 2.3420646492081645e-08, "loss": 0.5531, "step": 5651 }, { "epoch": 0.98, "learning_rate": 2.3038421901651064e-08, "loss": 0.5686, "step": 5652 }, { "epoch": 0.98, "learning_rate": 2.265933837748757e-08, "loss": 0.5655, "step": 5653 }, { "epoch": 0.98, "learning_rate": 2.228339603893792e-08, "loss": 0.5478, "step": 5654 }, { "epoch": 0.98, "learning_rate": 2.1910595004360767e-08, "loss": 0.559, "step": 5655 }, { "epoch": 0.98, "learning_rate": 2.154093539112667e-08, "loss": 0.5523, "step": 5656 }, { "epoch": 0.98, "learning_rate": 2.117441731561587e-08, "loss": 0.5524, "step": 5657 }, { "epoch": 0.98, "learning_rate": 2.0811040893218282e-08, "loss": 0.5567, "step": 5658 }, { "epoch": 0.98, "learning_rate": 2.0450806238340172e-08, "loss": 0.5582, "step": 5659 }, { "epoch": 0.98, "learning_rate": 2.009371346439082e-08, "loss": 0.5822, "step": 5660 }, { "epoch": 0.98, "learning_rate": 1.973976268379696e-08, "loss": 0.5557, "step": 5661 }, { "epoch": 0.98, "learning_rate": 1.938895400799279e-08, "loss": 0.5457, "step": 5662 }, { "epoch": 0.98, "learning_rate": 1.9041287547424404e-08, "loss": 0.5589, "step": 5663 }, { "epoch": 0.98, "learning_rate": 1.869676341154869e-08, "loss": 0.558, "step": 5664 }, { "epoch": 0.98, "learning_rate": 1.835538170883111e-08, "loss": 0.5456, "step": 5665 }, { "epoch": 0.98, "learning_rate": 1.8017142546752353e-08, "loss": 0.5572, "step": 5666 }, { "epoch": 0.98, "learning_rate": 1.768204603179835e-08, "loss": 0.5497, "step": 5667 }, { "epoch": 0.98, "learning_rate": 1.7350092269469155e-08, "loss": 0.5514, "step": 5668 }, { "epoch": 0.98, "learning_rate": 1.70212813642745e-08, "loss": 0.5598, "step": 5669 }, { "epoch": 0.98, "learning_rate": 1.66956134197338e-08, "loss": 0.5519, "step": 5670 }, { "epoch": 0.98, "learning_rate": 1.637308853837949e-08, "loss": 0.5648, "step": 5671 }, { "epoch": 0.98, "learning_rate": 1.6053706821750336e-08, "loss": 0.5503, "step": 5672 }, { "epoch": 0.98, "learning_rate": 1.5737468370400355e-08, "loss": 0.5526, "step": 5673 }, { "epoch": 0.98, "learning_rate": 1.5424373283889904e-08, "loss": 0.5541, "step": 5674 }, { "epoch": 0.98, "learning_rate": 1.5114421660791245e-08, "loss": 0.5417, "step": 5675 }, { "epoch": 0.98, "learning_rate": 1.4807613598688542e-08, "loss": 0.5549, "step": 5676 }, { "epoch": 0.98, "learning_rate": 1.4503949194173417e-08, "loss": 0.59, "step": 5677 }, { "epoch": 0.98, "learning_rate": 1.4203428542849396e-08, "loss": 0.5733, "step": 5678 }, { "epoch": 0.98, "learning_rate": 1.3906051739329684e-08, "loss": 0.5703, "step": 5679 }, { "epoch": 0.98, "learning_rate": 1.361181887723939e-08, "loss": 0.551, "step": 5680 }, { "epoch": 0.98, "learning_rate": 1.3320730049209973e-08, "loss": 0.5662, "step": 5681 }, { "epoch": 0.98, "learning_rate": 1.3032785346888122e-08, "loss": 0.5664, "step": 5682 }, { "epoch": 0.98, "learning_rate": 1.2747984860926876e-08, "loss": 0.5749, "step": 5683 }, { "epoch": 0.98, "learning_rate": 1.2466328680990069e-08, "loss": 0.5705, "step": 5684 }, { "epoch": 0.98, "learning_rate": 1.2187816895752324e-08, "loss": 0.5768, "step": 5685 }, { "epoch": 0.98, "learning_rate": 1.1912449592899055e-08, "loss": 0.5415, "step": 5686 }, { "epoch": 0.99, "learning_rate": 1.1640226859123138e-08, "loss": 0.5619, "step": 5687 }, { "epoch": 0.99, "learning_rate": 1.1371148780130459e-08, "loss": 0.5375, "step": 5688 }, { "epoch": 0.99, "learning_rate": 1.1105215440634364e-08, "loss": 0.5651, "step": 5689 }, { "epoch": 0.99, "learning_rate": 1.0842426924358996e-08, "loss": 0.5548, "step": 5690 }, { "epoch": 0.99, "learning_rate": 1.058278331403928e-08, "loss": 0.554, "step": 5691 }, { "epoch": 0.99, "learning_rate": 1.032628469141872e-08, "loss": 0.5504, "step": 5692 }, { "epoch": 0.99, "learning_rate": 1.0072931137252717e-08, "loss": 0.5558, "step": 5693 }, { "epoch": 0.99, "learning_rate": 9.822722731303025e-09, "loss": 0.5392, "step": 5694 }, { "epoch": 0.99, "learning_rate": 9.575659552344408e-09, "loss": 0.5663, "step": 5695 }, { "epoch": 0.99, "learning_rate": 9.3317416781602e-09, "loss": 0.5489, "step": 5696 }, { "epoch": 0.99, "learning_rate": 9.09096918554342e-09, "loss": 0.5644, "step": 5697 }, { "epoch": 0.99, "learning_rate": 8.853342150296762e-09, "loss": 0.5653, "step": 5698 }, { "epoch": 0.99, "learning_rate": 8.618860647232608e-09, "loss": 0.5521, "step": 5699 }, { "epoch": 0.99, "learning_rate": 8.387524750174125e-09, "loss": 0.5441, "step": 5700 }, { "epoch": 0.99, "learning_rate": 8.159334531953055e-09, "loss": 0.5558, "step": 5701 }, { "epoch": 0.99, "learning_rate": 7.93429006440971e-09, "loss": 0.5577, "step": 5702 }, { "epoch": 0.99, "learning_rate": 7.712391418396304e-09, "loss": 0.5456, "step": 5703 }, { "epoch": 0.99, "learning_rate": 7.493638663773617e-09, "loss": 0.5526, "step": 5704 }, { "epoch": 0.99, "learning_rate": 7.278031869412117e-09, "loss": 0.5635, "step": 5705 }, { "epoch": 0.99, "learning_rate": 7.0655711031908384e-09, "loss": 0.5682, "step": 5706 }, { "epoch": 0.99, "learning_rate": 6.856256432000719e-09, "loss": 0.5632, "step": 5707 }, { "epoch": 0.99, "learning_rate": 6.650087921739046e-09, "loss": 0.537, "step": 5708 }, { "epoch": 0.99, "learning_rate": 6.447065637316119e-09, "loss": 0.5597, "step": 5709 }, { "epoch": 0.99, "learning_rate": 6.2471896426474774e-09, "loss": 0.552, "step": 5710 }, { "epoch": 0.99, "learning_rate": 6.050460000662783e-09, "loss": 0.5457, "step": 5711 }, { "epoch": 0.99, "learning_rate": 5.856876773296938e-09, "loss": 0.5452, "step": 5712 }, { "epoch": 0.99, "learning_rate": 5.666440021497855e-09, "loss": 0.5635, "step": 5713 }, { "epoch": 0.99, "learning_rate": 5.479149805219796e-09, "loss": 0.5392, "step": 5714 }, { "epoch": 0.99, "learning_rate": 5.29500618342782e-09, "loss": 0.5445, "step": 5715 }, { "epoch": 0.99, "learning_rate": 5.1140092140966606e-09, "loss": 0.5458, "step": 5716 }, { "epoch": 0.99, "learning_rate": 4.9361589542107345e-09, "loss": 0.5544, "step": 5717 }, { "epoch": 0.99, "learning_rate": 4.761455459760811e-09, "loss": 0.5474, "step": 5718 }, { "epoch": 0.99, "learning_rate": 4.589898785750668e-09, "loss": 0.5652, "step": 5719 }, { "epoch": 0.99, "learning_rate": 4.421488986192657e-09, "loss": 0.5319, "step": 5720 }, { "epoch": 0.99, "learning_rate": 4.256226114105477e-09, "loss": 0.5609, "step": 5721 }, { "epoch": 0.99, "learning_rate": 4.094110221519731e-09, "loss": 0.5505, "step": 5722 }, { "epoch": 0.99, "learning_rate": 3.935141359475703e-09, "loss": 0.5469, "step": 5723 }, { "epoch": 0.99, "learning_rate": 3.779319578021134e-09, "loss": 0.5537, "step": 5724 }, { "epoch": 0.99, "learning_rate": 3.626644926214562e-09, "loss": 0.5526, "step": 5725 }, { "epoch": 0.99, "learning_rate": 3.4771174521208705e-09, "loss": 0.5452, "step": 5726 }, { "epoch": 0.99, "learning_rate": 3.3307372028179574e-09, "loss": 0.5554, "step": 5727 }, { "epoch": 0.99, "learning_rate": 3.1875042243911803e-09, "loss": 0.5637, "step": 5728 }, { "epoch": 0.99, "learning_rate": 3.0474185619333573e-09, "loss": 0.5566, "step": 5729 }, { "epoch": 0.99, "learning_rate": 2.9104802595480984e-09, "loss": 0.5587, "step": 5730 }, { "epoch": 0.99, "learning_rate": 2.7766893603486946e-09, "loss": 0.5469, "step": 5731 }, { "epoch": 0.99, "learning_rate": 2.6460459064570065e-09, "loss": 0.5528, "step": 5732 }, { "epoch": 0.99, "learning_rate": 2.5185499390023572e-09, "loss": 0.5541, "step": 5733 }, { "epoch": 0.99, "learning_rate": 2.3942014981270802e-09, "loss": 0.5479, "step": 5734 }, { "epoch": 0.99, "learning_rate": 2.273000622977639e-09, "loss": 0.5641, "step": 5735 }, { "epoch": 0.99, "learning_rate": 2.1549473517124e-09, "loss": 0.5662, "step": 5736 }, { "epoch": 0.99, "learning_rate": 2.0400417214994085e-09, "loss": 0.5424, "step": 5737 }, { "epoch": 0.99, "learning_rate": 1.9282837685141718e-09, "loss": 0.5493, "step": 5738 }, { "epoch": 0.99, "learning_rate": 1.8196735279407684e-09, "loss": 0.5449, "step": 5739 }, { "epoch": 0.99, "learning_rate": 1.7142110339740669e-09, "loss": 0.5444, "step": 5740 }, { "epoch": 0.99, "learning_rate": 1.6118963198163973e-09, "loss": 0.5606, "step": 5741 }, { "epoch": 0.99, "learning_rate": 1.5127294176797703e-09, "loss": 0.5706, "step": 5742 }, { "epoch": 0.99, "learning_rate": 1.416710358786988e-09, "loss": 0.5513, "step": 5743 }, { "epoch": 0.99, "learning_rate": 1.3238391733649824e-09, "loss": 0.5398, "step": 5744 }, { "epoch": 1.0, "learning_rate": 1.2341158906536976e-09, "loss": 0.547, "step": 5745 }, { "epoch": 1.0, "learning_rate": 1.1475405389016482e-09, "loss": 0.5508, "step": 5746 }, { "epoch": 1.0, "learning_rate": 1.0641131453659193e-09, "loss": 0.5759, "step": 5747 }, { "epoch": 1.0, "learning_rate": 9.83833736309947e-10, "loss": 0.5371, "step": 5748 }, { "epoch": 1.0, "learning_rate": 9.067023370101791e-10, "loss": 0.5611, "step": 5749 }, { "epoch": 1.0, "learning_rate": 8.327189717494133e-10, "loss": 0.5649, "step": 5750 }, { "epoch": 1.0, "learning_rate": 7.618836638190186e-10, "loss": 0.5521, "step": 5751 }, { "epoch": 1.0, "learning_rate": 6.941964355222653e-10, "loss": 0.5415, "step": 5752 }, { "epoch": 1.0, "learning_rate": 6.296573081665535e-10, "loss": 0.5645, "step": 5753 }, { "epoch": 1.0, "learning_rate": 5.682663020734059e-10, "loss": 0.5451, "step": 5754 }, { "epoch": 1.0, "learning_rate": 5.100234365706947e-10, "loss": 0.5462, "step": 5755 }, { "epoch": 1.0, "learning_rate": 4.5492872999264303e-10, "loss": 0.5678, "step": 5756 }, { "epoch": 1.0, "learning_rate": 4.029821996864858e-10, "loss": 0.5576, "step": 5757 }, { "epoch": 1.0, "learning_rate": 3.541838620069182e-10, "loss": 0.5434, "step": 5758 }, { "epoch": 1.0, "learning_rate": 3.0853373231720663e-10, "loss": 0.5635, "step": 5759 }, { "epoch": 1.0, "learning_rate": 2.6603182498807777e-10, "loss": 0.5419, "step": 5760 }, { "epoch": 1.0, "learning_rate": 2.2667815340216004e-10, "loss": 0.5594, "step": 5761 }, { "epoch": 1.0, "learning_rate": 1.904727299473219e-10, "loss": 0.5476, "step": 5762 }, { "epoch": 1.0, "learning_rate": 1.5741556602444362e-10, "loss": 0.5453, "step": 5763 }, { "epoch": 1.0, "learning_rate": 1.2750667203964563e-10, "loss": 0.5494, "step": 5764 }, { "epoch": 1.0, "learning_rate": 1.0074605740983957e-10, "loss": 0.5529, "step": 5765 }, { "epoch": 1.0, "learning_rate": 7.713373055939777e-11, "loss": 0.5662, "step": 5766 }, { "epoch": 1.0, "learning_rate": 5.666969892348384e-11, "loss": 0.5495, "step": 5767 }, { "epoch": 1.0, "learning_rate": 3.935396894250154e-11, "loss": 0.5449, "step": 5768 }, { "epoch": 1.0, "learning_rate": 2.5186546070976593e-11, "loss": 0.543, "step": 5769 }, { "epoch": 1.0, "learning_rate": 1.4167434766454435e-11, "loss": 0.5528, "step": 5770 }, { "epoch": 1.0, "learning_rate": 6.296638499492247e-12, "loss": 0.5556, "step": 5771 }, { "epoch": 1.0, "learning_rate": 1.5741597492180406e-12, "loss": 0.5444, "step": 5772 }, { "epoch": 1.0, "learning_rate": 0.0, "loss": 0.5595, "step": 5773 }, { "epoch": 1.0, "step": 5773, "total_flos": 0.0, "train_loss": 0.07439602913279147, "train_runtime": 4363.0883, "train_samples_per_second": 682.024, "train_steps_per_second": 1.323 } ], "logging_steps": 1.0, "max_steps": 5773, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "total_flos": 0.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }